Amfiteatru Economic Journal The Bucharest University of Economic Studies
Faculty of Business and Tourism Volume XXIII ● February 2021 ● No. 56
Quarterly publication
Amfiteatru Economic Journal is recognized and classified category A by National Council of Scientific Research from Romania
Topics of the following two issues
Issue no. 57/2021 – Challenges of IoT Technologies for Businesses and Consumers
Issue no. 58/2021 – Teleworking: Economic and Social Impact and Perspectives
The Journal is indexed/abstracted by the following international economic literature services: Clarivate Analytics (2008)
- Social Sciences Citation Index - Social Scisearch - Journal Citation Reports/Social Sciences Edition
EBSCO Publishing (2009) ProQuest LLC (2012) DOAJ – Directory of Open Access Journals (2010) EconLIT – Journal of Economic Literature (2006) SCOPUS – Elsevier B.V. Bibliographic Databases (2008) IBSS – International Bibliography of the Social Sciences (2006) RePEc – Research Papers in Economics (2004) Cabell’s Directory of Publishing Opportunities (2006) (Business Directories – Economics and Finance) ISSN 1582-9146 www.amfiteatrueconomic.ro
EDITORIAL BOARD Editor-in-Chief
Vasile Dinu, The Bucharest University of Economic Studies, Romania; Academy of Romanian Scientists
Managing Editor Laurenţiu Tăchiciu, The Bucharest University of Economic Studies, Romania
Associate Editors Armenia Androniceanu, The Bucharest University of Economic Studies, Romania Constantin Brătianu, The Bucharest University of Economic Studies, Romania; Academy of Romanian Scientists Cristina Circa, West University of Timișoara, Romania Dan-Cristian Dabija, Babeş-Bolyai University, Cluj-Napoca, Romania Cristian-Mihai Dragoş, Babeş-Bolyai University, Cluj-Napoca, Romania Irina Drăgulănescu, University of Studies of Messina, Messina, Italy Octavian-Dragomir Jora, The Bucharest University of Economic Studies, Romania Judit Oláh, University of Debrecen, Hungary Madălina Dumitru, The Bucharest University of Economic Studies, Romania Valentin Dumitru, The Bucharest University of Economic Studies, Romania Raluca-Gina Gușe, The Bucharest University of Economic Studies, Romania Valentin Hapenciuc, University “Ştefan cel Mare”, Suceava, Romania Nicolae Lupu, The Bucharest University of Economic Studies, Romania Čudanov Mladen, University of Belgrade, Belgrade, Serbia Alexandru Nedelea, University “Ştefan cel Mare”, Suceava, Romania Marieta Olaru, The Bucharest University of Economic Studies, Romania Cătălina Soriana Sitnikov, University of Craiova, Romania Włodzimierz Sroka, WSB University, Polonia Nicoleta Tipi, The University of Huddersfield, Huddersfield, United Kingdom George-Sorin Toma, University Bucharest, Romania Aharon Tziner, The Academic College of Netanya, Netanya, Israel Cristinel Vasiliu, The Bucharest University of Economic Studies, Romania Călin Vâlsan, Bishop’s University, Sherbrooke, Québec, Canada Milena-Rodica Zaharia, The Bucharest University of Economic Studies, Romania
Editorial Advisory Board Dan-Laurenţiu Anghel, The Bucharest University of Economic Studies, Romania Andrej Bertoncelj, University of Primorska, Koper, Slovenia Yuriy Bilan, University of Szczecin, Szczecin, Poland Slobodan Čerović, University Singidunum, Belgrade, Serbia Ung-il Chung, The University of Tokyo, Tokyo, Japan Lóránt Dénes Dávid, Szent István University, Gödöllő, Hungary; ordinary member of the European Academy of Sciences and Arts Emilian Dobrescu, Romanian Academy, Bucharest, Romania Veselin Draskovic, University of Montenegro, Kotor, Montenegro Delgado Francisco Jose, University of Oviedo, Spain Valeriu Ioan-Franc, Romanian Academy, Bucharest, Romania Romualdas Ginevicius, Vilnius Gediminas Technical University, Vilnius, Lithuania David B. Grant, Hanken School of Economics, Finlanda Nicolae Istudor, The Bucharest University of Economic Studies, Romania Dumitru Miron, The Bucharest University of Economic Studies, Romania Puiu Nistoreanu, The Bucharest University of Economic Studies, Romania Bogdan Onete, The Bucharest University of Economic Studies, Romania Rodica Pamfilie, The Bucharest University of Economic Studies, Romania József Popp, University of Debrecen, Hungary Idowu Samuel, London Metropolitan University, London, United Kingdom Ion Stancu, The Bucharest University of Economic Studies, Romania Daniel Stavarek, Silesian University, Karvina, Czech Republic Dalia Streimikiene, Vilnius University, Vilnius, Lithuania Bernhard Swoboda, Universitatea Trier, Germania Gheorghe Zaman, Romanian Academy, Bucharest, Romania
Founders Vasile Dinu, The Bucharest University of Economic Studies, Romania Sandu Costache, The Bucharest University of Economic Studies, Romania
Editorial Office Irina Albăstroiu, The Bucharest University of Economic Studies, Romania Mihaela Bucur, The Bucharest University of Economic Studies, Romania Simona Margareta Bușoi, ASE Publishing House, The Bucharest University of Economic Studies, Romania Răzvan Dina, The Bucharest University of Economic Studies, Romania Raluca Mariana Grosu (Assistant Editor), The Bucharest University of Economic Studies, Romania Silvia Răcaru, ASE Publishing House, The Bucharest University of Economic Studies, Romania Violeta Rogojan, ASE Publishing House, The Bucharest University of Economic Studies, Romania Daniel-Ion Zgură, The Bucharest University of Economic Studies, Romania
Vol. 23 • No. 56 • February 2021 3
Contents
Artificial Intelligence in Wholesale and Retail .................................................................. 5
Vasile Dinu
Artificial Intelligence in Wholesale and Retail
The Profound Nature of Linkage Between the Impact of the Use of Artificial
Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal ............................ 9
Theodor Purcărea, Valeriu Ioan-Franc, Ştefan-Alexandru Ionescu
and Ioan Matei Purcărea
The Impact of Artificial Intelligence on Consumers’ Identity and Human Skills ........ 33
Corina Pelau, Irina Ene and Mihai-Ionuț Pop
Artificial Intelligence in Retail: Benefits and Risks Associated
With Mobile Shopping Applications ................................................................................. 46
Victoria Stanciu and Sînziana-Maria Rîndașu
Consumers’ Perception of Risk Towards Artificial Intelligence Technologies
Used in Trade: A Scale Development Study .................................................................... 65
Pinar Aytekin, Florina Oana Virlanuta, Huseyin Guven, Silvius Stanciu
and Ipek Bolakca
Artificial Intelligence in Electronic Commerce: Basic Chatbots and
Consumer Journey ............................................................................................................. 87
Eliza Nichifor, Adrian Trifan and Elena Mihaela Nechifor
Risks of Observable and Unobservable Biases in Artificial Intelligence
Predicting Consumer Choice ........................................................................................... 102
Florian Teleaba, Sorin Popescu, Marieta Olaru and Diana Pitic
The Integration of Artificial Intelligence in Retail: Benefits, Challenges and
a Dedicated Conceptual Framework ....................................................................................... 120
Ionuț Anica-Popa, Liana Anica-Popa, Cristina Rădulescu and Marinela Vrîncianu
The Impact of Artificial Intelligence Use on the E-Commerce in Romania ................ 137
Adrian Micu, Angela-Eliza Micu, Marius Geru, Alexandru Căpățînă
and Mihaela-Carmen Muntean
Consumer Acceptance of the Use of Artificial Intelligence in Online Shopping:
Evidence From Hungary ................................................................................................. 155
Szabolcs Nagy and Noémi Hajdú
Benefits and Risks of Introducing Artificial Intelligence Into Trade
and Commerce: The Case of Manufacturing Companies in West Africa ................... 174
Zelin Zhuo, Frank Okai Larbi and Eric Osei Addo
4 Amfiteatru Economic
Economic Interferences
The Impact of COVID-19 on Romanian Tourism. An Explorative Case Study
on Prahova County, Romania ......................................................................................... 196
Christine Volkmann, Kim Oliver Tokarski, Violeta Mihaela Dincă and Anca Bogdan
On the Determinants of Fiscal Decentralization: Evidence From the EU ................... 206
Francisco J. Delgado
Food Quality Competition Among Companies and Government Food Safety
Supervision Under Asymmetric Product Substitution .................................................. 221
Ningzhou Shen, Yinghua Song, Dan Liu and Dalia Streimikiene
Are Positive and Negative Outcomes of Organizational Justice Conditioned
by Leader–Member Exchange? ...................................................................................... 240
Or Shkoler, Aharon Tziner, Cristinel Vasiliu and Claudiu-Nicolae Ghinea
Amfiteatru Economic recommends
Immigrant Entrepreneurship in Romania: Drawing Best Practices
From Middle Eastern Immigrant Entrepreneurs’ Experiences .................................. 260
Ozgur Ozmen, Raluca Mariana Grosu and Mariana Dragusin
The Governance Impact on the Romanian Trade Flows. An Augmented
Gravity Model................................................................................................................... 276
Anca Tamaș and Dumitru Miron
Socio-Economic and Macro-Financial Determinants and Spatial Effects
on European Private Health Insurance Markets .......................................................... 290
Gabriela Mihaela Mureșan, Cristian Mihai Dragoș, Codruța Mare,
Simona Laura Dragoș and Alexandra Pintea
Book Review: The Statistical Monograph of the Bucharest University
of Economic Studies. 100 Generations of Graduates .................................................... 308
Mihai Korka
Vol. 23 • No. 56 • February 2021 5
ARTIFICIAL INTELLIGENCE IN WHOLESALE AND RETAIL
Please cite this article as:
Dinu, V., 2021. Artificial Intelligence in Wholesale and Retail. Amfiteatru Economic,
23 (56), pp. 5-7.
DOI: 10.24818/EA/2021/56/5
A major phenomenon of the contemporary world, artificial intelligence (AI) represents the
ability of electronic equipment to perform duties and skills naturally associated with human
intelligence. John McCarthy (1955) defines it as a situation where "... the machine behaves
in a way that could be considered intelligent, if it were human." AI has experienced a
technological trend, which has exploded in the last decade, being a concept with continuous
evolution, offering a very attractive market, with many new opportunities for various
businesses. The field of commerce offers the widest applicable range to artificial
intelligence because it ensures contact with most of the population as potential clientele.
A comprehensive approach to researches into the penetration and proliferation of AI in the
wholesale and retail trade must be carried out along the value chain, from the producer of
the deliverables to the end user. It holds a wide scope: from the automatic launch of orders
to suppliers to the logistics of goods, from the receipt of products within the commercial
network to their merchandising on the shelf, from significant changes in the buying and
consumption habits of the demand carriers to the trader's communication with his clientele,
from the reduction of waste in the commercial network to the elimination of waste in
consumption. Automations in the sphere of goods’ circulation, including the introduction of
humanoid robots, nowadays favours the concept of intelligent commerce, once all retail
stakeholders are connected to the internet and a new dimension of research through Real
Time Data Analysis is developed. This approach provides scientific support for the
development of marketing strategies of the "Proximity Marketing" type, which
communicate to the real and potential customers the specific offer depending on their
buying behaviour.
Issue 56 of the Journal aims to publish the results of researches which tackle the multiple
valences of the implementation of artificial intelligence in the sphere of commerce,
focusing on the following topics: Benefits and risks of introducing artificial intelligence
into trade/commerce; The impact of using artificial intelligence in the trading activity on
purchasing and consumption behaviour; Ethical, legal and societal aspects related to the
promotion of artificial intelligence in trade; Artificial intelligence in customer relationship
management;
The partnership between the consumer and the smart technology confirms the reinvention
of retail organizations, the advancement of retailers on this direction involving the action on
new models and behaviours considering the adoption of the new technology in the context
of COVID-19. The paper “The profound nature of the connection between the impact of
using artificial intelligence in retail on buying and the consumers’ perceptions of artificial
intelligence on the path to the next normal” is focused on the nature of the link between the
impact of using artificial intelligence in retail on the consumers’ behaviour and their
6 Amfiteatru Economic
perceptions of Artificial Intelligence (AI) in order to give retailers an in-depth view of
changing buying and consuming behaviour in Romania on the path to the next normal,
which looks different from any of the years leading up to the current pandemic.
The presence of devices equipped with artificial intelligence in the daily lives of individuals
and consumers holds a number of advantages and disadvantages. Therefore, in order to
better integrate these devices into the lives of consumers, it is important to understand both
their advantages and disadvantages. The paper ”The impact of artificial intelligence on
consumers’ identity and human skills” analyses the relationship between the benefits of
artificial intelligence by increasing efficiency and the fascination created by them and the
main fears related to the human abilities of consumers. It also highlights the role of the
social circle in multiplying the benefits created by artificial intelligence, as well as its
impact in reducing fears related to artificial intelligence. This paper presents a model of
mediation between efficiency and fascination with artificial intelligence and consumer
perception of preserving identity and human skills, having as mediator the influence and the
model of the social circle.
Another paper that highlights the advantages and disadvantages of using artificial
intelligence in commerce is "Artificial intelligence in retail: benefits and risks associated
with mobile shopping applications". This paper explores the practical implications of using
mobile shopping apps, along with solutions based on AI to increase customer engagement,
improve the online shopping experience and encourage the buying impulse, also focusing
on data privacy, legal and ethical implications. This research provides practical insights into
the benefits of integrating IT solutions from the sphere of artificial intelligence into mobile
applications for online commerce, in an ethical manner, protecting users' data privacy and
their freedom of decision in accordance with their own personalities.
The use of artificial intelligence in commerce allows a better analysis of customer needs
and the development of effective marketing strategies. However, although these cutting-
edge technologies offer significant benefits to businesses, some risks may arise as these
technologies grow continuously and eventually become increasingly difficult to control. In
this context, it is important to know how consumers perceive the risks associated with the
use of artificial intelligence in commerce. Therefore, the purpose of the paper “Consumers’
perception of risk towards artificial intelligence technologies used in trade: a scale
development study” is to substantiate an instrument, called the Scale for Assessing
Consumer Perceptions on the Risks of Using Artificial Intelligence in Commerce, to
measure how consumers perceive the risks associated with artificial intelligence
technologies used in commerce.
The paper “Artificial intelligence in electronic commerce: basic chatbots and the consumer
journey” aims to empirically cover the impact of the use of artificial intelligence through
chatbots on online retail in terms of content implemented within the communication
process. The study makes an analysis of how the top ten retailers in Romania, chosen
according to the number of users, react to initiatives to communicate with the public,
through instant messages from the Facebook Messenger application (basic chatbots).
In the paper “Risks of observable and unobservable biases in artificial intelligence used for
predicting consumer choice” a brief summary of cognitive biases and the potential risks of
being replicated in the AI used in consumer choice prediction is presented. The paper also
highlights a separation of biases into two categories, namely observable and unobservable,
Vol. 23 • No. 56 • February 2021 7
and why unobservable biases with their multiplier effects can pose a double risk to AI,
affecting consumer choices.
The research referring to “The integration of artificial intelligence in retail: benefits,
challenges and a dedicated conceptual framework” brings to the fore a variety of advanced
solutions, benefits, but also risks that AI generates in retail, in different segments of the
value chain, briefly noted CECoR, respectively, from improving the customer experience
(Customer Experience , CE), due to virtual agents (chatbots, virtual assistants, etc.), to cost
reductions (Cost, Co) achieved by using smart shelves, to revenue growth (Revenue, R),
determined by product recommendations, offers and customized discounts. The proposed
conceptual framework is focused on the consumer profile and includes recommendations
on the implementation of AI from the perspective of CECoR factors, in a retail company.
As customers interact more and more with companies through digital channels and social
networks, marketers have realized the need to track these interactions and to measure their
performance. In this regard the paper regarding “The impact of artificial intelligence use on
e-commerce in Romania” highlights the role of management teams in e-commerce
companies to automate processes and streamline data flows through predictive analytical
platforms based on artificial intelligence algorithms. At the same time, the research
specifically aims to test the correlations between the intention of managers to automate
certain marketing processes through artificial intelligence algorithms and the desire to
identify customer satisfaction, respectively the use of a customer relationship management
application.
Using as a theoretical background the technology acceptance model (TAM) and an online
survey conducted in Hungary, the paper “Consumer acceptance of the use of artificial
intelligence in online shopping: evidence from Hungary” approaches the problem of
confidence and acceptance by the consumers of the artificial intelligence within online
commerce.
The paper „Benefits and risks of introducing artificial intelligence in commerce: the case of
manufacturing companies in West Africa” refers to an empirical research on a sample of
2.903 manufacturing companies from four West African countries, exploring the ways in
which AI is integrated into trade as well as its impact on customers and its efficiency in
sales processes.
Commerce has been at the forefront of using artificial intelligence, and online stores have
applied all the innovations in the market to attract and retain customers. The use of AI in
commerce brings a number of benefits including: predictive market analysis, facilitating the
decision-making process, systematizing the sales process, automating and optimizing the
data transcription process as well as improving the customer experience.
Editor in Chief,
Vasile Dinu
8 Amfiteatru Economic
Contents
Artificial Intelligence in Wholesale and Retail
The Profound Nature of Linkage Between the Impact of the Use of Artificial
Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal ............................ 9
Theodor Purcărea, Valeriu Ioan-Franc, Ştefan-Alexandru Ionescu
and Ioan Matei Purcărea
The Impact of Artificial Intelligence on Consumers’ Identity and Human Skills ........ 33
Corina Pelau, Irina Ene and Mihai-Ionuț Pop
Artificial Intelligence in Retail: Benefits and Risks Associated
With Mobile Shopping Applications ................................................................................. 46
Victoria Stanciu and Sînziana-Maria Rîndașu
Consumers’ Perception of Risk Towards Artificial Intelligence Technologies
Used in Trade: A Scale Development Study .................................................................... 65
Pinar Aytekin, Florina Oana Virlanuta, Huseyin Guven, Silvius Stanciu
and Ipek Bolakca
Artificial Intelligence in Electronic Commerce: Basic Chatbots and
Consumer Journey ............................................................................................................. 87
Eliza Nichifor, Adrian Trifan and Elena Mihaela Nechifor
Risks of Observable and Unobservable Biases in Artificial Intelligence
Predicting Consumer Choice ........................................................................................... 102
Florian Teleaba, Sorin Popescu, Marieta Olaru and Diana Pitic
The Integration of Artificial Intelligence in Retail: Benefits, Challenges and
a Dedicated Conceptual Framework ....................................................................................... 120
Ionuț Anica-Popa, Liana Anica-Popa, Cristina Rădulescu and Marinela Vrîncianu
The Impact of Artificial Intelligence Use on the E-Commerce in Romania ................ 137
Adrian Micu, Angela-Eliza Micu, Marius Geru, Alexandru Căpățînă
and Mihaela-Carmen Muntean
Consumer Acceptance of the Use of Artificial Intelligence in Online Shopping:
Evidence From Hungary ................................................................................................. 155
Szabolcs Nagy and Noémi Hajdú
Benefits and Risks of Introducing Artificial Intelligence Into Trade
and Commerce: The Case of Manufacturing Companies in West Africa ................... 174
Zelin Zhuo, Frank Okai Larbi and Eric Osei Addo
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 9
THE PROFOUND NATURE OF THE CONNECTION BETWEEN THE IMPACT
OF USING ARTIFICIAL INTELLIGENCE IN RETAIL ON BUYING
AND THE CONSUMERS’ PERCEPTIONS OF ARTIFICIAL INTELLIGENCE
ON THE PATH TO THE NEXT NORMAL
Theodor Purcărea1*, Valeriu Ioan-Franc2, Ştefan-Alexandru Ionescu3
and Ioan Matei Purcărea4 1)3)4) Romanian-American University, Bucharest, Romania
2) Romanian Academy, National Economic Research Institute, Bucharest, Romania
Please cite this article as:
Purcărea, T., Ioan-Franc, V., Ionescu, S.A. and
Purcărea, I.M., 2021. The Profound Nature of Linkage
Between the Impact of the Use of Artificial
Intelligence in Retail on Buying and Consumer
Behavior and Consumers’ Perceptions of Artificial
Intelligence on the Path to the Next Normal. Amfiteatru
Economic, 23(56), pp.9-32.
DOI: 10.24818/EA/2021/56/9
Article History
Received: 27 September 2020
Revised: 6 November 2020
Accepted: 29 November 2020
Abstract
The purpose of this paper is to investigate the impact of the use of Artificial Intelligence in
retail on buying and consumer behavior, better understanding how consumers perceive
Artificial Intelligence on the path to the Next Normal. The consumer-technology partnership
is confirming the reinvention of the retail organizations, retailers’ advancement on this path
involving acting on the new patterns and behaviors by considering the new technology
adoption within the context of COVID-19. Based on a comprehensive literature review on
the impact of Artificial Intelligence in retail a quantitative research design was employed to
demonstrate the nature of linkage between this impact and consumers’ perceptions of
Artificial Intelligence. The data collection was performed through survey conducted in a
supermarket chain in Romania, consumers’ perceptions of Artificial Intelligence being
considered a central metric for assessing the successful use of Artificial Intelligence-enabled
interactions within the context of the offline-online convergence confirmed by the Store of
the Future, and consumers’ use of mobile phones in their Omni channel shopping journey.
The present research, whose results were validated, offers retailers deep consumer insights
with regard to the buying and consumer behavior change in Romania on the path to the Next
Normal.
Keywords: artificial intelligence, retail, buying and consumer behaviour, consumers’
perceptions of artificial intelligence, store of the future, next normal
JEL Classification: D12, L81, M31, O33
* Corresponding author, Theodor Purcărea – e-mail: [email protected]
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
10 Amfiteatru Economic
Introduction
European Commission (2020) states “AI (Artificial Intelligence) is a collection of
technologies that combine data, algorithms and computing power”. The Green Paper on
Vertical Restraints in EC Competition Policy (1997) highlighted how increasingly blurred is
becoming the distinction between manufacturing, wholesaling and retail sectors, consumers’
perceptions of retail organisations and store formats being central to the retailing business.
The Green Paper on European Commerce (Ioan-Franc, 1998) presented commerce
(wholesale and retail) as the touchpoint between citizens and their local community, while
the electronic commerce (seen as being born global and driven by the Internet revolution)
was presented by the European Initiative in Electronic Commerce (1997) as revolutionising
the relationship between consumer and provider. Purcarea and Ioan-Franc (2000) found the
powerful linkage between the business evolution in the digital era, the generational marketing
and the NeWWW Consumer. As demonstrated by Purcarea and Purcarea (2008) Romania
become an important market for the large distribution chains which were struggling to invest
and differentiate by also combining offline and online sales. Purcarea, et al. (2018)
highlighted the significance of Amazon’s Just Walk Out technology, which enables shoppers
to enter a store, grab what they want, and just walk out, their credit card being charged for
the items in their virtual cart (computer vision being a subfield of AI). Just Walk Out
Shopping experience has been highlighted in 2018 as the world’s most advanced shopping
technology (Weise, 2018). Dogtiev (2018) has demonstrated that one of the biggest industries
in today’s world is the mobile app ecosystem, the app store being the major distribution
channel for mobile apps. And as shown by MIT Technology Review Insights (2018), AI
enables customer experience (CX) improvements to be made quickly and at scale. One of the
biggest trends in the app development industry is AI, Google and Microsoft have upgraded,
for instance, their translation apps with neural networks, releasing new AI-powered offline
language translator apps for iOS and Android, while Speech Recognition Technology,
Chatbots, Natural Language Technology, Machine Learning, Biometrics, Emotion
Recognition, Image Recognition and Text Recognition are the main AI technologies used in
mobile apps (Kanada, 2020). To make Omni channel shopping experiences more personal
retailers can easily deploy tools like Customer experience platforms, Social listening tools,
Chatbots, Account based marketing software (Delighted, 2020).
Over the last two years we have witnessed significant progress and achievements, such as:
the announcement of a new Prime Air delivery hybrid drone on the occasion of the
Amazon’s re:MARS 2019 AI & ML Conference (a new global AI event on Machine
Learning, Automation, Robotics, and Space (Anderson, 2019);
the increasing complexity of the consumer landscape, consumers’ behavior becoming
driven by speed and convenience, their homes welcoming try-before-you-buy, and mobile,
AR-based buying reaching a critical mass (Nielsen, 2019);
the improvement of retailers’ efficiencies across their workflows thanks to AI,
automation processes helping with both accurate data creation and providing context around
shoppers’ needs and behaviours (Accenture, 2020);
the relevance conferred by AI in the retail industry (Saker, 2020);
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 11
shoppers’ recognition and authentication – thanks to Capgemini’s Digital EXperience
Transforming Retail (DEXTR) – both via voice and facial recognition (Rajendran, 2020);
the way in which digital become critical – to understand on the path to the Next
Normal (considering critical areas of opportunity for leading in it: digital, transformation,
organization, resilience, and sustainability) – being made more relevant by COVID-19
(McKinsey, 2020);
the changing consumer dynamics (driven by technology adoption) in major areas
(Nielsen, 2020).
This paper focuses on the nature of linkage between the impact of the use of AI in retail on
buying and consumer behavior and consumers’ perceptions of AI (these perceptions being
considered a central metric for determining the successful use of AI-enabled interactions
within the context of the offline-online convergence confirmed by the Store of the Future,
and consumers’ use of mobile phones in their Omni channel shopping journey), with the
purpose of offering retailers deep consumer insights with regard to the buying and consumer
behavior change in Romania on the path to the Next Normal.
Little research has been done on the nature of the above mentioned linkage, despite the great
amount of space in between the information that has been revealed in the past and what
retailers’ decision-makers really need today within the context of the COVID-19 pandemic.
Although some work has been done to date, more studies are needed to fill the current
research gaps, including the need to ascertain the effect of consumers’ misinformation
regarding the benefits of AI-enabled interactions with retailers, and consumer anxiety
(considering consumers’ gap as the difference between their expectations and perceptions).
Hence the imperative of both to build greater trust in retailers, and of a clear communication
from them to consumers in terms of content focused on solutions based on AI to their
problems, empathizing with their current pains – the emotional bond being essential, and
highlighting how these solutions based on AI are making their life effortless, unworried and
secured. Much uncertainty still exists about the nature of the above mentioned linkage, and
as part of retailers’ practice of continuous learning they are under constant pressure of
redrafting their assumptions and understanding, being more sensitive to both the frequency,
and extent of changes in information on consumers’ perceptions. The more so what
consumers say may not coincide with their intentions and actions, not revealing their values
and emotions.
The results validated the structural research model and the associated hypotheses. The
structure of this paper is as follows: introduction, scientific literature review, methodology,
results and discussions, conclusions and references.
1. Review of the scientific literature
1.1. Store of the Future between connectivity and convergence
Earlier research by Purcarea (1994) found that the so-called Smart Store 2000 was a first
research-and-development centre for concepts and technologies which (beyond different
economic benefits) will make shopping more enjoyable and easier for consumers. Ristea,
Ioan-Franc and Purcarea (2005) focused on major developments of the economy of the
distributive trade. Ducrocq (2014) found that digital can allow stores to make the difference
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
12 Amfiteatru Economic
with e-commerce (which is becoming a powerful accelerator of globalization), the
technology making possible a new consumer relationship, intense and productive, also taking
into account both that the economic equation of the store of tomorrow cannot be solved
without resorting to digital tools, and that the complexity is a constituent of the retail of
tomorrow. Purcarea (2015) approached the store of the future by identifying major
developments with regard to the technological innovation, applications and success stories
in retail (Figure no. 1), and presented at SHOP 2015 Conference, Expo Milano 2015 the
foundation for the “Road Map for the Store of the Future” Project (Romanian contribution).
There were underlined significant aspects with regard to the 21st century retail customer, best
serving Omni channel shoppers, and being ready for the future store trends. Further research
by Worldwide Business Research showed in its Future Stores 2019 Retail Technology
Briefing that within the deepening of the so-called, in U.S., Retail Apocalypse major retailers
are rethinking their core strategy assumptions. As suggested by Atmar, et al. (2020) retailers
are under pressure of adequately defining the next normal and their competitive advantage,
better understanding the impact of COVID-19 on their core consumer segments. According
to Loeb (2020), many companies were impacted by the coronavirus pandemic (declaring
bankruptcy or closing some stores and cutting back on new projects), confirming this way
the above mentioned retail apocalypse (which began in 2010).
Figure no. 1. The Store of the Future between connectivity and convergence:
Technological innovation, applications, and success stories in retail
Source: Purcarea, 2015, p. 37
Begley, et al. (2020) underlined retailers’ need of building and maintaining their resiliency,
rethinking their strategies and business models, considering the significant changes in
consumer purchasing patterns and behaviors (such as e-commerce delivery, ordering through
online apps, and contactless pickup), leveraging analytics so as to better understand and react
to the evolutive shifts in consumer demand. In these unprecedented times retailers need to
create consumer impact by acting on the new patterns and behaviors exhibited across all
mediums (Twilio, 2020), and can can seize the opportunity of valorizing already well-known
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 13
empowered AI-based consumer use cases (DataRobot, 2020). Retailers need a clear
understanding of their business strategy and consumers’ buying behavior, so as to adequately
define their consumer, confirming a better understanding of the differences between
consumers, by tracking changes in their awareness (measuring awareness and knowledge),
attitudes (measuring beliefs and intentions), and usage (as a measure of consumers’ self-
reported behavior, measuring purchase habits and loyalty, consumers’ satisfaction providing
a leading indicator of consumer purchase intentions and loyalty), distinguishing seasonality
effects and noise (random movement) from signal, without forgetting, for instance, to consider
consumers’ willingness to recommend and to search, the estimation of the consumers’ store
visits using location-based tracking on mobile phones etc. (Bendle, et al. 2016).
Later research suggested that: consumers’ perceptions (as response relationship) are the link
between the brand inside the organization and the brand inside consumers (McEnally and de
Chernatony, 1999); consumers’ perceptions of choice alternatives are considered as
necessary ingredients of economists’ standard models (Anana and Nique, 2010); consumers
use only a limited number of informational signals to form their perceptions (Bauer, Kotouc
and Rudolph, 2012); consumers’ probability to purchase is depending on perceptions of a
brand (Seter, 2017). Consumers’ perceptions are indissolubly linked to retailers’ line that
shows profit or loss in times of crisis, retailers needing to be observant to trends and act
accordingly so as to build and maintain positive consumers’ perceptions towards their brands
(Keane, 2020). According to Gartner (2020), customer experience (CX) is defined as “the
customer’s perceptions and related feelings caused by the one-off and cumulative effect of
interactions with a supplier’s employees, systems, channels or products”, while CX
management (CEM or CXM) is seen as “the practice of designing and reacting to customer
interactions to meet or exceed their expectations, leading to greater customer satisfaction,
loyalty and advocacy”. Also, according to Gartner (2019), the continuous intelligence needed
for CX of the future will be provided by both human insights, and the speed and precision of
AI technologies and analytics. On the other hand, retailers also really need to consider both
the consumers’ unspoken fears which often go unasked (MECLABS Institute, 2018), and a
predictive model of the prospective consumer’s decision process (Johnson, 2019).
In approaching the store of the future, it is imperative that retailers understand:
To begin navigating to the next normal (which looks unlike any in the years preceding
the current pandemic), by acting across five stages: Resolve (determining the scale, pace, and
depth of business action), Resilience (going from cash management for liquidity and solvency
to acting on broader resilience plans), Return (reassessing their entire business system and
planning for contingent actions), Reimagination (considering the discontinuous shift in the
preferences and expectations of individuals as citizens, as employees, and as consumers;
reconsidering which costs are truly fixed versus variable; pushing the envelope of technology
adoption), and Reform (learning from social innovations and experiments), paying attention to
the imminent restructuring of the global economic order (Sneader and Singhal, 2020);
Why CX is an imperative to adapt to the next normal (which is still taking shape, while
customer expectations are continuing to shift in response) in retail. To recover faster from
the pandemic retailers must quickly reimagine their omnichannel approach so as to create a
distinctive CX, building more resilience in it, by taking initiative across five key actions:
strengthening the commitment to the digital course of action, without frictions and ensuring
that all digital channels are integrated and offer consistent services and experiences; injecting
innovation into Omni channel (by bringing an in-store feel to the digital experience,
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
14 Amfiteatru Economic
launching or diversifying delivery mechanisms, and partnering across retail to enhance
convenience); transforming store operations and win on ‘SafeX’ (which means to provide
safe(r) experiences so as to both alleviate consumers’ anxieties and enable a return to in-
person interactions); reimagining the physical network (by optimizing the footprint,
redefining the role of physical stores, and creating the store of the future); embracing an agile
operating model (Briedis, et al. 2020);
“The new consumer”, as the majority of global consumers have already changed their
shopping habits (Charm et al., 2020);
The overthrowing of the norms about retail, brand loyalty, and consumer behaviour
because of the COVID-19 pandemic; the onboarding into e-commerce; the rethinking of the
format of the store (making it much more comfortable and convenient for the consumer to visit);
the taking into account who are the generations most likely to change brands or retailers (Gen
Z cohort, as well as the higher-income cohort); the way of operating at multiple speeds while
making difference between reacting and shaping (Brady, Gregg and Kim, 2020).
1.2. Artificial Intelligence (AI) in Retail 4.0: Omni-channel retailing
Marsden (2017) did provide a research roundup with regard to how people feel (their hopes,
fears and meanings) about AI, pledging for a better understanding of these feelings if we are
to use or promote AI successfully. McKinsey approached Retail 4.0 and underlined trends
making clearly visible both the future role technology will assume in retail, and the need of
new expertise which combines technology-, marketing- and merchandising-savvy (Desai,
Potia and Salsberg, 2017). Capgemini Research Institute (2018) revealed its preoccupation
with regard to a better understanding of how consumers perceive AI. Bughin, et al. (2018)
also revealed new McKinsey Global Institute research findings (on the basis of a simulation
of the impact of AI on the world economy, considering five broad categories of AI: computer
vision, natural language, virtual assistants, robotic process automation, and advanced
machine learning). Further research (Purcarea, 2019) shows how retailers are preparing for
the impact of automation and AI technologies across all core functions, considering both the
workforce implications, and the shopping tendencies defined by generational gaps within the
reinvention of the shopper behavior by the digital revolution. According to IBM (2019),
organizations need trusted data and trusted AI (the superior enterprise decisions and
processes’ optimization and automation being informed by such data coupled with advanced
analytics and AI, including machine learning), organizations’ expectations for what these
data can do being reset by the interplay between people and AI (known as augmented
intelligence). Considering the spirit and the littera of IBM recommendations, retailers also
really need better learn:
to master the quality of the data used, reducing algorithmic bias, providing the needed
answers with evidence, and in order to have more intelligent interactions with their customers
to utilize AI for sentiment analysis and predictive analytics, becoming their customers’
trusted advisor as experience companies;
to make harmonious to context the humanized experience with the help of AI, which
reveal what makes human the moments in which customers are expressing their specific
preferences, building more trust.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 15
There is no doubt, as suggested by Ellis (2014) that the trending topic in data analytics is
real-time analytics, which according to White (2020) have benefits like timely and accurate
data, risk management (by revolutionizing the way systems can use data so as to predict
outcomes and formulate contingency plans), and less downtime (by anticipating customers’
needs and giving them the best possible CX on the basis of real-time data and predictive
analytics). Further research (Swerdlow and Cyr, 2020) shows that data is becoming an
essential part of CX, and the investments will focus on both gathering the data, and making
sure that data is going to the right places. According to Business Wire (2020): consumers are
helped both to find relevant information and perform tasks with actionable advice by
Intelligent Virtual Assistants (IVAs), while in order to further enhance their customers’
shopping experience retail stores implemented IVAs in their processes; there is a the growing
popularity among the consumers of such devices like chatbots (which are enabling ease of
accessibility) and smart speakers (mainly Amazon Alexa and Google Home) for the variety
of functions ensured. As shown by Market Research Future (2020), Proximity Marketing
(using the cellular technology in order to connect with the mobile-device users who are in
the proximity of a business, knowing the increasing dependency for accessing digital content)
was considerably stimulated by the new coronavirus crisis, being estimated a significant
growth in the upcoming period because of the rising use of the different types of proximity
marketing technology (such as: mainly BLE Beacon, then Near Field Communication - NFC,
GPS Geofencing, Wi-Fi etc.), retail being an important application area of this technology.
As suggested by IBM (2020): becoming digital is the start of a transformation (not only on
the basis of both data and technologies extracting its full value and informing then intelligent
workflows, but also of human-centric design for ever-better engagement of customers) into
a so-called Cognitive Enterprise having capability layers build on each other in an integrated
way, one of these capability layers (composed from: AI, Blockchain, Automation, IoT, 5G)
being made possible with exponential technology; AI is helping to reimagine and reconfigure
intelligent workflows, to personalize learning systems inside the organization, raising the
customer, employee and ecosystem experience (while ensuring a continuum between them)
thanks to an adequate human-technology partnership which is widely supporting continuous
reinvention. Recent research (Fergusson, 2020) has found that considering the inevitability
of the automation in large format store retail (leading to a better shopping experience)
retailers are needing achieving the harmonious balance between shoppers and robots, being
given significant examples of robots’ use in retail, including with reference to the traditional
competition between the giants Amazon and Walmart.
Recent Capgemini Research Institute research (2020) shows retailers’ need to invest in
emerging technologies (such as voice interfaces, facial recognition systems and computer
vision) which are required by the obvious shift towards touchless customer interactions,
considering critical success factors (such as incorporating feedback from consumers,
securing employee buy-in, and leveraging data to iterate and improve CX, including by
improving mobile-centric CX). Very recently, as suggested by Attar (2020), because of the
rising demand for distribution space within the context of the boom of the e-commerce due
to the pandemic, there is a clear interest of operators to drive further efficiencies with the
help of various types of autonomous warehouse solutions, such as: autonomous (driverless)
forklifts, autonomous inventory drones, and autonomous mobile robots.
And the last but not the least, it is worth also highlighting the recently released by Google,
Android app so-called Lookout, which includes a supermarket product detection and
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
16 Amfiteatru Economic
recognition model, allowing to run in real-time entirely on-device. This on-device system
enables a spectrum of new in-store experiences (such as the display of detailed product
information, customer ratings, product comparisons, smart shopping lists, price tracking
etc.). (Chen, 2020)
1.3. The evolution of the retail market in Romania
Mortari (2015) highlights the crucial cognitive practice of reflection. As suggested by
Purcarea and Ioan-Franc (2008) it is imperative to have a better understanding of the
operationalization of knowledge transfer and of the competitiveness of the consumer goods
distribution sector. According to Purcarea and Ratiu (2010) there is a real need of vision that
orientates us towards a new future, considering both consumers’ shifting needs, and stores’
needs of being stocked with the right products at the right prices at the time consumers wants
them, and also taking into account the continuous debate within the academic marketing
community about the so-called critical marketing providing critical understanding of the
marketing operations and mechanisms used by marketers for creating and supporting
customer values, ensuring a retail environment which fits the particular needs of consumers.
Retail Science from CBRE (2015) found that Romania ranks on the 22nd position globally
out of 67 markets in terms of attractiveness and future plans of retailers for expansion, being
necessary a deep understanding of consumer behaviour based on the coming together of
information and analytics, the intersection of data and CX being at the heart of the retail
environment.
A Romanian Competition Council (2018) report on e-commerce sector focused on marketing
strategies (with regard to electro-IT products, the main product category sold in Romania,
this category being followed by fashion and subsidiary articles) highlighted the increase in
the level of receptivity and maturity of supply and demand and analysed including the impact
of the Emag Marketplace (the only significant marketplace on the Romanian Market) on e-
commerce sector. According to Paunescu (2020), Emag CEO underlined on the occasion of
a webinar organized by GPeC in March 2020 that there were 20.625 active traders on Emag
Marketplace. The above mentioned report of the Competition Council made reference to
GPeC (eCommerce Prizes Gala), the most important e-commerce event in Romania. GPeC
is organizing a well-known online shops competition.
A survey performed between April 6th-8th 2020 by Colliers Retail Division revealed
(Colliers, 2020) how the COVID-19 pandemic affected retailers’ activity and expectations in
Romania, showing that 67% (mainly in the grocery, food & beverage, electronics, healthcare,
and fashion segments) of respondents (which planned to support their business in the context
of the COVID-19 crisis) focused on alternative sales channels (online commerce), while 60%
of respondents (the majority being from fashion and food & beverage segments) expressed
the belief that – due to both psychological medium term impact of the new coronavirus crisis
on the consumers’ appetite for shopping, and an expected contraction of their disposable
income – the recovery of their business will happen next year, in 2021. As underlined by
Roşca (2020) Kaufland (127 stores in 2019; it announced the opening of 13 stores in the
current financial year) is now followed by Carrefour (371 stores), Lidl (260) etc. and despite
the COVID-19 pandemic their investments will continue, the expansion champions
continuing to be Profi and Mega Image (known as active small-format retailers, supermarkets
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 17
and proximity stores). According to USDA (2020) there was a dramatically shift of
Romanian consumers in favor of online shopping beginning with the first Romanian COVID-
19 case. And as shown by Cristea (2020) Penny Market, for instance, is adapting to the
changes in consumer behavior by launching together with Auto Total the home delivery
service Lyvra, which allows customers both to place orders online, and receive the products
at home. According to Buhaescu (2020) even in the context of the COVID-19 pandemic more
than half of the Romanian consumers (who participated at on LinkedIn applied Deloitte
Romanian Consumer Trends questionnaire, between June 3- July 20, 2020) are still attracted
by a combination of online and offline shopping channels, BOPIS offering them both the
greatest safety, and convenience, while proximity is always a proper solution. It was also
revealed that: there is a predominance of the socially conscious shoppers who supports local
brands (55%) as being the most common profile, followed by the convenience seekers (33%,
preferring to shop in their neighborhood), then by the so-called bargain hunters (7%), and
stockpilers (6%, looking for storing a large supply for future use); retailers are challenged to
embraced technology more quickly than planned, considering the above mentioned
consumers’ preferences, including that for buying local products. Deloitte Romania (2020)
highlighted not only the strong impact on the local economy and urban mobility of the
COVID-19 pandemic, but also its effect on a presumably acceleration of the offline-online
convergence in retail (online shopping having a great potential). While a BOLD study by
Profero&iSense Solutions (2020) showed that that regarding the integration of AI solutions
65% of consumers stated that they feel comfortable with online retail (the highest percentage
for a field). And a Romanian Competition Council (2020) report on “The evolution of
competition in key sectors - 2020” has mentioned, among other aspects, that the large
commercial networks have highlighted the impact of medical and economic developments
on their activity (costs, adaptation to new consumer behavior, changes in trade flows, etc.).
2. Research methodology
Consumers are more sensitive to the experiences they have in their lives right now during the
COVID-19 pandemic. Their rational thinking (which is shaping their consumer and buying
decision) is both preceded, and influenced by emotion, which is subsequently activating the
resulting behavioural action. Retailers are more and more aware of their need to engage
consumers by memorable, personal, and emotional experiences, making them valuing the
time and energy spend with a retailer. The results delivered by retailers’ digital
transformation are depending on the digital consumer experience, essential to retailers’
success being the value drivers. The way in which a consumer is feeling during a retail
experience is expressed by his experiential value, which is now increasingly exceeding both
the economic value, and the functional value. It is therefore necessary that retailers should be
able: to identify which area is guiding the prevailing value for consumers when facing a
purchase situation, going beyond the fragmenting data; to find out what consumers feel (even
when they don’t know themselves), making difference between how they think, what they
think, and what they communicate about what they think or feel (by speaking, writing or in
some other way), by exploring more fully the unconscious thoughts and emotions through
blending of technology and psychology (Shaw, 2020).
Based on a comprehensive literature review of the impact of Artificial Intelligence in retail
on buying and consumer behavior, a quantitative research was employed (using a structural
equation model and an associated AI in Retail questionnaire) to demonstrate to the linkage
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
18 Amfiteatru Economic
between this impact and consumers’ perceptions of Artificial Intelligence. The research was
based on six hypotheses, consumers’ perceptions of AI (a single exogenous variable) being
seen as the input construct (allowing the statement of the below hypothesis H1), considering
our focus on questioning its impact on retailer’s successful use of AI-enabled interactions
with consumers, through the facilitating function of retailer’s online-offline channel
integration (a mediating variable allowing the statement of the below hypothesis H4). As
shown in the figure no. 2 below, retailer’s online-offline channel integration: depends on the
retailer’s alignment of Omni channel strategy to actions (an endogenous variable allowing the statements of the below hypotheses H2 and H3), the above last construct appearing as a
moderating factor with regard to both retailer’s online-offline channel integration, and
retailer’s successful use of AI-enabled interactions with consumers; impacts on retailer’s
successful use of AI-enabled interactions with consumers, being a mediating factor between
the consumers’ perceptions of AI (the input construct) and retailer’s successful use of
AI-enabled interactions with consumers.
Figure no. 2. The theoretical research model
The theoretical research model designed on the basis of a consecrated quantitative model
(Hair, et al. 2017) allows the exploration of the causal relationships between the above
mentioned main identified constructs (to which was added a fifth construct in the risky
perspective of a new wave of the new coronavirus crisis: Consumers’ needs to feel better
informed and educated, while learning to live with the new reality, with regard to the benefits
of AI interactions with retailers), reflecting the following hypothesized influences shown by
the SEM’s path diagram: Hypothesis 1 (H1), Consumers’ perceptions of AI has a positive
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 19
influence on retailer’s online-offline channel integration; Hypothesis 2 (H2), Retailer’s
alignment of Omni channel strategy to actions has a positive influence on retailer’s online-
offline channel integration; Hypothesis 3 (H3), Retailer’s alignment of Omni channel
strategy to actions has a positive and significant influence on retailer’s successful use of AI-
enabled interactions with consumers; Hypothesis 4 (H4), Retailer’s online-offline channel
integration has a positive influence on retailer’s successful use of AI-enabled interactions
with consumers; Hypothesis 5 (H5), There is a negative influence of Retailer’s successful
use of AI-enabled interactions with consumers on Consumers’ needs to feel better informed
and educated with regard to the benefits of AI interactions with retailers, this negative
influence being due to Consumers’ anxiety (a mental cost depending on trust and clear
content communication to Consumers) and friction (a physical cost depending on reducing
the burdens on Consumers in doing something with regard to AI-enabled interactions and
stimulating them to do it; Hypothesis 6 (H6), There is a negative influence of Consumers’
perceptions of AI on Consumers’ needs to feel better informed and educated with regard to
the benefits of AI interactions with retailers, this negative influence being also due to
Consumers’ anxiety and friction, especially in the risky perspective of a new wave of the new
coronavirus crisis.
The data collection in the quantitative study was performed through survey conducted in a
supermarket chain in Romania from July 6, 2020 to August 19, 2020, based on the above-
mentioned AI in Retail questionnaire consisting of a set of 39 questions answered by a sample
of 1140 respondents (table no. 1), the average time spent to answer the questions being 7.30
minutes. As type of questions there were used closed questions: limited choice, multiple
choice, checklist, partially closed. And as type of data sought (to be obtained from these
closed questions) there were used factual, opinion, and behavioral questions.
Table no. 1. Sample Structure (research) – Romania
Romania Younger
than 18
18-25
years
old
26-35
years
old
36-45
years
old
46-55
years
old
56-65
years
old
66 or
older
Total
Male 13 164 133 152 59 31 4 556
Female 10 178 141 155 55 37 8 584
Total 23 342 274 307 114 68 12 1140
3. Results and discussions
To test the hypotheses in the proposed model, we used the questionnaire survey method.
Later we used the Amos software package from IBM. We used a series of indicators to
determine the validity of the model. We would like to mention: the non-normed fit index,
also known as the Tucker-Lewis index - TLI, the Goodness of Fit Index - GFI, the adjusted
goodness of fit index - AGFI, the coefficient of determination - CD, the comparative fit index
- CFI, the standardized root mean square residual - SRMR, the weighted root mean square
residual - WRMR. The 5 latent variables and related items are presented in the table below
(table no. 2).
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
20 Amfiteatru Economic
Table no. 2. The 5 latent variables and related items
Retailer’s
alignment of
Omni channel
strategy
to actions
A
A1 Did you receive marketing messages from retailers regardless of the channel
used?
A2 In general, do you find identical retailers’ stocks (the variety and assortment
levels of goods) on any channel you search?
A3
In your interactions with retailers, have you noticed that your data,
information about your business is concentrated under a single account,
regardless of the channel through which you interacted?
A4 Did you find identical prices for identical products on the channels you
searched for?
A5 Did you receive marketing messages on your mobile devices to influence
purchasing being in close proximity to a retail store?
A6 Do the retailers you are interested in use bricks-with-clicks technologies?
A7
Do the retailers you are interested in offer services such as click-and-
collect/BOPIS (in-store pickup of products purchased online) or online
purchase of products while at a physical store?
A8 Were you able to return online purchased products in classic locations?
A9 Have you noticed more attentive staff or improved interaction software from
retailers?
A10 Have you been asked for personal data from retailers in order to create a
profile for you?
A11 Have you interacted with retargeting applications, personalized email
marketing, social media marketing, cart abandonment emails?
Retailers’ online
offline channel
integration
I
I1 Do you already use or willing to use a mobile wallet to speed up payment and
avoid carrying cash or cards?
I2 Are you an active user of retailers’ mobile applications?
I3 Are you already using or willing to use BOPIS (buy online, pickup in store)?
I4
Do you document yourself on several channels before making an „important”
purchase (excluding purchases of strict necessity, daily) in physical stores or
online?
I5 Have you ever completed an online order outside the physical store’s opening
hours?
Retailers’
successful use of
AI-mediated
interactions with
consumers
U
U1
Are you comfortable with using your mobile phone at different points (such
as self-identification, product information, in-store navigation, self-checkout,
and payments) in your in-store customer journey?
U2
Do you think that will be an increase of the touchless interactions (through
voice assistants, facial recognition, or apps) with retailers after the end of the
new coronavirus crisis?
U3 Are you already making or willing to make online purchases on the basis of
personalized ads which use personal data?
U4 Are you using or willing to use augmented and virtual reality to assess new
goods and services?
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 21
Consumers’
perceptions of AI
(benefits from
AI-enabled
interactions with
retailers)
P
P1 Higher personalization
P2 More trustworthy
P3 Improved privacy and security of personal data
P4 Reduction in effort from your end in your interaction with the retailer
P5 Faster resolution of support issues (such as chatbots/virtual agents over)
P6 24/7 availability
P7 Greater control over the interactions
P8 Other aspects
P9 Human-like behaviour and personality
P10 Ability to provide greater empathy
P11 Ability to understand human emotions and respond
P12 Human-like intellect
P13 Human-like voice
P14 Other human-like qualities (such as pleasant conversation)
Consumers’
needs to feel
better informed
and educated
with regard to
the benefits of AI
interactions with
retailers
N
N1 Are you anxious to minimize physical contact and maximize contactless when
interacting with a retailer?
N2
Do you have privacy concerns about potential data malpractices when
retailers are enabling interactions via AI? Do you want to be made aware of
it?
N3
Do you want retailer to provide an “opt-out” (a choice between adopting
different above mentioned technologies or traditional forms of interactions) so
as to avoid being anxious?
N4 Do you think the interaction with AI could lead to unpredictable, more
spontaneous behavior from you?
N5 Are you afraid to share your personal data in order to receive personalized
content and offers?
A – The retailers’ alignment of the Omni channel strategy to actions was evaluated on a
Likert scale with 5 levels: Yes, Partially true, Neutral, Rather no and No. Survey participants
answered questions A1-A11. Higher values reflect an advance in the implementation of the
Omni channel strategy.
I – The integration of the retailers’ offline and online channels was evaluated on a Likert
scale with 5 levels: Yes, Sometimes, Rarely, Very rarely and No. Survey participants
answered questions I1-I5. Higher values indicate more advanced integration.
U – Retailers’ successful use of AI-enabled interactions with consumers were assessed by
defining a 5-items variable, evaluated on a 5-level Likert scale: Yes, Partially True, Neutral,
Rather No, and No. Survey participants answered questions U1-U4. The higher the values,
the more successful the use of AI interactions are.
P – Consumers’ perceptions regarding the benefits of AI-enabled interactions with retailers
were assessed on a 5-level Likert scale: Yes, Sometimes, Rarely, Very Rarely and No. Survey
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
22 Amfiteatru Economic
participants answered questions P1-P14. Higher values show a positive perception, while low
values reflect a negative impact on consumers.
N – Consumers’ needs to feel better informed and educated with regard to the benefits of
AI interactions with retailers were assessed by defining a 5-items variable, evaluated on a 5-
level Likert scale: Yes, Partially true, Neutral, Rather no and No. Survey participants
answered questions N1-N5. Higher values indicate a high degree of misinformation and
consumer anxiety.
Table no. 3 shows the mean and the standard deviation for the variables included in the study.
Table no. 3. The mean and the mean square deviation for the variables included
in the study
Mean SD Mean SD
A
A1 3.663 1.145
P
P1 2.242 1.622
A2 3.213 1.506 P2 2.344 1.665
A3 3.896 1.357 P3 2.412 1.690
A4 4.380 0.990 P4 2.378 1.678
A5 2.741 1.475 P5 2.582 1.740
A6 3.476 1.261 P6 3.670 1.657
A7 3.699 1.365 P7 3.330 1.758
A8 4.398 1.137 P8 1.324 0.633
A9 3.887 1.429 P9 3.031 1.515
A10 3.464 1.201 P10 3.056 1.514
A11 3.777 1.097 P11 3.133 1.509
Mean SD P12 3.260 1.491
I
I1 2.650 1.697 P13 3.447 1.534
I2 4.604 0.835 P14 3.082 1.512
I3 3.073 1.682
I4 3.809 1.305
I5 3.408 1.408 Mean SD
Mean SD
N
N1 1.838 1.378
U
U1 3.985 1.362 N2 3.119 1.785
U2 3.092 1.786 N3 3.925 1.529
U3 2.786 1.776 N4 2.582 1.740
U4 2.492 1.699 N5 2.786 1.776
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 23
Taking a brief look, we notice that the average score for the retailer’s alignment of the Omni
channel strategy to actions varies between 2,741 which refers to the use of techniques to
influence the purchase, such as mobile technology to identify buyers in close proximity to a
retail store, up to a maximum of 4,398 to allow the return of products purchased online in
classic locations. Regarding the integration by the retailer of the online and offline channels,
the average score varies between 2.65 for the intention of using a mobile wallet to speed up
payment and to avoid carrying cash or cards and 4,604 for the degree of retailers’ mobile
applications usage. Regarding the retailer’s use of AI-enabled interactions with consumers,
the minimum score of 2,492 is recorded for the availability to use augmented and virtual
reality to evaluate new goods and services, and the maximum score of 3,985 refers to the use
of mobile phone at different points (such as self-identification, product information, in-store
navigation, self-checkout, and payments) during in-store customer journey. The score of
consumers’ perceptions regarding AI varies between 1,324 for aspects that were not
mentioned and 3,67 for the 24/7 availability of AI based online services. Consumers’ needs
to feel better informed and educated with regard to the benefits of AI interactions with
retailers average scores between 1,838 on anxiousness to minimize physical contact and
maximize contactless when interacting with a retailer, and 3,925 on providing an “opt-out”
(a choice between adopting different above mentioned technologies or traditional forms of
interactions) so as to avoid being anxious.
The reliability is estimated using the test proposed by Lee Cronbach in 1951 which measures
the internal consistency of the items of a questionnaire, or rather the reliability of the test. It
is the most used method to highlight the quality of measuring latent variables. Table no. 4
shows the scale reliability.
Table no. 4. Scale reliability
Scale α
Cronbach
Number
of Items
The retailers’ alignment of the Omni channel strategy
to actions 1-5 0.912 11
The integration of the retailers’ online and offline channels 1-5 0.831 5
Retailers’ successful use of AI-enabled interactions
with consumers 1-5 0.698 4
Consumers’ perceptions regarding the benefits of AI-enabled
interactions with retailers 1-5 0.883 14
Consumers’ needs to feel better informed and educated
with regard to the benefits of AI interactions with retailers 1-5 0.735 5
The higher our score, the better the measure. A score above 0.9 (as we have for the retailers’
alignment of the Omni channel strategy to actions) is considered excellent, a score between
0.8-0.9 (as we have for the integration of the retailers’ online and offline channels and the
Consumers’ perceptions regarding the benefits of AI enabled interactions with retailers) is
considered good, over 0.7 (as we have for the Consumers’ needs to feel better informed and
educated with regard to the benefits of AI interactions with retailers) is considered acceptable.
Below 0.7 we have a poor consistency, however we can consider that the value for the
Retailers’ successful use of AI-mediated interactions with consumers is at the limit, and if
we take into account the relatively small number of participants, we place this value in the
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
24 Amfiteatru Economic
acceptable category. Trying to remove from the items, we fail to increase the value of the α
Cronbach test, which is why we do not make any changes to the model. At the same time, a
large number of items will increase the value of α Cronbach, while a small number can lead
to low values. The increase in α Cronbach should be done by adding relevant questions. We
keep this observation for future research.
The evaluation of the structural model confirms the initial hypotheses. Figure no. 3 presents
the resulted structural model.
Figure no. 3. The structural model
Five of the six working hypotheses are validated, because the p-value is less than 5%. There
is also a negative but weak relationship between the retailers’ successful use of AI-mediated
interactions with consumers and the Consumers’ needs to feel better informed and educated
with regard to the benefits of AI interactions with retailers. Also, the associated risk is
approximately 11% (low evidence against the null hypothesis in favour of the alternative). The table below (table no. 5) shows the output also generated with the Amos software.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 25
Table no. 5: SEM output
Hypothesis Relation β p-value Decision
H1 IA 0,414 0.000 Valid model
H2 UA 0,281 0.008 Valid model
H3 UI 0.369 0.000 Valid model
H4 IP 0.621 0.000 Valid model
H5 NP -0.504 0.044 Valid model
H6 NU -0.172 0.114 11% risk
Conclusions
The findings of the research are to some extent constrained by certain limitations, some of which
provide opportunities for further research. We have a low representation of people over 46 years,
their share being only 17% of the total number. This could also be explained by the fact that
during the State of Alert period (July 6 - August 19, 2020), older people went less shopping,
having a higher risk of getting infected with the new COVID-19 coronavirus, as well as having
complications. Therefore, the final results should be seen with circumspection, as they are rather
relevant for the 18-45 age segment (Xennials + Millennials + Gen Z). Furthermore, the reference
period to which the closed questions were applied included holiday time which was strongly
impacted by the COVID-19 pandemic (staying safe being a priority – fear of disease spread, of
another disease outbreak, of unemployment etc. – and beyond financial issues or restrictions
and border closures traveling to a destination being sometime a real challenge, including
considering the changed way of celebrating different events or even the changed perception of
time), consumers’ perceptions when assessing something (by selecting, sensing, organizing,
interpreting, and choosing) in their buying decision process influencing their behaviour (being
necessary to better understand the fine border between what they need, want, prefer, expect, buy
and feel this way, sometimes being very difficult under such above mentioned unprecedented
circumstances). And retailers’ emotionally engagement with consumers involves an obvious
better understanding of consumers’ perceptions, becoming consumers experts based on
competency, adequate tools, and building trust.
The research developed a better understanding (needing to be continued) of the consumers’
needs to feel better informed and educated (while learning to live with the new reality) with
regard to the benefits of AI interactions with retailers (becoming more and more aware of:
the role of AI in both predicting patterns, and changing the scalability of behavioral data
analysis; the whole spectrum of benefits of AI for digital marketing), where mobile-based
contactless transactions (such as retail store self-checkouts which use mobile apps) need to
resonate more with consumers at the first attempt and a growing number of them are willing
to use in-home devices or virtual assistance to purchased products.
Greater geographic reach, more supermarket chains for instance, may help understand certain
differences which are important for the success of the future research, in the context of
increasing touchless interactions (through voice assistants, facial recognition, or apps) with
retailers after a presumably end of the new coronavirus crisis. We intend a calibration of the
model, adding a series of questions for the latent variable N - Consumers’ needs to feel better
informed and educated with regard to the benefits of AI interactions with retailers. We will
also try to complete the questionnaire so that the other age groups (especially Generation X
and baby boomers) are fairly represented. We also plan to extend the research to other Eastern
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
26 Amfiteatru Economic
European countries, while making a comparison of the habits and perceptions of people from
different geographical areas.
The above-mentioned findings confirm both that the Romanian buying and consumer
behavior is considerably changing on the path to the Next Normal, and that consumers’
perceptions (as response relationship) of AI represent a central metric for determining the
successful use of AI-enabled interactions within the context of the offline-online convergence strongly influencing Romanian consumers’ Omni channel shopping journey. In the same
time, there is clear evidence that the adoption by the Romanian consumers of the beneficial
innovations driven by AI show an almost identical way to European and US consumers.
Given these conditions, it is imperative for retailers’ organizations acting on the Romanian
market to take into account the above-mentioned practical insights and to adapt their
offerings to the evolving buying and consumer habits (by continuously tracking changes in
consumers’ awareness, attitudes and usage), including by harmonizing their messages to the
revealed new Romanian consumer mind-sets within the context of the fundamental shifts
determined by the current unprecedented crisis.
This research provide an interesting and promising areas to work on the linkage between
consumers’ general anxiety within the context of the new coronavirus crisis (economic
uncertainty, unemployment, financial insecurity) and the shifts in consumer behaviour and
consumers’ value perception (better experience, saving time and effort etc.; price; quality).
And this based on the revealed findings (the need to ascertain the effect of consumers’
misinformation regarding the benefits of AI-enabled interactions with retailers, and consumer
anxiety, considering consumers’ gap as the difference between their expectations and
perceptions; the imperative of both to build greater trust in retailers, and of a clear
communication from them to consumers in terms of content focused on solutions based on
AI to their problems, empathizing with their current pains, these solutions based on AI
making consumers’ life effortless, unworried and secured).
From the point of view of managerial implications, the findings of this research mean in
terms of retailers’ actions accelerating to earn consumers’ trust with trusted AI in order to
capture their data, to use AI to maximize their shopping experience by removing their key
pain points and ensuring them a more personalized experience, providing them with
everything they need at the right time alongside the touchpoints and micromoments with the
help of mobile apps, virtual conversational and voice assistants, robots, drones etc.
The perspective taken by this action-oriented research (based on inside knowledge but trying
to obtain both insider and outsider knowledge, so as to maintain the appropriate detachment)
allowed not only sensitivity and empathy, but also understanding of the matters, reflection,
and the identification of the necessary change. The hypotheses developed in the quantitative
study have derived from gaps in the literature, the research going beyond just collecting
evidence to support the postulated hypotheses, trying to identify alternative explanations
making more sense within the acceleration of the adoption of digital commerce and building
trust and emotional connection, one hand, and the emphasis placed on both the prioritization
of shopper safety, and the perception of what a good or service is worth to a consumer, on
the other hand. And this including in order to avoid neglecting opportunities for future
research. There is no doubt that on the path to the Next Normal, data science and the
disruptives technologies like AI will reshape the retail landscape, retailers responding with
agility and innovation in their omnichannel experience, better proceeding and strengthen
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 27
their ties to consumers by revolutionizing CX, store management, merchandising strategies,
and offering them new value at a cost that they are able to pay.
References
Accenture, 2020. Artificial intelligence in retail: Scale at speed, Retail Applied Intelligence,
Research Report. [online] Available at: <https://www.accenture.com/us-
en/insights/retail/artificial-intelligence-retail-value> [Accessed 15 August 2020].
Anana, E. and Walter Nique, W., 2010. Perception-Based Analysis: An innovative approach
for brand positioning assessment, Journal of Database Marketing & Customer Strategy
Management, [e-journal] 17, pp. 6-18. DOI: https://doi.org/10.1057/dbm.2009.32
Anderson, G., 2019. How long before Amazon launches its fleet of drones? RetailWire.
[online] Available at: <https://www.retailwire.com/discussion/how-long-before-amazon-
launches-its-fleet-of-drones/> [Accessed 16 August 2020].
Atmar, H., Begley, S., Fuerst, J., Rickert, S., Slelatt, R. and Tjon Pian Gi, M., 2020. The next
normal: Retail M&A and partnerships after COVID-19. [pdf] McKinsey & Company.
[online] Available at: <https://www.mckinsey.com/business-functions/m-and-a/our-
insights/the-next-normal-retail-m-and-a-and-partnerships-after-covid-19> [Accessed 22
July 2020].
Attar, W.E., 2020. Autonomous warehousing solutions. Supply Chain Analysis -IGD [online]
Available at: <https://supplychainanalysis.igd.com/topics/inventory-management/news-
article/t/autonomous-warehousingsolutions/> [Accessed 3 August 2020].
Bauer, J.C., Kotouc, A.J. and Rudolph, T., 2012. What Constitutes a “Good Assortment”? A
Scale for Measuring Consumers’ Perceptions of an Assortment Offered in a Grocery
Category. Journal of Retailing and Consumer Services, [e-journal] 19 (1), pp. 11-26.
DOI: 10.1016/j.jretconser.2011.08.002.
Begley, S., Coggins, B., Maloney, M. and Noble, S., 2020. The next normal in retail: Charting
a path forward, Retail and Consumer Packaged Goods Practices. [pdf] McKinsey &
Company. [online] Available at: <https://www.mckinsey.com/industries/retail/our-
insights/the-next-normal-in-retail-charting-a-path-forward> [Accessed 22 July 2020].
Bendle, N.T., Farris, P.W., Pfeife, P.E. and Reibstein, D.J., 2016. Marketing Metrics. The
Manager’s Guide to Measuring Marketing Performance. 3rd ed. New Jersey: Pearson.
Brady, D., Gregg, B. and Kim, A., 2020. Rapid revenue recovery after the crisis: Strategies
for success. [online] Available at: <https://www.mckinsey.com/business-
functions/marketing-and-sales/our-insights/rapid-revenue-recovery-after-the-crisis-
strategies-for-success > [Accessed 6 November 2020].
Briedis, H., Kronschnabl, A., Rodriguez, A. and Ungerman, K., 2020. Adapting to the next
normal in retail: The customer experience imperative. [online] Available at:
<https://www.mckinsey.com/industries/retail/our-insights/adapting-to-the-next-normal-
in-retail-the-customer-experience-imperative> [Accessed 29 July 2020].
Bughin, J., Seong, J., Manyika, J., Chui, M. and Joshi, R., 2018. Notes from the frontier:
Modeling the impact of AI on the world economy. Discussion Paper McKinsey Global
Institute. [online] Available at: <https://www.mckinsey.com/featured-insights/artificial-
intelligence/notes-from-the-frontier-modeling-the-impact-of-ai-on-the-world-
economy?> [Accessed 29 July 2020].
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
28 Amfiteatru Economic
Buhaescu, O., 2020. Retail-ul în timpul pandemiei: cum se pot transforma vulnerabilitățile
în oportunitati. Deloitte Romania. [online] Available at: <https://www2.deloitte.com/ro/
en/pages/business-continuity/articles/retail-ul-in-timpul-pandemiei-cum-se-pot-
transforma-vulnerabilitatile-in-oportunitati.html?> [Accessed 21 August 2020].
Business Wire, 2020. Global Intelligent Virtual Assistant Market (2020 to 2027) - Size, Share
& Trends Analysis Report - ResearchAndMarkets.com. [online] Available at:
<https://www.businesswire.com/news/home/20200422005328/en/Global-Intelligent-
Virtual-Assistant-Market-2020-2027> [Accessed 13 August 2020].
Capgemini, 2018. The Secret to Winning Customers’ Hearts with Artificial Intelligence: Add
Human Intelligence. [online] Available at: <https://www.capgemini.com/ca-
en/capgemini-week-of-innovation-networks-2018/technovision-2018-the-impact-of-ai/>
[Accessed 17 August 2020].
Capgemini, 2020. COVID-19 and the age of the contactless customer experience: Winning
the trust of consumers in a no-touch world. [online] Available at:
<https://www.us.sogeti.com/explore/research/reports/covid-19-and-the-age-of-the-
contactless-customer-experience/> [Accessed 17 August 2020].
CBRE, 2015. How Active is the Romanian Retail Market? [pdf] CBRE Romania Research.
Available at: <https://nrcc.ro//56959cbre romania_how-active-retail-market.pdf>
[Accessed 21 August 2020].
Charm, T., Perrey, J., Poh, F. and Ruwadi, R., 2020. 2020 Holiday Season: Navigating shopper
behaviors in the pandemic. McKinsey & Company Report. [online] Available at:
<https://www.mckinsey.com/business-functions/marketing-and-sales/solutions/periscope/
our-insights/surveys/2020-holiday-season-navigating-shopper-behaviors-in-the-
pandemic> [Accessed 6 November 2020].
Chen, C., 2020. On-device Supermarket Product Recognition. Google AI Blog, [blog] 11
August. Available at: <https://ai.googleblog.com/2020/07/on-device-supermarket-
product.html> [Accessed 20 August 2020].
Colliers, 2020. COVID-19 Survey: How does COVID-19 affect the retail market in Romania?
[online] Available at: <https://www2.colliers.com/en-ro/research/colliers-romania-
covid-19-retail-survey-2020> [Accessed 21 August 2020].
Communication from the Commission to the Council, the European Parliament, the
Economic and Social Committee and the Committee of the Regions, Commission of the
European Communities, Brussels, COM (97) 157 final A European Initiative in
Electronic Commerce. [pdf] Available at: <https://eur-lex.europa.eu/LexUriServ/
LexUriServ.do?uri=COM:1997:0157:FIN:EN:PDF> [Accessed 15 August 2020].
Consiliul Concurenţei, 2018. Raport al investigaţiei privind sectorul comerţului electronic -
componenta referitoare la strategiile de marketing. [pdf] Direcţia Cercetare. [online]
Available at: <http://www.consiliulconcurentei.ro/wp-content/uploads/2020/01/raport_
al_investigatiei_privind_sectorul_comertului_electronic.pdf> [Accessed 21 August
2020].
Consiliul Concurenţei, 2020. Raportul “Evoluția concurenței în sectoare cheie - 2020”.
[pdf] Direcţia Cercetare. Available at: <http://www.consiliulconcurentei.ro/wp-
content/uploads/2020/11/Raport-Sectoare-Cheie-2020.pdf> [Accessed 19 November
2020].
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 29
Cristea, M., 2020. PENNY products delivered for the first time directly to home via Lyvra.
Business Review, [online] Available at: <https://business-review.eu/business/retail/
penny-products-delivered-for-the-first-time-directly-to-home-via-lyvra-212391>
[Accessed 21 August 2020].
DataRobot, 2020. The Empowered Consumer. [online] Available at:
<https://www.datarobot.com/blog/the-empowered-consumer/> [Accessed 29 July 2020].
Delighted, 2020. Delighted’s 2021 retail customer experience guide: 3 must-embrace CX
trends. [online] Available at: <https://delighted.com/guides/retail-customer-experience-
2021> [Accessed 25 November 2020].
Deloitte Romania, 2020. Romanian Consumer Trends. [online] Available at:
<https://www2.deloitte.com/ro/en/pages/about-deloitte/consumer/romanian-consumer-
trends.html> [Accessed 21 August 2020].
Desai, P., Potia, A. and Salsberg, B. (2017). Retail 4.0: The Future of Retail Grocery in a
Digital World. [pdf] McKinsey & Company. Available at:
<https://www.mckinsey.com/~/media/mckinsey/dotcom/client_service/retail/articles/the
_future_of_retail_grocery_in_digital_world%20(3).pdf> [Accessed 15 August 2020].
Directorate General for Competition, COM (96) 721, Paper on Vertical Restraints in EC
Competition Policy. [pdf] Available at: <https://ec.europa.eu/competition/
antitrust/others/96721en_en.pdf> [Accessed 15 August 2020].
Dogtiev, A., 2018. App Download and Usage Statistics 2017. [online] Available at:
<http://www.businessofapps.com/data/app-statistics/> [Accessed 15 August 2020].
Ducrocq, C., 2014. Distribution. Inventer le commerce de demain. S.l: Pearson France.
Ellis, B., 2014. Real-Time Analytics: Techniques to Analyze and Visualize Streaming Data.
[e-book] Wiley. Available at: Wiley <ttps://www.wiley.com/en-
us/Real+Time+Analytics%3A+Techniques+to+Analyze+and+Visualize+Streaming+Da
ta-p-9781118837917> [Accessed 14 August 2020].
European Commission, 2020. On Artificial Intelligence - A European approach to excellence
and trust. White Paper, COM (2020) 65 final. [online] Available at:
<https://ec.europa.eu/info/sites/info/files/commission-white-paper-artificial-
intelligence-feb2020_en.pdf> [Accessed 16 August 2020].
Fergusson, F., 2020. Are Retail Robots Out to Get Us? [online] Available at:
<https://www.vendhq.com/blog/retail-robots/> [Accessed 29 July 2020].
Hair, J.F.Jr., Hult, G.T.M., Ringle, C.M. and Sarstedt, M., 2017. A primer on partial least
squares structural equation modelling (PLS-SEM). 2nd ed. Los Angeles, CA: SAGE.
IBM, 2019. Build Your Trust Advantage. Leadership in the era of data and AI everywhere.
[online] Available at: <https://www.ibm.com/downloads/cas/K1OGEMA9> [Accessed
11 August 2020].
IBM, 2020. Building the Cognitive Enterprise: Nine Action Areas. [online] Available at:
<https://www.ibm.com/downloads/cas/JKJA41PW> [Accessed 11 August 2020].
Ioan-Franc, V. ed., 1998. TEZAUR. [pdf] Bucharest: Academia Română, Institutul Naţional
de Cercetări Economice “Costin C. Kiriţescu”. Available at: <http://www.cide.ro/
tezaur%20XLVII.pdf> [Accessed 16 August 2020].
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
30 Amfiteatru Economic
Johnson, L., 2019. How to Create a Model of Your Customer’s Mind. [online] Available at:
<https://marketingexperiments.com/conversion-marketing/how-to-create-a-model-of-
your-customers-mind> [Accessed 21 August 2020].
Kanada, P., 2020. How to Use Artificial Intelligence in Mobile Apps. [online] Available at:
<https://readwrite.com/2020/09/13/how-to-use-artificial-intelligence-in-mobile-apps/>
[Accessed 5 December 2020].
Loeb, W., 2020. More Than 13,200 Stores Are Closing In 2020 So Far – A Number That
Will Surely Rise. Forbes, [online] Available at: <https://www.forbes.com/sites/
walterloeb/2020/07/06/9274-stores-are-closing-in-2020--its-the-pandemic-and-high-
debt--more-will-close/#60d62da1729f> [Accessed 14 March 2019].
Market Research Future, 2020. Proximity Marketing Market 2020-2023: Business Trends,
COVID-19 Impact Analysis, Global Segments, Competitor Strategy, Sales, Supply,
Demand and Regional Study. MarketWatch Press Release, [online] Available at:
<https://www.marketwatch.com/press-release/proximity-marketing-market-2020-2023 –
business-trends-covid-19-impact-analysis-global-segments-competitor-strategy-sales-
supply-demand-and-regional-study-2020-07-20?> [Accessed 14 August 2020].
Marsden, P., 2017. How People Feel About Artificial Intelligence: A Research Roundup.
[online] Available at: < https://digitalwellbeing.org/how-do-consumers-feel-about-
artificial-intelligence-a-list-of-research/#comments > [Accessed 20 August 2020].
McEnally, M. and de Chernatony, L., 1999. The Evolving Nature of Branding: Consumer
and Managerial Considerations. Academy of Marketing Science Review, [online] 2, p. 19.
Available at: <http://www.amsreview.org/articles/mcenally02-1999.pdf> [Accessed 27
August 2020].
McKinsey, 2020. Kate Smaje: Why businesses must act faster than ever on digitization,
McKinsey’s Kate Smaje on set with CNBC, Digitalour People, McKinsey & Company.
[online] Available at: <https://www.mckinsey.com/about-us/new-at-mckinsey-
blog/kate-smaje-mckinsey-digital-explains-why-businesses-must-act-fast-on-
digitization> [Accessed 15 August 2020].
MECLABS Institute, 2018. How to Model Your Customer’s Mind. [pdf] Available at:
<https://images.meclabs.com/uploads/content/How%20to%20Model%20Your%20Cust
omers%20Mind.pdf > [Accessed 21 August 2020].
MIT Technology Review Insights, 2018. Humans + bots: Tension and opportunity. [online]
Available at: <https://www.technologyreview.com/2018/11/14/239924/humans-bots-
tension-and-opportunity/> [Accessed 28 November 2020].
Mortari, L., 2015. Reflectivity in Research Practice: An Overview of Different Perspectives,
International Journal of Qualitative Methods, [e-journal] 14(5).
https://doi.org/10.1177/1609406915618045.
Nielsen, 2019. 5G, AR and frictionless commerce: Nielsen illuminates future of retail and
consumer packaged goods. [online] Available at: <https://www.nielsen.com/us/en/press-
releases/2019/5g-ar-and-frictionless-commerce-nielsen-illuminates-future-of-retail-and-
consumer-packaged-goods/> [Accessed 20 August 2020]
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 31
Păunescu, A., 2020. Situația actuală din eCommerce-ul românesc în contextul COVID-19.
Iulian Stanciu, CEO eMAG, în dialog deschis cu GPeC. GPeC Blog, [blog] 25 March.
Available at: <https://www.gpec.ro/blog/situatia-actuala-din-ecommerce-ul-romanesc-
in-contextul-covid-19-iulian-stanciu> [Accessed 21 August 2020].
Profero&iSense Solutions, 2020. Studiu BOLD by Profero&iSense Solutions: Creativity
meets A.I. 6 modalități în care brandurile pot folosi inteligența artificială pentru eficiență
și inovație în noul context. [online] Available at: <https://www.isensesolutions.ro/studiu-
bold-by-proferoisense-solutions-creativity-meets-a-i-6-modalitati-in-care-brandurile-
pot-folosi-inteligenta-artificiala-pentru-eficienta-si-inovatie-in-noul-context/>
[Accessed 21 August 2020].
Purcărea, T., 1994. Management comercial. Bucharest: Editura Expert.
Purcărea, T. and Ioan-Franc, V., 1999. Partenariat consommateur – entreprise productrice,
Observer. Romanian Economic Research, no. 9, C.I.D.E., Academie Roumaine.
Purcărea, T. and Ioan-Franc, V., 2000. MARKETING. Evoluţii, experienţe, dezvoltări
conceptuale. Bucharest: Editura Expert.
Purcarea, T. and Purcarea, A., 2008. Distribution in Romania at the shelf supremacy’s
moment of truth: competition and cooperation. Amfiteatru Economic, 10(24), pp. 9-25.
Purcarea, T. and Ioan-Franc, V., 2008. Operaţionalizarea transferului de cunoaştere şi
competitivitatea sectorului distribuţiei bunurilor de larg consum. Probleme Economice.
Institutul Naţional de Cercetări Economice (INCE), Academia Română, Vol. 307-308.
Purcarea, T. and Ratiu, M., 2010. The ongoing challenge: How to remain competitive in the
global service economy. Bucharest: Carol Davila University Press.
Purcarea, T., 2015. Road Map for the Store of the Future, World Premiere, May 4, 2015, at
SHOP 2015, Expo Milano 2015. Romanian Distribution Committee Magazine, 6(2),
pp. 36-45.
Purcarea, T., Purcarea, I. and Purcarea, A., 2018. The Impact of an Improved Smartphone
App’s User Experience on the Mobile Customer Journey on the Romanian Market. In:
Bucharest University of Economic Studies, International Conference on Economics and
Social Sciences. Bucharest, Romania, 16-17 April 2018. Bucharest: Bucharest University
of Economic Studies.
Purcarea, T., 2019. Retailers’ Reinvention in Harmony with the Shopping Tendencies.
Romanian Distribution Committee Magazine, 10(2), pp. 36-46.
Rajendran, R., 2020. Personalize retail with AI and robotics for spot-on recommendations.
[online] Available at: <https://www.capgemini.com/resources/personalize-retail-with-ai-
and-robotics-for-spot-on-recommendations/> [Accessed 17 August 2020].
Ristea, A.L., Ioan-Franc, V. and Purcărea, T., 2005. Economia distribuţiei. Bucharest:
Editura Expert.
Roşca, C., 2020. Cei mai mari jucători din economie. Topul reţelelor de comerţ. Business
Magazin, [online] Available at: <https://www.businessmagazin.ro/actualitate/cei-mai-
mari-jucatori-din-economie-topul-retelelor-de-comert-19353069> [Accessed 21 August
2020].
Saker, B., 2020. The Impact of Artificial Intelligence in Retail. Total Retail, [online]
Available at: <https://www.mytotalretail.com/article/the-impact-of-artificial-
intelligence-in-retail/> [Accessed 15 August 2020].
AE The Profound Nature of Linkage Between the Impact of the Use of Artificial Intelligence in Retail on Buying and Consumer Behavior and Consumers’
Perceptions of Artificial Intelligence on the Path to the Next Normal
32 Amfiteatru Economic
Sneader, K. and Singhal, S., 2020. Beyond coronavirus: The path to the next normal. [online]
Available at: <https://www.mckinsey.com/industries/healthcare-systems-and-services/
our-insights/beyond-coronavirus-the-path-to-the-next-normal> [Accessed 21 August
2020].
Swerdlow, F. and Cyr, M., 2020. Future of Retail 2020: Retail Industry CIOs Will Invest in
Automation, Data, And Employees. [online] Available at: <https://go.forrester.com/
blogs/future-of-retail-2020-retail-industry-cios-will-invest-in-automation-data-and-
employees/ [Accessed 14 August 2020].
Twilio, 2020. Creating Consumer Impact in Retail. [e-book] Available at:
<https://pages.twilio.com/rs/294-TKB-300/images/Creating_Consumer_Impact_E-
book-TB-FIN-4441.pdf> [Accessed 3 August 2020].
USDA, 2020. Romania: COVID-19 Transforms Romanian Retail and Food Service Sectors.
USDA Foreign Agricultural Service, Attaché Reports. [online] Available at:
<https://www.fas.usda.gov/data/romania-covid-19-transforms-romanian-retail-and-
food-service-sectors> [Accessed 21 August 2020].
Weise, E., 2018. Amazon opens its grocery store without a checkout line to the public. USA
TODAY, 21 Jan.
White, A., 2020. What is Real Time Analytics? – A Complete Overview. [online] Available
at: <https://www.izenda.com/real-time-analytics-platform-overview/> [Accessed 12
August 2020].
Worldwide Business Research, 2019. Future Stores 2019 Retail Technology Briefing.
[online] Available at: <https://futurestores.wbresearch.com/downloads/future-stores-
2019-retail-technology-briefing> [Accessed 14 March 2019].
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 33
THE IMPACT OF ARTIFICIAL INTELLIGENCE ON CONSUMERS’ IDENTITY
AND HUMAN SKILLS
Corina Pelau1, Irina Ene2 and Mihai-Ionuț Pop3 1)2)3) Bucharest University of Economic Studies, Romania
Please cite this article as:
Pelau, C., Ene, I. and Pop, M.I., 2021. The Impact of
Artificial Intelligence on Consumers’ Identity and
Human Skills. Amfiteatru Economic, 23(56),
pp. 33-45.
DOI: 10.24818/EA/2021/56/33
Article History
Received: 30 September 2020
Revised: 10 November 2020
Accepted: 8 December 2020
Abstract
The development of artificial intelligence is one of the main paradigms of the contemporary
society, which will radically change the existence of individuals and our society and it will
have important effects on the economy. The use of artificial intelligence in the daily work of
individuals and in the relationship between companies and consumers has a great number of
advantages such as the increased efficiency, a high degree of fascination in interaction, but
in the same time there are several fears related to its development in the future. Due to its
great data storage capacity about the behavior of individuals and the processing speed of this
data, there is a risk that the forms of artificial intelligence will become smarter than humans
and thus intervene in the decisions made by them. Through the constant use of artificial
intelligence, there is a high risk of manipulation of consumers as well as a high degree of
dependence on intelligent technologies. This close relationship between the user and artificial
intelligence can reduce an individual’s cognitive abilities and can affect their thinking,
personality and relationships with its social circle. This paper presents a mediation model
between the efficiency and fascination with artificial intelligence and the consumers’
perception of preserving their self-identity and human skills, having as mediator the influence
and model of the social circle. The research results show that a higher degree of efficiency
and fascination, as well as a positive influence from the social circle decrease the consumers’
perception of reduction of human skills in relation to artificial intelligence. Moreover, the
social circle mediates the relationship between efficiency and fascination produced by
artificial intelligence and the perception of preserving human abilities.
Keywords: artificial intelligence, robots, consumers, social circle, consumers’ self-identity,
human abilities
JEL Classification: M21, M31
Corresponding author, Corina Pelau – e-mail: [email protected]
AE The Impact of Artificial Intelligence on Consumers’ Identity and Human Skills
34 Amfiteatru Economic
Introduction
Our daily life is increasingly dominated by the presence of various forms of artificial
intelligence, which accompany us in almost all activities. From finding the direction with the
help of GPS, to monitoring our heart rate during sports activities, we always have a smart
device that monitors our activity and makes recommendations for a better life. This trend of
automation is also found in the relationship with companies that provide us with various
products and services. For example, to communicate better with the mobile phone, we have
Siri for Apple phone users (Apple, 2020) or Bixby for Samsung users (Samsung, 2020), while
for banking services we communicate with George as the bank’s interface (BCR, 2020). This
trend of using robots or artificial intelligence systems is increasingly present in the daily lives
of consumers and in the relationship they have with various companies. In order to better
understand this interaction, it is important to know the advantages and disadvantages that the
development of artificial intelligence has in contemporary society.
The objective of this paper is to determine the relationship between the influence of efficiency
and fascination with artificial intelligence, the influence of the social circle and the perception
of preserving the identity of consumers in relation to various forms of artificial intelligence.
The presence of devices equipped with artificial intelligence in the daily lives of individuals
and consumers has a number of advantages and disadvantages. In order to integrate these
devices as much as possible, it is important to understand both the advantages and the
disadvantages that they bring to people's lives. In this paper we will analyse the relationship
between the benefits of artificial intelligence characterized by an increased efficiency and the
fascination created by them and the main fears related to the human abilities of consumers.
There will be also analysed the multiplying effect of the role of the social circle on the
benefits created by artificial intelligence, as well as its impact in reducing the fears associated
to artificial intelligence. Most of the recent studies related to the acceptance of new
technologies refer mainly to the relationship between the individual and the technology
(Davis et al., 1989; Venkatesh et al., 2012) and they focus less on the influence of the social
circle on the perception of artificial intelligence. Although the influence of the social circle
has been included in models related to the use of artificial intelligence, this has been more
correlated with the evaluation of how to use them and less on the reduction of fears related
to the use of artificial intelligence (Gursoy et al., 2019). The paper is structured in five parts.
The first part presents the literature review with emphasis on the main variables used in the
subsequent model. The second part presents the research methodology and the collection of
data. The next two parts are dedicated to the confirmatory factor analysis for determining the
values of the variables and the testing of the mediation model based on the previously
validated variables. The last part is dedicated to the discussions and conclusions based on the
results of the mediation model.
1. Literature review on the role of artificial intelligence in economics
The development of various forms of artificial intelligence and robots has been one of the
biggest challenges of contemporary society. On one hand, the development of smart
technologies contributes to the efficiency of the various daily activities of consumers and
companies, thus shifting to a better quality of life (Pelau and Ene, 2018). On the other hand,
the presence and integration of artificial intelligence into people’s daily lives has long been
debated (Kaplan and Haenlein, 2020).
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 35
One of the main topics of debate related to artificial intelligence is the degree of trust and
acceptance of artificial intelligence by users (Hengstler et al., 2016). Several authors
emphasize the importance of initial trust in these new technologies (McKnight et al., 2002;
Lee and See, 2004), but also the way of integrating them into the daily work (MacVaugh and
Schiavone, 2010) of individuals. Previous research has shown that the consumers’ acceptance
of artificial intelligence and robots depends on the specific situation in which they are used.
For instance banking robots are much more easily accepted by the consumer in comparison
to intelligent cooking or smart legal consulting devices (Pelau and Ene, 2020). There are
several factors that affect the general acceptance or rejection of robots. The increased
performance (Gursoy et al., 2019; Lu et al., 2019), the social pressure for using intelligent
devices (Hall and Henningsen, 2008; Hsu and Lin, 2008), the entertainment and fascination
created by the device (Fryer et al., 2017) as well as the easiness of its use (Law et al., 2018;
Allam et al., 2019) are just some of the factors affecting the willingness of using intelligent
devices.
Higher performance refers to the situation where robots or artificial intelligence systems can
provide more services in a better, more efficient or faster way than humans (Lu et al., 2019).
By doing so, robots increase people’s comfort and well-being and increase companies'
profits. The involvement of artificial intelligence or robots in the commercial activity leads
to an increased efficiency of processes, due to their greater capacity to store information
compared to human personnel, which leads to a faster processing of orders and a greater
degree of personalization for customers (West et al., 2018). Hedonic pleasure and fascination
refer to the positive feelings associated with the use of robots and the satisfaction of using
new technologies for personal interests and entertainment (Fryer et al., 2017). Several studies
consider hedonic motivation as one of the main factors that affect in a positive way and are
an important facilitator of consumers’ willingness to use new types of technologies such as
the artificial intelligence (Law et al., 2018; Allam et al., 2019). The fascination towards
artificial intelligence is important in the use of robots as an interface for communication and
interaction with customers (Marinova et al., 2017; Wirtz et al., 2018). If artificial intelligence
is accepted as a mean of interaction, a large number of service employees can be replaced by
robots, thus increasing the efficiency and profitability of companies. For this reason, the
perceived efficiency as well as the consumers’ fascination towards the use or interaction with
forms of artificial intelligence are important factors for their acceptance and are included in
hypothesis I1 of this study, according to which a high degree of efficiency and fascination
towards the artificial intelligence will increase the social pressure of using artificial
intelligence and robots.
The effort involved in using robots describes the perceived difficulty of the consumer in using
a new technology or a new form of artificial intelligence. In this regard, several theories have
been developed that describe how to accept new technologies (Davis et al., 1989; Venkatesh
et al., 2012). Several studies have shown that the difficulty depends on the ability of
consumers to learn how to use these new technologies (Kim and Baek, 2018; Lu et al., 2019;
Gursoy et al. 2019; Ashfaq et al., 2020).
The social circle is another important factor that influences the adoption of a technology. If
more people in a social group use a new technology, the individual will be forced to use that
technology (Hall and Henningsen 2008; Hsu and Lin, 2008). Consumers tend to adopt a new
product or technology if the social group they belong to or want to join appreciate that product
or technology. Especially when a consumer has little information about a new technology, it
AE The Impact of Artificial Intelligence on Consumers’ Identity and Human Skills
36 Amfiteatru Economic
tends to imitate the behavior of the social group to which it belongs (Venkatesh et al., 2012;
Dabija et al., 2017; Gursoy et al., 2019). Due to the fact that the influence of the social circle
is an important validation factor of a new technology, it is included in hypothesis I2 of the
research presented in this article, which postulates that the impact of the social circle will
decrease the perception of losing its own identity because of the use of artificial intelligence.
In opposition to this, the consumers also have negative feelings associated to robots because
of their capacity of manipulating its owner, the deterioration of social relations and the threat
of losing its own self-identity (Kaplan and Haenlein, 2020). One of the main fears associated
with artificial intelligence is the threat of manipulation. The use of robots and artificial
intelligence is closely linked to the collection of data by the system. A consumer can use a
robot or a smart system only if he or she is willing to provide a series of personal data and
commands to the system. Given that the robot or artificial intelligence will gather more and
more information about its user, it will have the ability to learn from experience and will have
the ability to influence the people's decisions (Kaplan and Haenlein, 2020). The question here
is to what extent people will have the power to control these artificial intelligence systems or
whether they will make their own decisions. There are already theories that forms of artificial
intelligence can become smarter than their creators, through their unlimited ability to learn
and store information and thus make decisions for their own interest and not in their owners’
interest (Rinesi, 2015; Kaplan and Haenlein, 2020). Opinions in this regard are divided. Some
authors believe that it will be a long time before robots have their own motivation and
reasoning (Haladjian and Montemayor, 2016), while others consider it to be just a challenge
of intelligent engineering (Graziano, 2015). Another fear related to the development of
artificial intelligence is the loss of the self-identity and human abilities of individuals. The
increasingly pronounced anthropomorphic characteristics of robots and our willingness to
give them their own identity (by giving them names and by providing them human rights)
will increase the social power of robots and may threaten people’s self-identity.
Anthropomorphic characteristics refer to the physical, mental and behavioral characteristics
of a robot that mimics human features (Kim and McGill, 2018). On one hand, consumers
may consider the human aspect to be friendlier. On the other hand, an anthropomorphic
appearance could threaten people’s self-identity and distinctiveness (Rosenthal-von der
Pütten and Krämer, 2014). Moreover, there is a growing trend of human emotional
dependence on machines or robots. Many authors predict an increased addiction to the
emotional and physical relationship with gender humanoid robots (Pfadenhauer, 2015;
Gonzales-Jimenez, 2018). The use of artificial intelligence and machines in communication
between consumers also increases the risk of deteriorating social relations. Communication
takes place more and more with the help of machines (smartphones, laptops, tablets) to the
detriment of human contact.
The use of artificial intelligence in companies will have a positive impact on their profits, but
at the same time more people will lose their jobs, as tasks currently performed by humans
will be replaced by artificial intelligence and robots. This creates a sense of uncertainty
among people and it is one of the main reasons for not accepting automation projects. The
threat of replacing humans with robots is described in a model proposed by Huang & Rust
(2018) based on the degrees and types of intelligence of robots. They propose several models
of co-integration of humans and robots at the workplace and in the provision of services.
They analyse several models of co-existence, ranging from the situation in which robots can
perform tasks that humans do not want to perform, to the situation in which humans and
robots are equal co-workers. It also describes a hypothetical situation in which artificial
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 37
intelligence is seen as an extension of humans and they coexist through integration (Huang
and Rust, 2018). For this reason, the preservation of human abilities, following the use and
interaction with the forms of artificial intelligence is included in hypothesis I3 of this study
and in the mediation model, by which a degree of efficiency and fascination towards artificial
intelligence decreases the perception of losing its own identity because of the use of artificial
intelligence and robots.
The use of artificial intelligence and robots in the daily lives of consumers is associated with
positive emotions and attitudes, but also with fears and negative feelings about the directions
in which they will develop. In this paper we propose a mediation model that tests the extent
to which positive aspects of efficiency and fascination towards robots alleviate negative
feelings about the loss of identity and human skills in the relationship and interaction with
various forms of artificial intelligence.
2. Methodology
The main scope of this research is to determine the relationship between efficiency and
fascination of users towards artificial intelligence, the influence of the social circle and the
pressure to use various forms of artificial intelligence and the perceived threat of losing self-
identity and human skills through the use of artificial intelligence and robots. For this scope,
there have been defined several objectives and a mediation model has been designed, having
the following hypotheses:
The first objective refers to the influence of the perceived benefits of using artificial
intelligence, which will increase the number of those who use such devices in the everyday
life. As more people use artificial intelligence devices, there will be a social pressure to
expand their use. Therefore the first hypothesis is formulated as follows:
1st Hypothesis (I1): Increased efficiency and fascination towards artificial intelligence
increases social pressure to use forms of artificial intelligence and robots.
The second objective refers to the impact that the spread of artificial intelligence has on the
consumers’ fears, namely to reduce human skills because of the use of artificial intelligence.
The more the use of artificial intelligence becomes a common phenomenon, the more it will
become a common activity and it will become part of the daily routine of consumers and
individuals in general. For this purpose, the second hypothesis has been formulated as
follows:
2nd Hypothesis (I2): The impact of the social circle reduces the perception of losing self-
identity while using artificial intelligence.
The third objective refers to the direct influence that the advantages brought by artificial
intelligence have on the reduction of human abilities. The benefits perceived by the consumer
reduce the fears that they have in relation to artificial intelligence and increase the positive
perception of their acceptance. For this, the third hypothesis was formulated:
3rd Hypothesis (I3): An increased degree of efficiency and fascination towards artificial
intelligence reduces the perception of losing self-identity caused by the use of artificial
intelligence and robots;
AE The Impact of Artificial Intelligence on Consumers’ Identity and Human Skills
38 Amfiteatru Economic
The last objective integrates the relations presented in the previous objectives and postulates
the mediation model, formulated in the following hypothesis:
4th Hypothesis (I4): The impact of the social circle mediates the relationship between the
perceived degree of efficiency and fascination towards artificial intelligence and the
perception of losing self-identity through the use of various forms of artificial intelligence.
(Figure no. 1)
Figure no. 1: Proposed mediation model
In order to confirm this model from an empirical point of view, a quantitative research has
been carried out which had as objective the evaluation of the three variables included in the
mediation model. The questionnaire included 24 items with statements about the users'
attitudes and perceptions about the efficiency and fascination experienced while using
artificial intelligence, the influence of the social circle on the peer pressure of using artificial
intelligence and the perception of losing self-identity and human abilities by using artificial
intelligence. The evaluation of the 24 statements was performed using a Likert scale with
values between 1 and 7, 1 representing total disagreement, and 7 representing total
agreement. The research took place in December 2019 and the subjects have been selected
randomly from the adult urban population, with the condition that there is a homogeneous
distribution by gender. The sample contained 740 valid responses and has a total Cronbach-
Alpha value of 0.901. The structure of the questionnaire includes similar values for women
(50.5%) and men (49.5%). The majority of the subjects have ages between 20 and 30 years
(61.6%), being complemented by subjects with ages between 30 and 40 years (14.3%), 40
and 50 years (7.5%), 50 and 60 years (11.6%), more than 60 years (1.8%) and less than 20
years (2.9%).
In order to determine the inclusion of the items to the three variables, a confirmatory factor
analysis has been performed. Based on the obtained results, the values of the variables
included in the model in figure no. 1 have been calculated and the hypotheses and the
mediation model have been tested. The results of the two analyses are presented in the
following chapters.
3. Results of the confirmatory factor analysis
The value of the Kaiser-Meyer-Olkin indicator KMO = 0.896 as well as the value of
p = 0.000 of the Bartlett test show an appropriate adequacy of the items used for a factor
analysis. The analysis of the loadings of items indicates an ideal number of 3 factors as it can
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 39
be seen in table no. 1. It should be noted that in order to determine the three factors, all items
with values greater than 0.600 have been taken into account.
The first factor contains items related to the high degree of efficiency obtained from the use
of artificial intelligence (4 items), to improving the quality of life (3 items), about the
fascination of using artificial intelligence (2 items), the desire to use a robot or some form of
artificial intelligence (3 items) and the desire to learn commands for the use of artificial
intelligence (1 item). This factor contains the largest number of items (13), due to the
correlation between the statements related to efficiency, fascination, and desire to use and
learn commands for artificial intelligence. The Cronbach-Alpha value α = 0.912 shows a high
relevance of the items in this factor. The average value of the items is M = 4.906.
Table no. 1: The results of the confirmatory factor analysis
1st
Factor
2nd
Factor
3rd
Factor
The robot is more efficient in carrying out activities .706
The robot has a higher accuracy in performing tasks .704
There are fewer errors if the tasks are performed by robots .605
The robot performs activities more quickly .629
The activities performed by the robot make my life easier .687
I have more free time because of the robot .672
I can focus on complex activities, if the robot helps me with
certain tasks .667
The interaction with the robot is fun .700
The interaction with the robot is fascinating due to the degree of
novelty .637
I am willing to learn the necessary commands to optimize the
activity with a robot .647
I like to interact with a robot that helps me in daily activities .727
I want to have a robot to help me in my daily activities .702
I am willing to buy a robot to help me in my daily activities .651
All my friends have robots that help them in their daily activities .621
All the people I appreciate use robots for their daily activities .626
I don't think I'm addicted to robots .645
I consider that the activity with the robot does not affect my
personality .728
I believe that working with the robot does not reduce my human
abilities .705
The second factor includes 3 items related to the dependency on artificial intelligence, the
reduction of human abilities and personality determined by the use of robots. All items in
this factor are formulated with the help of a negation, so a higher value indicates a greater
disagreement. For this reason the variable was called “reducing the degree of self-identity
caused by the use of artificial intelligence”. It should be noted that items related to
communication and interaction with others do not correlate in this factor. The Cronbach-
Alpha value α = 0.834 shows a good adequacy of the items for this factor. The average value
of the items is M = 5.031. This variable can be also found in other studies such as Huang and
Rust (2018) or Kaplan and Haenlein (2020). Both articles present a slightly pessimistic
variant of the direction of development of artificial intelligence, by which they aim to draw
a signal to the dangers that may occur from their evolution.
AE The Impact of Artificial Intelligence on Consumers’ Identity and Human Skills
40 Amfiteatru Economic
The third factor contains two items related to the way in which the social circle uses robots
and other forms of artificial intelligence. It is interesting to note that this factor correlates
only based on the model taken or imitated from others, namely “all my friends” and “all the
people I appreciate use robots in daily activities” and does not include items related to how
the respondent is appreciated by others. For this factor Cronbach-Alpha has a value of α =
0.867, proving a good fit, and the average of the items is M = 3.182.
Out of all items, only 6 statements did not have loadings greater than 0.6 for any of the three
factors resulted from the confirmatory factor analysis, so they were not used in the subsequent
mediation model. In spite of the fact that that one of the most common ways to use intelligent
devices is found in the consumers’ way of communication, by using smartphones or smart
tablets, the two items related to the interference of artificial intelligence in interpersonal
communication does not correlate in any of the determined factors. This situation requires
further investigation in future research.
4. Results of the mediation model and discussions
Based on the three variables obtained in the confirmatory factor analysis, the mediation
model has been developed with efficiency and fascination towards artificial intelligence as
an independent variable, the influence of the social circle as a mediator and the perception of
losing self-identity by using artificial intelligence as a dependent variable. The mediation
model has been tested using the Bootstrapping method based on 5000 distinct samples for a
confidence interval of 0.95. This has been done by using the Process Macro, developed by
Hayes (2018) in SPSS 20.0. The obtained results can be observed in Figure no. 2.
Figure no 2: Final mediation model
It can be observed that there is a significant positive a-path relationship between the
independent variable efficiency and fascination towards artificial intelligence and the
mediator influence of the social circle, having βa=0.593, p=0.000 and the confidence interval
CIa = [0.488; 0697]. An increased degree of efficiency and fascination towards artificial
intelligence creates models in society that use these new forms of technology and can become
an amplifier in the way artificial intelligence is promoted and accepted. This confirms
hypothesis I1 that a high degree of efficiency and fascination towards artificial intelligence
increases the social influence of using forms of artificial intelligence and robots.
There is also a significant positive relationship for the b-path, between the mediator and the
dependent variable related to losing self-identity by using artificial intelligence. This
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 41
relationship has a coefficient βb = 0.072, p = 0.024 and a confidence interval of Cib = [0.009;
0135]. The fact that friends or admirers use forms of artificial intelligence reduces the
perception of dehumanization caused by the use of artificial intelligence. These results
confirm hypothesis I2, namely that the impact of the social circle reduces the perception of
losing self-identity while using artificial intelligence.
The direct effect (c’-path) between the independent variable and the dependent variable is
also significant, having βc’=0.390, p=0.000 and the confidence interval CIc’= [0.291; 0488].
Thus, it can be said that efficiency and fascination towards artificial intelligence positively
influence the reduction of the perception of dehumanization towards the use of artificial
intelligence. This result also confirms Hypothesis I3 Hypothesis 3, namely that an increased
degree of efficiency and fascination towards artificial intelligence reduces the perception of
losing self-identity caused by the use of artificial intelligence and robots. This result is
confirmed in other research by two different variables, namely that efficiency (Gursoy et al.,
2019; Lu et al., 2019) and fascination and hedonic pleasure (Fryer et al., 2017) positively
influence the degree of accepting artificial intelligence.
The total effect (c-path) between the independent variable and the dependent variable has
also significant values for the coefficient βc=0.433, p=0.000 and the confidence interval
CIc=[0.341; 0524]. Due to the fact that the coefficient of the total effect is higher than the
coefficient of the direct effect, both being significant, it results that there is an influence of
the mediator. The indirect mediation effect has a value of β=0.043, and its significance is
validated by the confidence interval CI =[0.005; 0.083] and the fact that the value 0 is not
included in it. This confirms the last hypothesis I4, according to which the impact of the
social circle mediates the relationship between the perceived degree of efficiency and
fascination towards artificial intelligence and the perception of losing self-identity
determined by the use of forms of artificial intelligence. Although there is not a big mediation
effect, it exists and it is significant, proving that the models from the social circle amplifies
the effect of efficiency and fascination with artificial intelligence in relation to reducing fears
about the loss of human abilities and self-identity. This influence is also confirmed by other
studies such as Hall and Henningsen (2008); Hsu and Lin (2008); Gursoy et al. (2019), in
which the social circle has an influence on the degree of acceptance of artificial intelligence.
The way of correlating the variables resulting from the confirmatory factor analysis shows
that the influence is given by the behavior patterns of the social circle rather than by the
appreciation of others. Thus, if someone in a consumers’ social circle uses artificial
intelligence, there is a chance that this behavior will be imitated.
Conclusions and discussions
The results of the mediation model confirm that an increased degree of efficiency and
fascination in the use of artificial intelligence, mediated by the positive influence of the social
circle, diminishes the perception of loss of identity and human abilities in relation to artificial
intelligence. It can be said that the benefits obtained from the use of artificial intelligence and
robots reduce the fears that the consumer has in relation to them, having a series of theoretical
and managerial implications.
From a theoretical point of view, the degree of innovation brought by this model is given by
the inclusion of the social circle influence as a mediator in the relationship between efficiency
and fascination with artificial intelligence and preservation of consumer identity. Existing
AE The Impact of Artificial Intelligence on Consumers’ Identity and Human Skills
42 Amfiteatru Economic
models, especially the technology acceptance model, refer to concepts that belong only to the
individual or the consumer and less to the rest of the world. The results of this research show
that social models of using artificial intelligence can reduce the fear of losing human identity
in relation to artificial intelligence and it is a multiplier of the perception about the efficiency
and fascination of smart devices. From a managerial point of view, in order to increase the
degree of acceptance of robots and artificial intelligence, it is beneficial to use interaction
models with robots in the communication with the consumers or with individuals in general.
For services, this can be done by presenting interaction examples between consumers and
artificial intelligence in advertising, which can be later distributed through traditional media
channels or social networks. In order to promote the use of artificial intelligence within a
company, training sessions are recommended in which this type of interaction is given as an
example. However, it is important that the implementation of artificial intelligence and robots
should be done with caution because there are a number of pessimistic scenarios in which
artificial intelligence and robots can become smarter than their creators and control their
actions. To avoid this negative direction, both consumers and policy makers need to be aware
of the potential dangers so that through clear regulations they can optimize the use of artificial
intelligence in contemporary society and in the economy.
One of the limitations of the research concerns the way in which the variables are correlated,
especially for factor 1. The initial aim of the research was to test separately the efficiency of
artificial intelligence, the consumers' fascination with it and the learning processes for their
use. Based on the confirmatory factor analysis, the first two variables, respectively efficiency
and fascination are correlated and thus were used as a single variable. For the availability of
learning commands for the use of artificial intelligence, only one item correlates with factor
1, while the rest of the items were removed from the research.
The result obtained from the mediation model has both positive and negative valences. On
one hand, consumer fears about the use of artificial intelligence can be diminished by their
benefits and the influence of the social circle. This way the consumers can be easily
convinced to use various forms of artificial intelligence. On the other hand, there is a risk that
the well-being brought by the use of artificial intelligence as well as the positive emotions
created by the fascination of their use will reduce the vigilance of consumers in relation to
the potential dangers of the development of artificial intelligence. Our recommendation is to
take advantage of the benefits of artificial intelligence, but in the same time to remain vigilant
to the potential dangers associated with this evolution. We are aware of the fact that due to a
high profitability, artificial intelligence will be increasingly present in the daily lives of
consumers and in their relationship with companies. For this reason, the benefits cannot be
ignored, but a clear regulation is needed for the use of artificial intelligence. The way of
regulating artificial intelligence can be achieved in several ways, namely through the
legislation related to the use of these technologies, through the degree of innovation of
technologies but also at the level of each user. At the legislative level, it is recommended to
develop rules for the operation and use of artificial intelligence by establishing what is
allowed and what is not allowed to be achieved with the help of robots and artificial
intelligence. Legal regulations can have a wide range of applications from ethical issues (here
we refer in particular to the use of companion robots), to issues related to their rights (for
example by granting citizenship to the robot Sophia and implicitly equal rights to humans)
and up to the degree of autonomy of robots. From a technological point of view, we refer to
the degree of innovation in the sense of the similarity to people. Of course, from a scientific
point of view it is fascinating to create robots similar to humans, with similar emotions and
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 43
feelings as humans, but in the same time, the existence of these technologies create the
temptation to use them for commercial purposes and it is difficult to predict to what extent
these robots could be controlled in the future. On a personal level, each individual must be
aware of both the benefits of a robot or a device equipped with artificial intelligence, but also
of the risks to which he is exposed by granting access to personal information. For this reason,
any consumer must be aware that artificial intelligence brings certain benefits that must serve
the individual, and that the rights of the robot are limited (for example by the possibility of
shutting down or disengaging a robot).
In conclusion, the development of artificial intelligence can be an extraordinary thing for
individuals and for humanity, if, at all levels, we will know how to manage the use of robots
for the benefit of humans. In the economic activity, artificial intelligence and robots will play
a significant role in increasing process efficiency and reducing costs, but at the same time
their implementation must be considered carefully, in order to maintain interpersonal
relationships between the company and its customers or employees.
References
Allam, H., Bliemel, M., Spiteri, L., Blustein, J. and Ali-Hassan, H., 2019. Applying a
multidimensional hedonic concept of intrinsic motivation on social tagging tools: A
theoretical model and empirical validation. International Journal of Information
Management, 45, pp. 211-222.
Apple, 2020. Siri does more than ever. Even before you ask. [online] Available at:
<https://www.apple.com/siri/> [Accessed 3 December 2020].
Ashfaq, M., Yun, J., Yu, S. and Loureiro, S.M.C., 2020. I, Chatbot: Modeling the
determinants of users’ satisfaction and continuance intention of AI-powered service
agents. Telematics and Informatics, [e-journal] https://doi.org/ https://doi.org/
10.1016/j.tele.2020.101473.
BCR, 2020. Internet Banking de la BCR, cu tot ce ai nevoie, într-un singur loc. [online]
Available at: <https://www.bcr.ro/ro/persoane-fizice/digital-banking/george> [Accessed
3 December 2020].
Dabija, D.C., Băbuţ, R., Dinu, V., Lugojan, M., 2017. Cross-Generational Analysis of
Information Searching based on Social Media in Romania. Transformations in Business
& Economics, 16(2(41)), pp.248-270.
Davis, F.D., Bogozzi, R.P. and Warshaw, P.R., 1989. User acceptance of computer
technology: A comparison of two theoretical models. Management Science, 35,
pp. 982-1003.
Fryer, L.K., Ainley, M., Thompson, A., Gibson, A. and Sherlock, Z., 2017. Stimulating and
sustaining interest in a language course: An experimental comparison of Chatbot and
Human task partners. Computers in Human Behavior, 75, pp. 461-468.
Gonzalez-Jimenez, H., 2018. Taking the fiction out of science fiction: (Self-aware) robots
and what they mean for society, retailers and marketers. Futures, 98, pp. 49-56.
Graziano, M.S.A., 2015. Build-a-brain: We could build an artificial brain that believes itself
to be conscious. Does that mean we have solved the hard problem? [online] Available at:
<https://aeon.co/essays/can-we-make-consciousness-into-an-engineering-problem>
[Accessed 24 February 2018].
AE The Impact of Artificial Intelligence on Consumers’ Identity and Human Skills
44 Amfiteatru Economic
Gursoy, D., Oscar Hengxuan Chi, O.H., Lu, L. and Nunkoo, R., 2019. Consumers acceptance
of artificially intelligent (AI) device use in service delivery. International Journal of
Information Management, 49, pp. 157-169.
Haladjian, H.H. and Montemayor, C., 2016. Artificial consciousness and the consciousness-
attention Dissociation. Consciousness and Cognition, 45, pp. 210-225.
Hall, B. and Henningsen, D.D., 2008. Social facilitation and human–computer interaction.
Computers in Human Behavior, 24(6), pp. 2965-2971.
Hayes, A.F., 2018. Introduction to Mediation, Moderation and Conditional Process Analysis
– A Regression-Based Approach. New York: Guilford Press.
Hengstler, M., Enkel, E. and Duelli, S., 2016. Applied artificial intelligence and trust-The
case of autonomous vehicles and medical assistance devices. Technological Forecasting
& Social Change, 105, pp. 105-120.
Hsu, C.L. and Lin, J.C.C., 2008. Acceptance of blog usage: The roles of technology
acceptance, social influence and knowledge sharing motivation. Information &
Management, 45(1), pp. 65-74.
Huang, M.H. and Rust, R., 2018. Artificial Intelligence in Service. Journal of Service
Research, 21(2), pp. 155-172.
Kaplan, A. and Haenlein, M., 2020. Rulers of the world, unite! The challenges and
opportunities of artificial intelligence. Business Horizons, 63(1), pp. 37-50.
Kim, H.Y. and McGill, A.L., 2018. Minions for the rich? Financial status changes how
consumers see products with anthropomorphic features. Journal of Consumer Research,
45(2), pp. 429-450.
Kim, S. and Baek, T.H., 2018. Examining the antecedents and consequences of mobile app
Engagement. Telematics and Informatics, 35, pp. 148-158.
Lee, J.D. and See, K.A., 2004. Trust in automation: designing for appropriate reliance.
Human Factors, 46(1), pp. 50-80.
Lu, L., Cai, R. and Gursoy, D., 2019. Developing and validating a service robot integration
willingness scale. International Journal of Hospitality Management, 80, pp. 36-51.
MacVaugh, J. and Schiavone, F., 2010. Limits to the diffusion of innovation: a literature
review and integrative model. European Journal of Innovation Management, 13(2),
pp. 197-221.
Marinova, D., de Ruyter, K., Huang, M.H., Meuter, M.L. and Challagalla, G., 2017. Getting
smart: Learning from technology-empowered frontline interactions. Journal of Service
Research, 20(1), pp. 29-42.
McKnight, D.H., Choudhury, V. and Kacmar, C., 2002. The impact of initial consumer trust
on intentions to transact with a web site: a trust building model. Journal of Strategic
Information Systems, 11(3), pp. 297-323.
Pelau, C. and Ene, I., 2018. Consumers’ perception on human-like artificial intelligence
devices. In: R. Pamfilie, V. Dinu, L. Tăchiciu, D. Pleșea and C. Vasiliu, BASIQ
International Conference: New Trends in Sustainable Business and Consumption - 2018.
Heidelberg, Germany, 11-13 June 2018. Bucharest: ASE Publishing House.
Pelau, C. and Ene, I., 2020. Interaction between consumers and emerging forms of artificial
intelligence: a discriminant analysis. Studia Universitatis Vasile Goldis, 30(2), pp. 1-12.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 45
Pfadenhauer, M., 2015. The contemporary appeal of artificial companions: Social robots as
vehicles to cultural worlds of experience. The Information Society, 31(3), pp. 284-293.
Rinesi, M., 2015. The price of the Internet of Things will be a vague dread of a malicious
world. [online] Available at: <https://ieet.org/index.php/IEET2/more/rinesi20150925>
[Accessed 24 February 2018].
Rosenthal-von der Pütten, A.M. and Krämer, N.C., 2014. How design characteristics of
robots determine evaluation and uncanny valley related responses. Computers in Human
Behavior, 36, pp. 422-439.
Samsung, 2020. Realizați mai multe cu Bixby. [online] Available at:
<https://www.samsung.com/ro/apps/bixby/> [Accessed 3 December 2020].
Venkatesh, V., Thong, J. and Xu, X., 2012. Consumer acceptance and user of information
technology: Extending the unified theory of acceptance and use of technology. MIS
Quarterly, 36, pp. 157-178.
West, A., Clifford, J. and Atkinson, D., 2018. “Alexa, build me a brand”: An investigation
into the impact of artificial intelligence on branding. The Business & Management
Review, 9(3), pp. 321-330.
Wirtz, J., Patterson, P. G., Kunz, W. H., Gruber, T., Nhat Lu, V., Paluch, S. and Martins, A.,
2018. Brave new world: Service robots in the frontline. Journal of Service Management,
29(5), pp. 907-931.
AE Artificial Intelligence in Retail: Benefits and Risks Associated With Mobile Shopping Applications
46 Amfiteatru Economic
ARTIFICIAL INTELLIGENCE IN RETAIL: BENEFITS AND RISKS
ASSOCIATED WITH MOBILE SHOPPING APPLICATIONS
Victoria Stanciu1 and Sînziana-Maria Rîndașu2 1)2) Bucharest University of Economic Studies, Romania
Please cite this article as:
Stanciu, V. and Rîndașu, S.M., 2021. Artificial
Intelligence in Retail: Benefits and Risks Associated
With Mobile Shopping Applications. Amfiteatru
Economic, 23(56), pp. 46-64.
DOI: 10.24818/EA/2021/56/46
Article History
Received: 29 September 2020
Revised: 30 October 2020
Accepted: 3 December 2020
Abstract
The objective of the study is to examine the practical implications of using artificial
intelligence (AI) based solutions in the case of retail mobile applications, to enhance the
online shopping experience and improve the engagement by also having in mind the
privacy of the users. We examined 117 shopping applications available in the Google Play
market and investigated the permissions required for each application and the categories of
personal data collected from the users. Based on the information gathered, we provided
practical methods to integrate artificial intelligence-based solutions to offer a new set of
services, partially unavailable in physical stores. Some of the permissions identified, if
exploited by malicious users, can affect individuals’ privacy. The fact that artificial
intelligence is a fast-developing technology constitutes the main challenge in the effort of
creating proper regulations. This research provides practical directions regarding the
benefits of integrating artificial intelligence solutions in retail mobile applications in an
ethical manner, protecting the users’ privacy.
Keywords: artificial intelligence, machine learning algorithms, retail, ethics, privacy,
mobile shopping applications
JEL Classification: L81, K40, O33
Corresponding author, Sînziana-Maria Rîndasu – e-mail: [email protected]
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 47
Introduction
The pandemic changed the consumers’ habits, especially in the case of non-essential goods,
where the sales volume decreased significantly in the second and third quarter of 2020.
Since social distancing continued to represent the leading solution to reduce the exposure of
the individuals to the virus, many consumers started using online shopping to the detriment
of physical stores (Barnes, 2020; Nguyen et al., 2020). In this context of significant quick
changes, retailers’ attention is moving to e-commerce in an attempt to respond to the new
needs of the individuals. The online shopping has received a significant level of attention
during the last years and researchers highlighted that the key aspects important to the
consumers are represented by the ease of use, the design of the website or application, and
the trust regarding the usage of the personal data collected (Ha and Stoel, 2009; Lian and
Yen, 2014; Al-Debei, Akroush and Ashouri, 2015; Natarajan, Balasubramanian and
Kasilingam, 2017).
The adoption of artificial intelligence (AI) based solutions in retail and wholesale is rapidly
transforming the industry by enhancing the entire process since this domain generates a
vast amount of data that can be leveraged using this data-driven technology. Over the years,
different studies have investigated the practical ways in which AI-based solutions are being
used in retail and wholesale. However, less attention has been paid on leveraging mobile
applications to provide a better experience than the one in-store, using AI, especially in this
current context where customers tend to avoid unnecessary social interactions. Mobile
shopping applications, compared with online shopping by using different browsers, provide
more leverage to the companies to increase the engagement (Ho and Chung, 2020; Li, Zhao
and Pu, 2020) and facilitate the use of various AI features that can be embedded in the
applications.
Theoretically, there is no limit for AI to expand and facilitate enhancements in any
industry, so the boundaries should be set by ethical guidelines and principles (Kreutzer and
Sirrenberg, 2020) to create a technology that works in favour of the individual and the
collective well-being.
The purpose of the current study is to investigate practical implications regarding the use of
mobile shopping applications along with AI-based solutions to improve engagement,
enhance the online shopping experience, and encourage impulse buying, also focusing on
privacy, legal, and ethical implications. A series of mobile shopping applications available
in the Google Play market has been examined to achieve the research’s objective. This
approach is used to determine the main permissions required, which of those permissions
can be leverage by using AI, the security implications, and how the personal data collected
is used to train machine learning algorithms to better respond to the needs of the users. The
current research aims to provide support for retail companies to adopt more AI-based
solutions that can be integrated into mobile shopping applications, to increase the business
objective ethically.
This paper is structured in four parts. The first part presents the relevant literature review
for this research, focusing on current AI solutions used in retail, legal aspects, and ethical
challenges. The second part details the methodology used to conduct the research,
mentioning the purpose and objectives. The third part reflects the study’s results, which
present the main permissions of mobile applications that can be exploited by AI solutions,
the types of data collected by companies that offer the mobile applications, along with the
AE Artificial Intelligence in Retail: Benefits and Risks Associated With Mobile Shopping Applications
48 Amfiteatru Economic
main challenges regarding privacy. The last part presents the conclusions, the limits of the
present study, and the future research directions.
1. Literature review
Researchers tend to have different approaches when presenting the distinct fields of AI
(Buhalis, et al., 2019; Dignum, 2019; Chowdhary, 2020; Girasa, 2020; Kreutzer and
Sirrenberg, 2020). While in some studies the types of AI are categorised based on the IT-
related functionalities, others are using a taxonomy based on the evolution in time, the
principal sub-fields of applications being: natural language processing, natural image
processing, expert systems, and robotics. In terms of the capabilities and evolution,
researchers classify AI as Weak Artificial Intelligence (or Artificial Narrow Intelligence),
Artificial General Intelligence, and Strong Artificial Intelligence. Another classification of
AI was given by Hintze (2016), which considers that there are four main categories of AI:
reactive, limited memory, theory of the mind, and self-aware.
Machine learning (ML), one of the most known and used AI techniques in the retail field, is
part of the robotic sub-field and the narrow AI, having different types of implementation:
supervised learning (predicts), unsupervised learning (discovers patterns), reinforcement
learning (decision-making), and deep learning (multiple neural networks) (LeCun, Bengio
and Hinton, 2015; Goodfellow, Bengio and Courville, 2016; Dignum, 2019; Luce, 2019;
Kreutzer and Sirrenberg, 2020), all based on algorithms that are processing data, learning,
and developing new models to answer the needs of the companies.
1.1. AI applications in retail
Retailers all around the globe started to adopt AI solutions to support the development in
this sector. Amazon was one of the most active pioneers in automating this domain with the
use of AI in different areas of retail, from warehouse robots that can move items and
prepare orders to be shipped (Bogue, 2016), drones that will bring the goods to the
customers (Singireddy and Daim, 2018), anticipatory shipping (patented in 2011), that
should address a significant part of the logistical issues (Spiegel, et al., 2011), to checkout-
free shops (Polacco and Backes, 2018). The next step might be shipping the products to the
client, without an initial order, based on the algorithms’ predictions and the users’
behaviour and, if the product is not desired by the customer, it can be returned (Shankar,
2018).
Chatbots, also called conversational agents, are becoming extremely attractive to
companies due to the applicability as this technology promises to improve engagement,
facilitate the interaction, and provide a personalised shopping experience. They can use
natural language processing, natural image processing, and ML algorithms. A recent study
(Rese, Ganster and Baier, 2020), investigating a series of market research surveys, showed
that the acceptance rate of chatbots varies between 10% and 50%.
IoT devices, like the Samsung fridge (Gritti, Önen and Molva, 2019), allow people to avoid
a series of activities; the fridge is making a list of goods and sends the orders automatically
to the grocery shop for the products needed. However, a recent study focused on
autonomous shopping (De Bellis and Johar, 2020) identifies four significant psychological
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 49
barriers regarding the adoption of these systems: reduced control and autonomy, reduced
meaningful experience, reduced individuality and identity, and reduced social
connectedness.
The increasing adoption of augmented reality (AR) applications brings significant benefits
to the retail area by allowing customers to visualise products in a completely new and
innovative manner by using magic mirrors, virtual fitting rooms, and different techniques
that help clients make the best purchase decision and increase their level of satisfaction
(Poushneh and Vasquez-Parraga, 2017).
A major drawback for retailers to adopt AI-based solutions was the initial adoption costs.
Now there are several companies (Slyce – image search for recommendations, Oracle –
chatbots) that started to provide AIaaS (Artificial Intelligence as a Software), aiming to
facilitate the integration of these technologies for the retailers. Therefore, the expectations
are that during the next years almost all online stores will use AI to enhance the shopping
experience.
1.2. Regulating Artificial Intelligence
At the European Union (EU) level, several instruments can be used to regulate the legal
impact of the AI applications on individuals and society, both in terms of hard law and soft
law. However, the main challenges are derived from the constant expansion and evolution
of AI techniques. In this context of fast-developing technologies, the hard law (binding
legal instruments and regulations) if applied, might not be sufficient for a medium and long
timeframe, while the soft law (agreements, principles, and declarations that are not legally
binding) and ethical frameworks, lack an oversight mechanism and create difficulties for
programmers to develop AI-based applications.
Since the expansion of AI is touching more and more individuals’ personal lives and due to
the need for a legal framework, the Council of Europe’s Committee of Ministers set up in
September 2019 the Ad Hoc Committee on Artificial Intelligence – CAHAI. The objective
of CAHAI, established for two years, until 2021, is to examine, through dialogue with
stakeholders, “the feasibility and potential elements of a legal framework for the
development, design and application of artificial intelligence”, so AI applications will
respect the compliance with the rule of law. Furthermore, CAHAI should assess whether
the current legal instruments are suitable for addressing the present and future challenges
brought by the development of AI.
The need for regulating AI arises from the necessity of having personal data processed in a
fair, transparent, and accountable manner by avoiding discrimination and creating trust for
stakeholders and shareholders. The regulation of AI poses challenges derived from the
continuous development of the algorithms. In this respect, the lawmakers and standard
setters should be aware that the risk assessment should be carried on a systematic base,
using preferably a principle-based approach instead of a rule-based regulation.
The regulation of AI applications impacts the commerce industry that had already
embedded in its processes AI solutions. A common issue lately identified in the retail area
is price discrimination (PD) that refers to the use of ML algorithms aiming to sell a good or
service at the maximum price the consumer is willing to pay. Based on a series of labels
that the algorithm assigns to the individuals, based on the past activity and personal data,
AE Artificial Intelligence in Retail: Benefits and Risks Associated With Mobile Shopping Applications
50 Amfiteatru Economic
researchers showed that some categories of consumers are offered a particular good or
service at a higher price (Hannak, et al., 2014; Larson, Mattu and Angwin, 2015; Calvano,
et al., 2019). As per this, because of the retail companies’ clustering algorithms, some
individuals are offered a higher price, even though the delivery costs are the same for every
potential client. While PD mechanisms might not use complex AI techniques to target
specific categories of individuals, the algorithms can instead use a limited series of
attributes, like the location; in theory, AI can use unsupervised learning to obtain a finer-
grained PD (Gautier, Ittoo and Van Cleynenbreugel, 2020). The current anti-discrimination
laws might not be sufficient for this scenario of indirect discrimination and regulators have
to address this issue by highlighting the need for developers to be more accountable for the
algorithms that they design, preventing, in this way, PD.
Retailing represents the perfect environment for the use and growth of AI since it collects a
significant amount of information regarding consumers and their behaviour. Because of
this, the way in which the algorithms are processing the data has received a lot of research
attention by studying how the privacy of individuals is affected and possible technical
solutions to mitigate the risks (Els, 2017; Alguliyev, Aliguliyev and Abdullayeva, 2019;
Hao, et al., 2019; Mazurek and Małagocka, 2019; Kreutzer and Sirrenberg, 2020; Singh,
Rathore and Park, 2020; Thinyane and Sassetti, 2020).
1.3. Ethical challenges
AI solutions started to be researched and used since 1950, but the progress became
significant in the last two decades, impacting individuals’ lives systematically. This
incredible progress, however, raised a question: to what point AI can influence humanity?
Kreutzer and Sirrenberg (2020) consider that “There are – technically – (almost) no limits
to the possible fields of application of Artificial Intelligence. The limits should therefore be
set by ethical standards”. In the 2019 Artificial Intelligence Index Report (Perrault, et al.,
2019) are presented 12 ethical challenges highlighted by researchers that should be
addressed in the development of AI-based solutions. The primary concerns are focused on
fairness – using a dataset that is not discriminatory and does not contain elements that
might lead to algorithmic bias, interpretability – the degree to which the user can
understand the cause of the decision and can predict future results, explicability – the active
characteristic of the algorithm that brings clarifications regarding the undergoing processes
in providing an output, transparency – the level of information provided by an AI-based
system, during the decision-making process, accountability – the accountability of the
stakeholders involved in developing the algorithms regarding the implications generated by
using it, and data privacy – users knowing how their personal data is processed in the
developing and usage of the AI-based systems.
One primary technical concern is algorithmic bias, which is represented by a series of
systematic and repeated errors, that can lead to discrimination due to using inadequate data
(incomplete, unrepresentative, or already biased) to develop the algorithms.
In the relevant literature, several examples of algorithmic biases can be found (Bozdag,
2013; Lambrecht and Tucker, 2019; Taati, et al., 2019). However, in some cases, it is
uncertain if the bias was created by the inadequate dataset used to train the algorithms or if
the algorithms have learned to maximise the chances of achieving the objective. Although
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 51
not all forms of algorithmic biases are discriminatory, there is an ethical need to address
this issue.
Shopping represents a therapeutic activity for individuals (Babin, Darden and Griffin,
1994) and retailers are nowadays focusing on enhancing the customers’ experience.
However, there is a fine line between shopping therapy and compulsive purchases
(Hirschman, 1992), so companies should focus their efforts on supporting a correct
shopping behaviour, especially in the context of introducing AI-based solutions that are
capturing easier the attention of individuals.
Another critical aspect that is supported by ML algorithms in the retail area is to predict the
users’ needs and provide anticipatory delivery of the products, not in the warehouse, but
directly to the customers, that can return the goods if they are not satisfied with them,;
however, this seems to impact the control of the individual over the decisions made to
acquire a good. While this is not a trespassing of privacy, it raises concerns regarding the
impact that this might have on the individuals’ personal lives.
2. Research methodology
Taking into consideration that smartphones gained a tremendous amount of popularity,
have an extremely favourable acceptance rate (Deloitte, 2019), and provide significant
support to the retailers, we considered appropriate to explore the relationship between AI
solutions and mobile applications, especially now when the customers’ behaviour changed
as a result of the pandemic, the number of visits to physical stores decreased worldwide,
and the online shopping started to be preferred, the smartphones being a significant vector
in this switch.
This research aims to identify practical solutions for the use of mobile shopping
applications for both essential and non-essential goods, together with solutions based on AI
systems, to increase customer engagement, encourage impulse buying, provide a
significantly better shopping experience, and analyse potential risks that may affect the
confidentiality of data collected from the users. Thus, the general and specific research’s
objectives we set out were:
O1. Analysing how privacy can be affected by using mobile applications. To achieve
this goal, we have refined the research direction through the following specific objectives:
O1.1. Identifying the mobile applications’ permissions that may affect the
confidentiality of data provided by users.
O1.2. Examining the privacy policies of the companies that sell products through
mobile applications, to determine the main types of information collected, the risks and
how the data is protected.
O2. Identifying how the AI used in mobile shopping applications supports the
development of competitive advantage. To achieve this goal, we have refined the research
direction through the following specific objectives:
O2.1. Analysing the permissions of mobile applications to determine which of them
can be used in conjunction with AI solutions to increase customer engagement.
AE Artificial Intelligence in Retail: Benefits and Risks Associated With Mobile Shopping Applications
52 Amfiteatru Economic
O2.2. Identifying permissions of mobile applications that can be exploited through
the use of AI to encourage impulse buying and provide consumers with a better experience
compared to physical store purchases.
The general approach of the research consists of conducting a cross-sectional study; the
objective is to capture several perspectives on the data collected from mobile shopping
applications, in order to provide practical recommendations to achieve the study’s aim.
Currently, in the Google Play store, there are over three million applications deployed, from
which over 110.000 are included in the shopping category and, by looking at the popularity
in terms of numbers of downloads, around 200.000 of them (6.67%) have been downloaded
more than 100.000 times.
In this exploratory research, we analysed 117 shopping applications available in Google
Play, randomly selected exclusively based on the popularity. The criteria taken into
consideration when selecting the application was to have at least 100.000 downloads.
Within the selected applications, products from different categories are sold (table no. 1).
The data has been collected during July-August 2020. Some of the applications analysed
are: Zalando - Shopping & Fashion, iHerb, The Home Depot, Best Buy, The Kroger,
Decathlon International, and PetSmart. Since the AI solutions are diverse and have a high
degree of complexity, for the selection of the applications included in this study the
existence or lack of usage of AI was not taken into account. Although in some cases users
can clearly identify the existence of an AI-based program, such as searching for a product
by uploading an image, conversational robots, or personalised recommendations, AI is
much more complex and even if the user is not informed about the programs used by the
company to determine a specific response, this does not indicate that these applications do
not use AI. For these reasons, we have not exclusively selected applications whose
description and functionality suggested the use of ML algorithms or any other program in
the field of AI.
Table no. 1. Distribution of the applications analysed by category
and number of downloads Category of
applications/ Number
of downloads
100k+ 500k+ 1mil+ 5mil+ 10mil+ 50mil+ 100mil+ Total
Clothes, shoes, and
accessories 4 6 19 6 10 - 1 46
Electronics - - 2 - 1 - - 3
Groceries - 3 16 - 2 - - 21
Home improvement - 2 4 - 3 - - 9
Multiple categories - - 7 4 3 1 - 15
Pet shops - - 4 - - - - 4
Skincare and
perfumes 5 2 6 1 2 - - 16
Sportswear and
equipment - - 2 - - - - 2
Vehicle and
accessories - - 1 - - - - 1
Total 9 13 61 11 21 1 1 117
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 53
From these 117 applications analysed, 15 of them refer to multiple categories of products,
including clothes, cosmetics, sports, home improvement, electronics, and many others.
Since it was difficult to allocate these applications to a particular category based on the
market share, we considered appropriate to assign them into a distinct category.
The companies that own these applications are from 30 different countries: Australia,
Brazil, Canada, Chile, Hong Kong, India, Israel, Japan, Nigeria, Russia, South Africa,
South Korea, Switzerland, Turkey, United Arab Emirates, United Kingdom, United States
of America, and 13 countries from the EU (Austria, Czech Republic, Finland, France,
Germany, Italy, Latvia, Netherlands, Poland, Portugal, Romania, Spain, and Sweden). To
have a complete understanding of what kind of data and permissions are requested by the
retailing companies, we considered necessary to have a view from different countries that
have a separate set of regulations regarding personal data.
The data collected from the applications and privacy policies have been analysed from
different perspectives based on the research’s purpose, for all of the 117 applications
included in this present study. By examining the permissions and privacy policies, we used
a content analysis to capture all the information put at the disposal of the smartphones’
users, as this research method has an adequate level of applicability in the retail area,
because it supports the avoidance of biased results and “provides an empirical starting point
for generating new research evidence” (Kolbe and Burnett, 1991). We have also used
statistical methods for the data collected to test if there is a significant difference between
applications, depending on the companies’ location.
3. Results and discussion
By analysing the permissions requested by the applications, we identified 63 different kinds
of permissions, from 13 different categories: photos/media/file, storage, Wi-Fi connection
information, camera, location, microphone, phone, device ID & call information, identity,
device ID & app history, contacts, calendar, and others. In the Google Play market, there is
a particular category, called others, where are presented the permissions that cannot be
included in one of the 12 categories listed above. After collecting the data from all the
applications, 1.854 permissions have been identified, an application having, on average, 16
different permissions (table no. 2).
Table no. 2. Distribution of the number of permissions of the applications analysed
by category and number of downloads Category of
applications/ Number
of downloads
100k+ 500k+ 1mil+ 5mil+ 10mil+ 50mil+ 100mil+ Average
Clothes, shoes, and
accessories 49 79 300 94 177 - 31 16
Electronics - - 46 - 22 - - 23
Groceries - 35 282 - 41 - - 17
Home improvement - 26 51 - 52 - - 14
Multiple categories - - 87 87 43 16 - 16
Skincare and
perfumes 60 24 92 16 35 - - 14
AE Artificial Intelligence in Retail: Benefits and Risks Associated With Mobile Shopping Applications
54 Amfiteatru Economic
Category of
applications/ Number
of downloads
100k+ 500k+ 1mil+ 5mil+ 10mil+ 50mil+ 100mil+ Average
Pet shops - - 65 - - - - 16
Sportswear and
equipment - - 26 - - - - 13
Vehicle and
accessories - - 18 - - - - 18
Average number of
permissions per
number of downloads
12 13 16 18 18 16 31 -
To achieve the specific objective O1.1., we examined the 63 different types of permissions
requested by the applications following their specifics. We then computed the frequency,
and classified them into three categories (low, medium, and high), based on the risk of
personal or sensitive data collection and exposure, correlated with the technical
functionality. The majority of the permissions have a medium and high level of exposure
(53.96%), while 29 permissions do not pose any significant threat to personal data.
To examine whether there is a difference between EU and non-EU countries, in terms of
the number of total permissions and the number of medium and high-risk permissions per
application, we conducted a one-way ANOVA test (single-factor analysis of variance)
(table no. 3), using the Statistical Package for Social Sciences (SPSS). We have also
included UK companies in EU countries, due to the fact that they comply with the same
personal data protection regulation.
Table no. 3. ANOVA Test
Descriptive
N Mean Std.
Deviation
Std.
Error
95% Confidence
Interval for Mean Min. Max.
Lower
Bound
Upper
Bound
Total_
number_of_
permissions
Non-EU 52 17.15 5.308 .736 15.68 18.63 8 31
EU 65 14.80 4.342 .539 13.72 15.88 6 25
Total 117 15.85 4.916 .454 14.95 16.75 6 31
Number_of
_medium_
and_high-
risk_
permission
Non-EU 52 9.88 3.507 .486 8.91 10.86 3 21
EU 65 8.20 2.694 .334 7.53 8.87 2 13
Total 117 8.95 3.181 .294 8.37 9.53 2 21
Test of Homogeneity of Variances
Levene Statistic df1 df2 Sig.
Total_number_of_
permissions 1.760 1 115 .187
Number_of_medium_
and_high-risk_permission 2.926 1 115 .090
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 55
ANOVA
Sum of Squares Df Mean
Square F Sig.
Total_number_of_
permissions
Between Groups 160.062 1 160.062 6.964 .009
Within Groups 2643.169 115 22.984
Total 2803.231 116
Number_of_medium
_and_high-risk_
permission
Between Groups 81.985 1 81.985 8.636 .004
Within Groups 1091.708 115 9.493
The result of the ANOVA analysis is showing that usually, the applications analysed where
the company is in the EU have generally fewer permissions than the applications deployed
by companies from countries outside the EU. This result might be an effect of the General
Data Protection Regulation (GDPR) adopted in all EU states in 2018, that reinforced
personal data processing rules. Although states from all over the world have privacy laws,
the regulations’ effect is more visible in Europe.
The application are making constant efforts to detect any malicious application. However,
there has been evidence that the initial screening performed by the specialists was not
always successful and led to the deployment of malware applications that had the potential
of affecting the users’ privacy (Rahman, et al., 2017; Bhat and Dutta, 2019), as the ability
of such programs is evolving and manages to bypass the security measures (Wadkar, Di
Troia and Stamp, 2020).
The awareness level of the users regarding the risks generated by the usage of mobile
applications have been assessed during the last years, and the studies have highlighted that
the majority is not concerned with the permissions of the applications that they are
installing, or they fail to understand them (Felt, et al., 2012; Ramachandran, et al., 2017;
Ngobeni and Mhlongo, 2019).
In terms of the functionality of the applications examined in this paper, we identified four
substantial risks that can significantly affect the users’ privacy:
Eavesdropping – 21% of the sample analysed include this feature (record audio) to
facilitate the vocal search, but malicious applications can use this permission to eavesdrop
on users. Even though this phenomenon has not been identified at a large scale, it is,
however, possible for an application that received this permission to eavesdrop on users
(Kröger and Raschke, 2019). Analysing the companies’ privacy policies that deployed
these applications, we have not identified information regarding the storing methods of
audio records or their future use.
Secretly transferring data – 12% of the applications have the permission to connect
and disconnect from Wi-Fi and 5% are allowed to change the network connectivity and
both of these, in association with another permission, “Full network access” (that is present
in all applications requiring an internet connection) can act as a malware, secretly collect,
then transmit the user’s data. In relation to mobile shopping, these two permissions are used
to increase the customers’ exposure to the offers and shopping applications.
AE Artificial Intelligence in Retail: Benefits and Risks Associated With Mobile Shopping Applications
56 Amfiteatru Economic
Install malware without the users’ knowledge – in 6% of the applications analysed
we identified the permission called “Download files without notification” that allows the
applications to download and install programs without the knowledge of the user, even
malware programs that can then be used to spy on the activity and then leak private data.
Online shopping applications will use this permission to download mandatory updates
without first notifying the users, avoiding, in this way, scenarios in which they might access
a non-functional application.
Gather unprovided data without the users’ knowledge – a permission called “Read
sensitive log data” has been identified in five cases. This allows the application to read the
data provided and stored by the other applications installed and might access personal data
of the user. This feature can be used to collect more information regarding the consumers
and give a better personalised shopping experience, but if the data has not been provided
with the users’ knowledge, this practice raises some ethical questions.
Through this analysis of medium and high-risk permissions in terms of the ability to affect
privacy, we achieved the specific objective O1.1. of the research.
The specific objective O1.2. imposed the retailers’ privacy policies that deployed these
applications. For this purpose, we identified the main types of personal data collected: full
name, profile picture, delivery address, phone number, email address, credit/debit card
information, passport ID, and drivers’ license number. Besides these categories of data,
other types of information were collected, such as location, postal code, gender, age, date of
birth, hobbies, buying preferences, estimated income, shopping patterns and behaviours, the
reason for purchasing a particular product, and in one case we identified that the company
was storing information about the customers’ education level. All these details are needed
by the retailers to address the users’ needs better, but they should protect personal privacy
by using the best current available solutions to avoid data leakages. The majority of the
companies specified in the privacy policy that they are using encryption methods to protect
the data of the users transmitted via the internet, usually Secure Sockets Layer (SSL), an
encryption-based Internet security protocol, but researchers have highlighted that this
technology has a series of vulnerabilities, that might affect the accuracy of the data
transmitted (Ramírez-López, et al., 2019; Tang, et al., 2019). Only in a limited number of
cases details about how the data stored is protected were provided.
A recent experiment carried out by a well-known cybersecurity expert, Bob Diachenko,
showed that unprotected databases are usually attacked 18 times per day and during the first
eight hours after deploying the database, 175 attacks have been performed by hackers
(Bischoff, 2020).
There are several techniques that companies can use to protect the privacy of the users,
such as leveraging AI, to identify earlier data breaches and sending red flags to respond
faster to a possible attack. Also, developers can train ML algorithms to categorise data
based on the level of confidentiality to use the proper encryption techniques, testing and
identifying potential risk areas and automatically apply patches. Since the majority of
mobile applications analysed collect information about the credit/debit cards used by the
customers to make payments, AI-based solutions can be used to detect any fraudulent
activity based on the previous behaviour and preferences of the users.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 57
In two cases, we identified that the retailers were collecting data about the users’ income,
and in one situation, information about the education level was stored. Although it might
seem harmless, this kind of data can encourage algorithms to provide a biased outcome that
might result in PD. Moreover, in some of the privacy policies was mentioned that the
company would store any data that the user provides, an aspect that raises concerns
regarding the way the information is being processed and the output provided. Wiener
highlighted in 1960 that “we had better be quite sure that the purpose put into the machine
is the purpose which we really desire”. Although this quote is more than 60 years old, it has
a pearl of significant wisdom for the current status of automation, where narrow
intelligence algorithms are using all the data provided and generate an outcome based on
that, so developers should act with diligence and avoid any inappropriate data processing.
The achievement of the specific objectives O1.1. and O1.2. led to the accomplishment of
the first general research objective formulated; thus, we were able to perform an analysis of
how privacy can be affected by the usage of mobile shopping applications.
Aiming to achieve the specific objective O2.1. we conducted a brief analysis of the main
elements that determine consumers to make impulse purchases, then we examined what
permissions within the analysed mobile applications can be exploited by using AI
techniques to encourage the impulse buying. We also identified permissions that can be
used to improve the online shopping experience.
Shoppers tend to make more impulsive purchases in stores than online (First Insight, 2019), but
the reasons behind this behaviour are not fully understood. Perceived experience and a series of
organic variables are making consumers keener to engage in impulse purchasing (Liu, Li and
Hu, 2013), and the current shift generated by the pandemic influenced the impulsive buying
habits as well (Roggeveen and Sethuraman, 2020), so retailers should find appropriate ways to
create favourable environments for the clients to embrace more online shopping. While
browsing on websites using smartphones can be difficult for consumer, mobile applications are
more user-friendly and provide a better online shopping experience.
The main attributes that facilitate the engagement in online impulse buying are: visual
appeal, ease of use, product availability, facilitated payment process, promoting intensity,
and a series of organic variables such as normative evaluation and instant gratification (Liu,
Li and Hu, 2013; Leong, Jaafar and Ainin, 2018). Recommendation algorithms can
improve impulse buying (Hostler, et al., 2011) by gathering the users’ preferences and
making predictions based on previous purchases. However, developers and retailers should
keep in mind that impulse buying is unplanned and spontaneous, so they should avoid
obnoxious advertising and send pop-ups or notifications only at a suitable time.
Two of the applications analysed have a permission called “Read battery statistics”, which
allows battery data collection and, in this way, developers can predict, with the use of ML
algorithms, when it is a suitable time for sending notifications so they can receive more
attention from the user. This is a low-level risk permission and companies should use it
more to increase the exposure time. A permission called “Reorder running apps”, identified
in only five applications, can be exploited to gain the clients’ attention and encourage
impulsive buying behaviours by exposing more the user to that application. Another
permission, “Control vibrations”, is present in 81% of the cases analysed and the main
purpose is to draw the users’ attention to the notifications sent by the application so the
developers can leverage this; however, this might be considered slightly intrusive. “Run at
AE Artificial Intelligence in Retail: Benefits and Risks Associated With Mobile Shopping Applications
58 Amfiteatru Economic
startup”, identified in 56% of the cases, allows the application to be more visible to the
consumer and increases the chances of impulsive shopping, being an efficient feature to
improve the chances of accessing the online store application. Nevertheless, this advertising
should be carried out ethically, without severely influencing the control of the users.
One of the most accepted applications that AI provides is augmented reality (AR) to
support the decisions of the users (Alves and Reis, 2020) like the tool created by IKEA, that
customers can use to decide whether a particular piece of furniture or decoration will be
suitable for their house. Analysing the data collected, we identified that 79% of the
applications require the permission to take pictures and videos, most of them using this
feature to allow the users to upload photos and videos with their reviews, but the companies
can leverage this permission by using AR techniques to allow the customers to virtually try
an outfit and reduce the return ratio for clothes, shoes, and accessories. Also, for skincare
and cosmetics, pictures can be uploaded for the user to decide what makeup shade to select.
Another practical application to improve the shopping experience is to put at the disposal of
the client AI algorithms (natural image processing and ML) that will assist them in making
the right decision when looking for a particular product, based on the client’s preferences.
For example, the user can upload a picture with a specific outfit (85% of the applications
have the permission to read the storage space) and the algorithm can make suggestions by
trying to find the best match available at that particular moment.
To increase the ease of use and improve the experience, the developers could leverage
another permission identified in 11% of the cases, “Create accounts and set passwords”,
which allows the users to connect to a specific store, without manually having to create an
account, by linking the new account to the personal data stored on the phone. Although the
risk of this permission is medium, using it in a proper ethical manner, the users’ personal
data should be collected and stored transparently.
The achievement of the specific objective O2.2. imposed the analysis of permissions, which,
along with AI, can support increasing the level of customer engagement. We started from the
idea that improving customer engagement is a critical objective in the retail industry, and
companies are now interested in introducing chatbots to facilitate the interactions and provide
real-time availability to the users to handle their queries. In terms of functionality, these
programs can use a text-based or voice-based search (speech to text transcription). As
presented above, 21% of the applications analysed asked for permission to record audio – this
feature can be used by the client to discuss with the chatbots or even to do vocal searches for
the products needed. Also, the users can have the possibility to review whether their demand
has been solved and if the output was in line with the expectations and, in this way, the ML
algorithms behind the chatbots can improve after each interaction with the client and be able to
generate personalised outputs. While chatbots’ acceptance rate is still relatively low, trust can
be increased by being transparent with the users.
Data analytics techniques available in the online shopping applications gather data
regarding the time spent by a user viewing a particular item and also how many times the
application has been accessed. As presented above, the majority of the applications studied
had permission to control vibration to advise the user about the notifications received.
Using ML algorithms, the companies can analyse if the advertising is intrusive or not and if
it can affect the clients’ behaviour – for example, it can be examined if the notifications are
being deleted without accessing the data. In this way, the application can be programmed to
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 59
send only several notifications per month instead of daily. By doing so, companies can
improve the rating from the client by avoiding obnoxious advertising.
Conclusions
After conducting the current research, we were able to draw several conclusions regarding the
practical implications and opportunities of using AI-based solutions for mobile shopping
applications. From a total of 63 distinct permissions examined after analysing the 117
applications included in the study, we identified several permissions that retailers can use to
enhance the online shopping experience, improve engagement, and encourage impulse
buying. However, the majority of these permissions that can be successfully leveraged using
AI-based solutions have been encountered only in a limited number of cases. Companies
should focus more on mobile applications to facilitate the shopping experience being now
able to provide a series of new services to their customers such as virtual fitting rooms,
chatbots that can provide instant product recommendation and ML algorithms that search for
a particular product after analysing an image provided by the user. One major drawback of the
integration of AI in e-shopping seems to be the cost of adoption, an issue that we consider
will start to fade once the number of AIaaS solutions increases. The majority of companies
are already using a series of data analytics services provided by search engines, such as
Google, to gather non-personal data about their clients, so the expectations are that most
retailers will start on the medium-term leveraging AI-based solutions.
Another finding was that some of the applications ask for permissions that can affect the
privacy of the users and if these permissions are exploited maliciously, they can even
eavesdrop on the user, secretly collect data, and install malware programs on the personal
smartphones or tablets. Therefore, it is essential for retailers when deploying AI solutions
to keep in mind the customers’ privacy, as data breaches pose both financial and
reputational risks. Also, the statistical analysis performed showed that in the case of EU
based companies, there is a lower number of total and possible harmful permissions, which
highlights that these retailers might be a little more cautious regarding the privacy of
individuals, even though similar regulations as in the EU started to be adopted by the
majority of the countries worldwide.
After examining the privacy policies of the companies focusing on the types of data they
can leverage to improve the ML algorithms for products recommendations and predict the
future needs of the users, we identified that some of the companies are collecting data that
can lead to biased or discriminatory algorithms, such as estimated income and level of
education. Retailers have at their disposal a wide range of information to successfully
develop algorithms, but they must collect only the data they will need for conducting
business in an ethical manner.
Grasping AI-based solutions by embedding them in mobile shopping applications represent
an excellent opportunity for retailers as the customers’ preferences change and they opt for
a personalised shopping experience more suitable to their needs. Since AI is a fast-
developing technology, ethical and legal challenges are the main concern for researchers
and developers. As the retail companies have a vast amount of structured and unstructured
data at their disposal, they can lead the way in the development of ethical AI solutions that
will be used for the individual and collective well-being.
In the relevant literature, the issue of privacy of the personal data provided by users
represented an intensely researched topic (Rahman, et al., 2017; Bhat and Dutta, 2019;
AE Artificial Intelligence in Retail: Benefits and Risks Associated With Mobile Shopping Applications
60 Amfiteatru Economic
Kröger and Raschke, 2019), as well as the analysis of the reasons that determine a
consumer to make a purchase or not (Liu, Li and Hu, 2013; Leong, Jaafar and Ainin, 2018;
Ho and Chung, 2020), including AI elements. However, we have not identified in the
literature any study that analyses the potential of mobile application permissions to
encourage the impulse buying, provide a better experience compared to visiting a physical
store, and to increase customer engagement in an ethical manner, focusing on the
personality and confidentiality of data provided by users to online retailers. Therefore, our
study brings an element of originality to AI research in the field of retail.
The main limitation of this study is caused by the partial lack of transparency in the privacy
policies, drafted by companies that use these applications for online shopping. Although
these policies are public, companies, for competitiveness reasons do not fully present the
processes in which they use anonymized user data; therefore, this aspect did not allow us to
determine the possibility of developing subjective or discriminatory algorithms.
The present research can be continued in two directions: (1) extending the study over a
more extended period to see to what extent the shopping experience has been improved as a
result of using mobile applications for online commerce; and (2) analysing medium and
high-risk permissions, through a comparative study in countries that have adopted GDPR
and those that have not.
References
Al-Debei, M.M., Akroush, M.N. and Ashouri, M.I., 2015. Consumer attitudes towards
online shopping: The effects of trust, perceived benefits, and perceived web. Internet
Research, [e-journal] 25(5), pp.707-733. https://doi.org/10.1108/IntR-05-2014-0146.
Alguliyev, R.M., Aliguliyev, R.M. and Abdullayeva, F.J., 2019. Privacy-preserving deep
learning algorithm for big personal data analysis. Journal of Industrial Information
Integration, [e-journal] 15, pp.1-14. https://doi.org/10.1016/j.jii.2019. 07.002.
Alves, C. and Reis, J.L., 2020. The Intention to Use E-Commerce Using Augmented
Reality-The Case of IKEA Place. In: Á. Rocha, C. Ferrás, C. Montenegro Marin and V.
Medina García, The 2020 International Conference on Information Technology &
Systems. Bogota, Columbia, 5-7 February 2020. Cham: Springer.
Babin, B.J., Darden, W.R. and Griffin, M., 1994. Work and/or fun: measuring hedonic and
utilitarian shopping value. Journal of consumer research, [e-journal] 20(4), pp.644-656.
https://doi.org/10.1086/209376.
Barnes, S.J., 2020. Information management research and practice in the post-COVID-19
world. International Journal of Information Management, [e-journal] 55, pp.1-4.
https://doi.org/10.1016/j.ijinfomgt.2020.102175
Bhat, P. and Dutta, K., 2019. A survey on various threats and current state of security in
android platform. ACM Computing Surveys, [e-journal] 52(1), pp.1-35. https://doi.org/
10.1145/3301285.
Bischoff, P., 2020. Unsecured databases attacked 18 times per day by hackers. [online]
Available at: <https://www.comparitech.com/blog/informati on-security/unsecured-
database-honeypot/> [Accessed 12 August 2020].
Bogue, R., 2016. Growth in e-commerce boosts innovation in the warehouse robot market.
Industrial Robot, [e-journal] 43(6), pp. 583-587. https://doi.org/10.1108/IR-07-2016-0194.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 61
Bozdag, E., 2013. Bias in algorithmic filtering and personalization. Ethics and information
technology, [e-journal] 15(3), pp.209-227. https://doi.org/10.1007/s10 676-013-9321-6.
Buhalis, D., Harwood, T., Bogicevic, V., Viglia, G., Beldona, S. and Hofacker, C., 2019.
Technological disruptions in services: lessons from tourism and hospitality. Journal of
Service Management, [e-journal] 30(4), pp.484-506. http://dx.doi.org/10.1108/JOSM-
12-2018-0398.
Calvano, E., Calzolari, G., Denicolò, V. and Pastorello, S., 2019. Algorithmic pricing what
implications for competition policy?. Review of industrial organization, [e-journal]
55(1), pp.155-171. https://doi.org/10.1007/s11151-019-09689-3.
Chowdhary, K.R., 2020. Fundamentals of Artificial Intelligence. New Delhi: Springer Nature.
Council of Europe, 2020. CAHAI Ad Hoc Committee on Artificial Intelligence. [pdf]
Council of Europe. Available at: <https://rm.coe.int/leaflet-cahai-en-june-2020/
16809ed7fd> [Accessed 28 July 2020].
De Bellis, E. and Johar, G.V., 2020. Autonomous Shopping Systems: Identifying and
Overcoming Barriers to Consumer Adoption. Journal of Retailing, [e-journal] 96(1),
pp.74-87. https://doi.org/10.1016/j.jretai.2019.12.004.
Deloitte, 2019. Technology, Media, and Telecommunications Predictions 2020 [pdf]
Deloitte Development LLC. Available at: <https://www2.deloitte.com/global/en/
insights/industry/technology/technology-media-and-telecom-predictions> [Accessed 12
August 2020].
Dignum, V., 2019. Responsible Artificial Intelligence: How to Develop and Use AI in a
Responsible Way. New Delhi: Springer Nature.
Els, A.S., 2017. Artificial Intelligence as a Digital Privacy Protector. Harvard Journal of
Law & Technology, 31(1), p.217.
Felt, A.P., Ha, E., Egelman, S., Haney, A., Chin, E. and Wagner, D., 2012. Android
permissions: User attention, comprehension, and behavior. In: USENIX Association,
Proceedings of the eighth symposium on usable privacy and security. Washington,
D.C., U.S.A., July 2012. New York: Association for Computing Machinery
First Insight, 2019. The State of Consumer Spending: In-Store Impulse Shopping Stands the
Test of Time. [online] Available at: <https://www. firstinsight.com/white-papers-
posts/the-state-of-consumer-spending-report> [Accessed 12 August 2020].
Gautier, A., Ittoo, A., and Van Cleynenbreugel, P., 2020. AI algorithms, price
discrimination and collusion: a technological, economic and legal perspective.
European Journal of Law and Economics, [e-journal] pp.1-31. https://doi.org/
10.1007/s10657-020-09662-6.
Girasa, R., 2020. AI as a Disruptive Technology. In: R. Girasa ed., 2020. Artificial
Intelligence as a Disruptive Technology. Cham: Palgrave Macmillan, pp.3-21.
Goodfellow, I., Bengio, Y. and Courville, A., 2016. Deep learning (Vol. 1). Cambridge:
MIT press.
Gritti, C., Önen, M. and Molva, R., 2019. Privacy-preserving delegable authentication in
the internet of things. In: ACM (Association for Computing Machinery), Proceedings of
the 34th ACM/SIGAPP Symposium on Applied Computing. Limassol, Cyprus, 8-12
April 2019. New York: Association for Computing Machinery.
AE Artificial Intelligence in Retail: Benefits and Risks Associated With Mobile Shopping Applications
62 Amfiteatru Economic
Ha, S. and Stoel, L., 2009. Consumer e-shopping acceptance: Antecedents in a technology
acceptance model. Journal of business research, [e-journal] 62(5), pp.565-571.
https://doi.org/10.1016/j.jbusres.2008.06.016.
Hannak, A., Soeller, G., Lazer, D., Mislove, A. and Wilson, C., 2014. Measuring price
discrimination and steering on e-commerce web sites. In: ACM (Association for
Computing Machinery), Proceedings of the 2014 conference on internet measurement
conference. Vancouver, BC, Canada, 5-7 November 2014. New York: Association for
Computing Machinery.
Hao, M., Li, H., Luo, X., Xu, G., Yang, H. and Liu, S., 2019. Efficient and privacy-
enhanced federated learning for industrial artificial intelligence. IEEE Transactions on
Industrial Informatics, [e-journal] 16(10), pp.6532-6542. https://doi.org/10.1109/TII.20
19.2945367.
Hintze, A., 2016. Understanding the four types of AI, from reactive robots to self-aware
beings. [online] Available at: <http://theconversation .com/understanding-the-four-types-
of-ai-fromreactive-robots-to-self-aware-beings-67616> [Accessed 14 August 2020].
Hirschman, E.C., 1992. The consciousness of addiction: Toward a general theory of
compulsive consumption. Journal of Consumer Research, [e-journal] 19(2),
pp.155-179. https://doi.org/10.1086/209294.
Ho, M.H.W. and Chung, H.F., 2020. Customer engagement, customer equity and
repurchase intention in mobile apps. Journal of Business Research, [e-journal] 121,
pp.13-21. https://doi.org/10.1016/j.jbusres.2020.07.046
Hostler, R.E., Yoon, V.Y., Guo, Z., Guimaraes, T. and Forgionne, G., 2011. Assessing the
impact of recommender agents on on-line consumer unplanned purchase behavior.
Information & Management, [e-journal] 48(8), pp.336-343. https://doi.org/10.1016/
j.im.2011.08.002.
Kolbe, R.H. and Burnett, M.S., 1991. Content-analysis research: An examination of
applications with directives for improving research reliability and objectivity. Journal
of consumer research, [e-journal] 18(2), pp.243-250. https://doi.org/10. 1086/209256.
Kreutzer, R.T. and Sirrenberg, M., 2020. Understanding Artificial Intelligence.
Switzerland: Springer Nature Switzerland AG
Kröger, J.L. and Raschke, P., 2019. Is my phone listening in? On the feasibility and
detectability of mobile eavesdropping. In: S. Foley, 33th IFIP Annual Conference on
Data and Applications Security and Privacy. Charleston, U.S.A., 15-17 July 2019.
USA: Springer.
Lambrecht, A. and Tucker, C., 2019. Algorithmic bias? An empirical study of apparent
gender-based discrimination in the display of STEM career ads. Management Science,
[e-journal] 65(7), pp.2966-2981. https://doi.org/10.1287/ mnsc.2018.3093.
Larson, J., Mattu, S. and Angwin, J., 2015. Unintended Consequences of Geographic
Targeting. Technology Science, [e-journal]. http://dx.doi.org/10.7910/DVN/ VEBPCZ.
LeCun, Y., Bengio, Y. and Hinton, G., 2015. Deep learning. Nature, [e-journal] 52,
pp.436-444. doi:10.1038/nature14539.
Leong, L.Y., Jaafar, N.I. and Ainin, S., 2018. The effects of Facebook browsing and usage
intensity on impulse purchase in f-commerce. Computers in Human Behavior,
[e-journal] 78(1), pp.160-173. https://doi.org/10.1016/j.chb.2017.09.033.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 63
Li, X., Zhao, X. and Pu, W., 2020. Measuring ease of use of mobile applications in e-
commerce retailing from the perspective of consumer online shopping behaviour
patterns. Journal of Retailing and Consumer Services, [e-journal] 55, pp. 1-12.
Lian, J.W. and Yen, D.C., 2014. Online shopping drivers and barriers for older adults: Age
and gender differences. Computers in Human Behavior, [e-journal] 37, pp.133-143.
https://doi.org/10.1016/j.chb.2014.04.028.
Liu, Y., Li, H. and Hu, F., 2013. Website attributes in urging online impulse purchase: An
empirical investigation on consumer perceptions. Decision Support Systems, [e-journal]
55(3), pp.829-837. https://doi.org/10.1016/j.dss.2013.04.001.
Luce, L., 2018. Artificial Intelligence for Fashion: How AI is Revolutionizing the Fashion
Industry. San Francisco: Apress.
Mazurek, G. and Małagocka, K., 2019. Perception of privacy and data protection in the
context of the development of artificial intelligence. Journal of Management Analytics,
[e-journal] 6(4), pp.344-364. https://doi.org/10.1080/23270012.2019.1671243.
Natarajan, T., Balasubramanian, S.A. and Kasilingam, D.L., 2017. Understanding the
intention to use mobile shopping applications and its influence on price sensitivity.
Journal of Retailing and Consumer Services, [e-journal] 37, pp.8-22.
https://doi.org/10.1016/j.jretconser.2017.02.010.
Ngobeni, A. and Mhlongo, S., 2019. Towards Enhancing Security in Android Operating
Systems–Android Permissions & User Unawareness. In: IEEE, 2019 2nd International
Conference on Computer Applications & Information Security. Riyadh, Saudi Arabia,
1-3 May 2019. Riyadh: IEEE
Nguyen, H.V., Tran, H.X., Van Huy, L., Nguyen, X.N., Do, M.T. and Nguyen, N., 2020.
Online Book Shopping in Vietnam: The Impact of the COVID-19 Pandemic Situation.
Publishing Research Quarterly, [e-journal] 36, pp.437-445. https://doi.org/10.1007/
s12109-020-09732-2
Perrault, R., Shoham, Y., Brynjolfsson, E., Clark, J., Etchemendy, J., Grosz, B., Lyons, T.,
Manyika, J., Mishra, S. and Niebles, J.C., 2019. Artificial Intelligence Index Report
2019. [pdf] Stanford: Stanford University. Available at: <https://hai.
stanford.edu/sites/default/files/ai_index_2019_report.pdf> [Accessed 28 July 2020].
Polacco, A. and Backes, K., 2018. The amazon go concept: Implications, applications, and
sustainability. Journal of Business and Management, [e-journal] 24(1), pp.79-92.
http://dx.doi.org/10.6347%2fJBM.201803_24(1).0004.
Poushneh, A. and Vasquez-Parraga, A.Z., 2017. Discernible impact of augmented reality on
retail customer's experience, satisfaction and willingness to buy. Journal of Retailing
and Consumer Services, [e-journal] 34, pp.229-234. https://doi.org/ 10.1016/
j.jretconser.2016.10.005.
Rahman, M., Rahman, M., Carbunar, B. and Chau, D.H., 2017. Search rank fraud and
malware detection in Google Play. IEEE Transactions on Knowledge and Data
Engineering, [e-journal] 29(6), pp.1329-1342. https://doi.org/10.1109/TKDE.
2017.2667658.
Ramachandran, S., Dimitri, A., Galinium, M., Tahir, M., Ananth, I.V., Schunck, C.H. and
Talamo, M., 2017. Understanding and granting android permissions: A user survey. In:
J. Ortega-Garcia, 2017 International Carnahan Conference on Security Technology.
Madrid, Spain, 23-26 October 2017. Spain: IEEE.
AE Artificial Intelligence in Retail: Benefits and Risks Associated With Mobile Shopping Applications
64 Amfiteatru Economic
Ramírez-López, F.J., Varela-Vaca, Á.J., Ropero, J., Luque, J. and Carrasco, A., 2019. A
Framework to Secure the Development and Auditing of SSL Pinning in Mobile
Applications: The Case of Android Devices. Entropy, [e-journal] 21(12), p.1136.
https://doi.org/10.3390/e21121136.
Rese, A., Ganster, L. and Baier, D., 2020. Chatbots in retailers’ customer communication:
How to measure their acceptance?. Journal of Retailing and Consumer Services,
[e-journal] 56, pp.102-176. https://doi.org/10.1016/j.jretconser. 2020.102176.
Roggeveen, A.L. and Sethuraman, R., 2020. How the COVID Pandemic May Change the
World of Retailing. Journal of Retailing, [e-journal] 96(2), pp. 169-171.
https://doi.org/10.1016/j.jretai.2020.04.002.
Shankar, V., 2018. How artificial intelligence (AI) is reshaping retailing. Journal of
retailing, [e-journal] 94(4), pp.vi-xi. https://doi.org/10.1016/S0022-4359(18)30076-9.
Singh, S.K., Rathore, S. and Park, J.H., 2020. Blockiotintelligence: A blockchain-enabled
intelligent IoT architecture with artificial intelligence. Future Generation Computer
Systems, [e-journal] 110, pp.721-743. https://doi.org/10.1016/j. future.2019.09.002
Singireddy, S.R.R. and Daim, T.U., 2018. Technology Roadmap: Drone Delivery–Amazon
Prime Air. In: T. Daim, L. Chan and J. Estep eds., 2018. Infrastructure and Technology
Management. Cham: Springer, pp. 387-412.
Spiegel, J.R., Mckenna, M.T., Lakshman, G.S. and Nordstrom, P.G., Amazon Technologies
Inc., 2011. Method and system for anticipatory package shipping. U.S. Pat. 8,086,546.
Taati, B., Zhao, S., Ashraf, A.B., Asgarian, A., Browne, M.E., Prkachin, K.M., Mihailidis,
A. and Hadjistavropoulos, T., 2019. Algorithmic bias in clinical populations –
evaluating and improving facial analysis technology in older adults with dementia.
IEEE Access, [e-journal] 7, pp.25527-25534. https://doi.org/10.1109/ACCESS.2019.
2900022.
Tang, J., Li, J., Li, R., Han, H., Gu, X. and Xu, Z., 2019. SSL Detecter: Detecting SSL
Security Vulnerabilities of Android Applications Based on a Novel Automatic Traversal
Method. Security and Communication Networks, [e-journal] 2019, pp.1-21.
https://doi.org/10.1155/2019/7193684.
The New York City Council, 2017. A Local Law in relation to automated decision systems
used by agencies. Technical Report. [pdf] The New York City Council. Available at:
<https://legistar.council.nyc.gov/LegislationDetail.aspx?ID=313781
5&GUID=437A6A6D-62E1-47E2-9C42-461253F9C6D0> [Accessed 28 July 2020].
Thinyane, H. and Sassetti, F., 2020. Towards a Human Rights-Based Approach to AI: Case
Study of Apprise. In: UNU-CS United Nations University Institute on Computing and
Society, 11th International Development Informatics Association Conference. Online,
25-27 May 2020. Macau: Springer Nature Switzerland AG.
Wadkar, M., Di Troia, F. and Stamp, M., 2020. Detecting malware evolution using support
vector machines. Expert Systems with Applications, [e-journal] 143, p.1-22.
https://doi.org/10.1016/j.eswa.2019.113022.
Wiener, N., 1960. Some moral and technical consequences of automation. Science,
131(3410), pp.1355-1358.
Artificial intelligence in wholesale and retail AE
Vol. 23 • No. 56 • February 2021 65
CONSUMERS’ PERCEPTION OF RISK TOWARDS ARTIFICIAL
INTELLIGENCE TECHNOLOGIES USED IN TRADE:
A SCALE DEVELOPMENT STUDY
Pinar Aytekin1, Florina Oana Virlanuta2*, Huseyin Guven3, Silvius Stanciu4
and Ipek Bolakca5
1),5) Izmir Democracy University, Turkey. 2),4)Dunărea de Jos University of Galati, Romania.
3)Karabaglar Guidance and Research Center, Turkey.
Please cite this article as:
Aytekin, P., Virlanuta, F.O., Guven, H., Stanciu, S.
and Bolakca, I., 2021. Consumers’ Perception of Risk
Towards Artificial Intelligence Technologies Used in
Trade: A Scale Development Study. Amfiteatru
Economic, 23(56), pp. 65-86.
DOI: 10.24818/EA/2021/56/65
Article History
Received: 25 September 2020
Revised: 8 November 2020
Accepted: 21 December 2020
Abstract
In today’s digitalizing world, internet, mobile technologies, nanotechnologies and learning
algorithms continue to develop and gain essential places in our lives. The use of artificial
intelligence in wholesale and retail trade enable better analysis of customer requests and the
development of effective marketing strategies. However, although these cutting-edge
technologies provide significant advantages to businesses, some risks may arise as these
technologies continuously develop, and it eventually becomes harder to control the pace of
development. Many famous scientists and entrepreneurs are worried that artificial
intelligence could have negative consequences for humanity if it does not develop safely,
and they suggest that urgent measures need to be taken as they believe that it may pose a
significant threat to humanity. The starting point of this study is to learn the point of view
of consumers on this technology, as well as scientists or entrepreneurs.
The purpose of this study is to determine how consumers perceive these risks. In this
direction, the literature focused on the issues related to artificial intelligence technologies in
trade was examined. In light of the information obtained from the literature, the Artificial
Intelligence Technologies Used in Trade Risk Perception Scale (AITUTRPS) was
developed.
Keywords: artificial intelligence, trade technique, risk perception, scale development
JEL Classification: D81, L81, M21
* Corresponding author, Florina Oana Virlanuta - e-mail: [email protected]
AE Consumers’ Perception of Risk Towards Artificial Intelligence Technologies Used in Trade: A Scale Development Study
66 Amfiteatru Economic
Introduction
Technologies such as artificial intelligence, robotics and mechatronic developments, big
data, Internet of Things, augmented reality, data mining, hologram and wearable electronics
(Dirican, 2015), which have been used in business life in recent years, have made it
necessary for companies to restructure their sales strategies. Applying artificial intelligence
technologies to sales strategies enables businesses to know their customers very well. In
this direction, it is possible to provide a personalized experience by sending messages,
content, and even small surprises specific to each customer. Businesses can also optimize
their sales campaigns by using artificial intelligence technologies in their marketing
activities. For example, it can be determined which loyal customers have not made
purchases from the company recently, and even unique campaigns can be offered to these
customers, and they can be regained (Kietmann, Paschien and Trieen, 2018).
With artificial intelligence technology, making transactions faster and accessing large
databases has reached a level that people cannot entirely control. For this reason, many
scientists and entrepreneurs recommend taking urgent measures regarding artificial
intelligence technology in order not to create a significant threat to humanity (Brockman,
2015). Also, some famous scientists have concerns that this technology may have
consequences for humanity if it does not develop safely (Bostrom, 2014; BBC, 2015;
Russell, 2015; The Guardian, 2015). The possibility that artificial intelligence technologies
may harm humanity as they develop is a matter of concern.
The purpose of this study is to develop a scale to measure how consumers perceive risks
related to artificial intelligence technologies used in trade. The lack of a specific scale to
measure the risk perceptions of consumers towards artificial intelligence technologies used
in trade makes this study necessary and essential in the literature. First, in the study, the
artificial intelligence technologies used in trade and the risks encountered were mentioned.
In the methodology, the stages of the scale development study were discussed. Finally, the
findings obtained from the analyses were included, and recommendations were made
within the scope of the findings and results.
1. Literature review
1.1. Artificial Intelligence Technology and Its Application in Trade
Artificial intelligence is a system that can make logical calculations, solve a problem on its
own and reprogram (Sterne, 2017). Artificial intelligence technology has reached the human
level in a shorter time than it was expected in some areas, and it has already developed in a
way beyond human control. In 1997, the world-famous chess champion Garry Kasparov was
defeated in a chess game by a computer called Deep Blue (Silver, 2012). Watson, an artificial
intelligence program developed by IBM in 2011, defeated two players in the word game
Jeopardy (PC World, 2011). In 2015, the software named Cepheus had theoretically analysed
the poker game “Heads-up Limit Texas Hold’em” and learned to bluff upon playing million
times (Bowling, et al., 2015). In this direction, it is understood that artificial intelligence is a
technology that learns very quickly, develops, and progresses rapidly.
European Research Project READ has developed Transcribus software specifically for
digitizing old books. This software uploads the captured photos to the platform server and
then digitizes them by reading the manuscripts (Euronews, 2018). Investments in artificial
Artificial intelligence in wholesale and retail AE
Vol. 23 • No. 56 • February 2021 67
intelligence technologies, which progressively develop to the present, continue to increase.
This investment, which was 9.5 billion dollars in 2018, is predicted to be 118.6 billion
dollars by 2025 (Statista, 2020).
Collecting a large amount of customer data, predicting the next step of the customer, and
using concepts such as machine learning to improve customer relations are among the use
of artificial intelligence technologies in wholesale and retail trade (Tjepkema, 2019). In
today’s business world, marketers must get to know their target audience well and predict
their possible behaviours in the increasing information stack.
Providing personalized content about brands using artificial intelligence technology is a
typical example of the use of artificial intelligence in trade. For example, thanks to an
application run by IBM Watson and the famous sportswear company Under Armor,
information and data such as the route the athlete, the duration of sports activity, the diet,
and the weather are evaluated, and recommendations for exercise and nutrition are made. In
London, M&C Saatchi positioned a camera with artificial intelligence technology on a
digital poster, and with the help of the camera, the facial movements of the people looking
at the poster were analysed and depending on the level of interest in the poster, the strategy
was modified, and the advertisements reached the target exactly (Ulutaş Ertuğrul, 2019).
In wholesale and retail trade, augmented reality applications are among the artificial
intelligence technologies that have been used in recent years. Augmented reality is about
combining the real and virtual world in real-time in a 3D environment. For example, the
Marshall augmented reality application allows the customer to make colour tests and
determine the most suitable colour for their home before painting the walls (Ayvaz, 2014).
This allows the customer to experience different paint colours, albeit virtually, and to make
a more comfortable decision.
Hologram technology is the virtual presentation of an object in 3D. For example, a
customer who wants to buy a sneaker can think as if he/she is holding the shoe thanks to
hologram technology on the Internet, which helps him make decisions easier (Kulabaş,
2017). Thiraviyam (2018) also mentioned that while watching the make-up video of a
famous person on YouTube, cosmetic products appear before the consumer and machine
vision and natural language processing technology are used to make these products
attractive to the customer.
Levi’s has created a virtual assistant that works with artificial intelligence to help users
who want to shop from websites via Facebook Messenger. While browsing the website,
users can get fashion suggestions, communicate from their mobile phones or laptops, and
experience personalized shopping from wherever they want (Levi Strauss & Co, 2017).
Therefore, the personalized experience offered through the virtual assistant will enable the
customer to visit the website and shop more.
West Elm, which offers home decoration products, tries to offer personalized
recommendations with the help of Pinterest Style Finder by making use of the customer
data on Pinterest, which is the world’s most extensive idea-collection on the Internet; that
is, the company first recognizes the previous posts of customers and modifies its
recommendation according to their colour and style preferences (O’Shea, 2017).
The American Mail Company UPS allocates $1 billion a year for technology investments and
makes use of artificial intelligence technologies to save over $200 million annually in all its
AE Consumers’ Perception of Risk Towards Artificial Intelligence Technologies Used in Trade: A Scale Development Study
68 Amfiteatru Economic
global operations. By using an artificial intelligence chatbots to improve customer experience,
UPS can respond wholly and quickly to customers’ questions about tracking their shipments.
These chatbots can even learn pricing information and connect to chat with customers through
many social communication channels. UPS also uses ORION (On-Road Integrated
Optimization and Navigation), a highly sophisticated artificial intelligence platform developed
by the company that is at the centre of the company’s operations. This platform plans and
optimizes the routes of loads received by UPS drivers with algorithms. Thus, as soon as
packages are received, fast, timely and efficient routes are created (Şentürk, 2018). In terms of
time and effort spent, this application is very advantageous for a mail company.
The artificial intelligence-based smart personal shopping assistant ShopBot in eBay’s
Facebook Messenger application can recommend the most suitable products from the list of
one billion products. This assistant, with which customers can easily communicate, asks
them questions to better understand their requests and needs and directs them to suitable
products accordingly (Pittman, 2016). Thus, the customer can easily find an affordable and
wished product without wasting much time.
Generally, it is difficult for a business to provide a personalized experience to its customers
if the number of customers is high. However, it becomes easier to offer this experience with
artificial intelligence technology. However, many scientists have warned that continuously
developing artificial intelligence technologies may have adverse effects. The CEO of Tesla
Motors, entrepreneur and investor Elon Musk stated that the security of artificial
intelligence technology is fundamental and he donated $10 million to an artificial
intelligence research institute that aims to make artificial intelligence beneficial to
humanity (The Guardian, 2015).
Stephen Hawking stated that artificial intelligence is very advanced and useful. However,
he said he was worried that it could reach a level that could surpass human intelligence.
According to Stephen Hawking, “Artificial intelligence can continue to improve itself and
even reformat itself. Human beings, limited to a prolonged biological evolution, cannot
compete with this kind of power” (Jones, 2014). In the 100-page report published by
international security experts on the probable dangers of artificial intelligence, they warned
the developers of artificial intelligence to make more efforts to prevent the abuse of
technology (BBC, 2018). This report shows that artificial intelligence is beginning to be
recognized as dangerous if not used correctly and reliably in the future.
1.2. Possible Risks Related to Artificial Intelligence Technologies Used in Wholesale
and Retail Trade
Artificial intelligence technology, as mentioned above, is a technology that provides
benefits in many areas, but it can cause dangerous consequences. Shahar Avin, from the
Centre for Existential Risk Research at the University of Cambridge, spoke of scenarios
that artificial intelligence could be used maliciously soon. According to these scenarios,
“AlphaGo game, developed by Google’s artificial intelligence company DeepMind, may be
used by hackers to find new ways to exploit new patterns and codes, a malicious person
may purchase a drone and target a specific person by training it with face recognition
software; also, bots can be automated, fake videos may be used for political manipulation,
or hackers may even use voice synthesis to impersonate targets” (Wakefield, 2018).
Artificial intelligence in wholesale and retail AE
Vol. 23 • No. 56 • February 2021 69
Artificial intelligence technology is also used in many marketing-related areas such as
lowering advertising costs, providing better content to customers, audience targeting and
market segmentation. 88% of marketers believe that investing in artificial intelligence
technology will have a positive effect (Econsultancy, 2018), but although artificial
intelligence technology contributes positively to the wholesale and retail trade field, it can
also have negative consequences when not used correctly and carefully.
Marketers realize that in every field they choose to use artificial intelligence technologies,
marketing strategies are reshaped, and the business is significantly affected. On the other
hand, the threats that this technology may bring along should not be ignored. In 2016, a
driverless vehicle developed by Nvidia set off in Monmouth, New Jersey, and this vehicle did
not even follow a single instruction provided by the engineer or programmer. The car learned
to drive by itself, watching how the people in the other vehicles were driving. Possibilities
such as the driverless vehicle hitting the tree in the future or suddenly stopping at the green
light worry the experts (Knight, 2017). This example shows that some security issues may
arise. Another security issue occurred in Amazon’s warehouse in the state of New Jersey.
Twenty-four employees were hospitalized as an artificial intelligence robot accidentally broke
the bear removal spray in the warehouse (CNNTURK, 2018). It is worrying to see that
gradually improving artificial intelligence causes such an accident and harms the employees.
Therefore, it is critical to control such robots very well and prepare the necessary environment
for safe and decent operation. Otherwise, unexpected results may occur.
There are growing concerns about artificial intelligence (AI) violating personal privacy. The
world-famous Nike company receives customized shoe orders. However, while the customer
chooses the shoe, the brand can collect various data in the background and develop various
strategies by using estimated models (Antonio, 2017). In this sense, using customer data
without informing the customer for future sale strategies is an ethical question.
In the digital world, persuasion architecture can be built to reach billions thanks to artificial
intelligence technology (big data and machine learning). Ads can target individuals one by
one, penetrate them on a personal level by detecting their weak points, and these ads can
even be sent to everyone’s phone screen. Especially, companies such as Facebook, Google,
and Amazon use algorithms to click on ads (Tüfekçi, 2018). There are concerns that
computers have begun to take over business skills, including the quality of thinking, as they
become more and more established in our lives (Dedeoğlu, 2006).
Although artificial intelligence is an essential innovation in trade applications, they cause
more and more people to lose their jobs. According to the May 2017 report of Cornstone
Capital Group, approximately 6 million job losses are foreseen in the retail sector (Shavel,
Vanderzeil and Currier, 2017). In the retail industry, especially salespeople, cashiers,
transport workers, and order clerks have roles open to automation. E-commerce giant
Amazon has developed the Amazon Go name as a cashier-free store form. In this format,
the content of shopping carts is automatically detected, and sensors perform the billing
process when customers leave the store (Morris, 2017). At Cafe X in San Francisco, coffee
is prepared by a robot. This robot prepares the coffee within a minute as specified when
ordered, receives the payment, and waits for the customer to come and receive the coffee
within six minutes (Yamak, 2017). If such examples increase, job loss will be inevitable for
baristas as well. The job loss problem also applies to the management level. In a study
conducted by Young and Cormier (2014), the idea of whether robots should be managers or
AE Consumers’ Perception of Risk Towards Artificial Intelligence Technologies Used in Trade: A Scale Development Study
70 Amfiteatru Economic
not was investigated. After all, although a human manager is likely to be perceived as an
authority, about 46% of the participants obeyed the robots.
Using a system based on artificial intelligence has some drawbacks because artificial
intelligence has quite a few shortcomings in terms of morality and ethical values (Parry,
Cohen and Bhattacharya, 2016). The relationship of artificial intelligence with ethical
values has become heavily debated, especially in recent years. In particular, the problem of
emotions not being entirely modelled causes paradoxes, such as how an artificial
intelligence-based system should make decisions at the stage of ethical dilemmas (Köse,
2018a). Judgment quality, which enables to distinguish between good/bad, right/wrong, and
the quality of decision making by combining reason and conscience, are critical in artificial
intelligence. Whether artificial intelligence can bear responsibility for a problem, its
accountability, obligations, and which powers can be assigned/delegated to artificial
intelligence are considered essential issues (Dedeoğlu, 2006). It is a question of how
artificial intelligence can make sense of ethical values that are cared for by people. It is a
matter of curiosity to what extent a machine that learns on its own and whose behaviour
cannot be directed will comply with ethical values in the future (Köse, 2018b). When a
driverless vehicle’s brakes fail, it is a matter of concern how it will make a sensible
decision when it accelerates into a crowded crosswalk, faced with the difficult choice of
mowing down a large group of older people or steering a woman pushing the stroller (The
Associated Press, 2017).
As it is often challenging to document the processes performed by artificial intelligence
systems, it may take a long time for such applications to gain trust (Henkoğlu, 2019). In an
experiment conducted at the University of Bologna, the two opposing algorithms did not
fight each other to find the lowest and most competitive price but instead misled users by
offering high-priced products. The fact that the artificial intelligence algorithms did not
leave any trace that they performed joint action was found to be alarming for the scientists
who conducted the experiment. This shows that AI algorithms can learn to cooperate and
oppose people without communicating or being told to do so. According to the popular
science and technology magazine called Mechanics, such algorithms currently account for
the vast majority of prices in online marketplaces such as Amazon (Altan, 2019). These
experiences show that we may encounter ethical issues related to artificial intelligence.
2. Research methodology
While the use of artificial intelligence in wholesale and retail trade is gradually increasing,
it also brings some problems. These problems can change the perspective on this
technology. As mentioned above, some risks may arise regarding this technology. The
purpose of this study is to determine how consumers perceive these risks. In this direction,
it is aimed to develop the Artificial Intelligence Technologies Used in Trade Risk
Perception Scale (AITUTRPS) to measure the risk perceptions of consumers towards
artificial intelligence technologies used in trade. The absence of a scale on this subject in
the literature makes the research valuable. Since this research is a perception scale
development study, scanning method was used. Large-scale samples are not required as
item analysis, and consistency of items with the whole scale will be examined instead of the
validity and reliability of the scale in pilot applications. If the number of items on the scale
in the pilot application is up to 30, then a sample size of around 50 will be sufficient (Seçer,
2015). Since the pilot application of the study was conducted on 51 people, it meets the
Artificial intelligence in wholesale and retail AE
Vol. 23 • No. 56 • February 2021 71
sample size. If multivariate analyses are to be used in the research, the conditions required
by these analyses should be considered as the sample size.
In cases where exploratory and confirmatory factor analysis will be used, a sample size of
over 200 is considered sufficient in terms of reliability (Kline, 2005). Bryman and Cramer
(2001) stated that the value obtained by multiplying the number of items in the scale by five
or ten should be taken as a criterion in determining the number of participants. While Ho
(2006) suggests that the sample size should not fall below 100, he mentions that this
number should exceed five times the number of variables (items), from a more acceptable
point of view, the number of items should be ten times. Since the main application was
carried out on 205 people, sufficient sample size was met to perform factor analysis.
The data obtained as a result of the application of the scales were subjected to statistical
evaluation with SPSS 22.0 and AMOS 20.0 statistical data analysis programs. The data
collection tool used in the study consists of two parts: The Artificial Intelligence
Technologies Used in Trade Risk Perception Scale (AITUTRPS) and the Demographic
Information Form. The gender, age, profession, income, and education levels of the
participants were asked in the Demographic Information Form. The applications during the
design of AITUTRPS are as follows:
a) Item Creation Stage
At this stage, the sources in the literature such as: Dedeoğlu, 2006; Bostrom, 2014; Young
and Cormier, 2014; Brockman, 2015; Russell, 2015; The Guardian, 2015; Parry, Cohen and
Bhattacharya, 2016; Antonio, 2017; Knight, 2017; Morris, 2017; Shavel, Vanderzeil and
Currier, 2017; The Associated Press, 2017; Yamak, 2017; BBC, 2018; CNNTURK, 2018;
Köse, 2018a,b; Tüfekçi, 2018; Wakefield, 2018; Altan, 2019; Henkoğlu, 2019 were used to
form the scale, and while preparing these items, attention was paid not to include more than
one judgment/perception. Besides, positive items used in the scale were graded as
I strongly agree and I agree, while negative items were graded as Strongly disagree and
Disagree. The Undecided option was used for items that did not contain a positive or
negative statement.
b) Consultation to Expert Opinion Stage
There are three types of validity: scope (content), compliance and structure. One or more of
these validity types are suitable according to the characteristics of the measurement tool
used in the study. At this stage, the adequacy of the scale in terms of scope (content)
validity was examined. Content validity exactly means the degree to which items represent.
This validity is provided when the items are first selected for the test (Field and Hole,
2019). In determining the scope validity of the scale, the opinions of economics and
engineering faculty members and business administration professionals were consulted.
Expert opinions were obtained through a form that provides three categories of evaluation:
item needed, item useful but not sufficient and item unnecessary. Some modifications have
been made by considering the opinions and criticisms made by experts. The scope validity
of AITUTRPS was ensured by considering the collective opinions of the experts.
c) Preliminary Trial Phase
A draft scale consisting of a total of 20 items was created for the pre-trial and applied to
51 people in total in October 2019. In the study, the reliability (internal consistency) of the
scale to determine the participants’ risk perceptions about artificial intelligence
AE Consumers’ Perception of Risk Towards Artificial Intelligence Technologies Used in Trade: A Scale Development Study
72 Amfiteatru Economic
technologies used in wholesale and retail trade was examined by item analysis, and in
addition to item analysis based on the difference between lower-upper group averages, item
analysis based on correlation was also performed. The Cronbach alpha reliability
coefficient is widely used to determine the reliability of the scales used to measure
psychological characteristics. The Cronbach Alpha Reliability Coefficient of the scale
consisting of 20 items was determined as (α=0.899). The Cronbach Alpha reliability
coefficient value is exceptionally reliable over 0.90, highly reliable between 0.90 and 0.80,
and reliable between 0.79 and 0.70 (Cohen, Manion and Morrison, 2007). According to
these values, the draft scale seems to be highly reliable. While item selection was made
using the item analysis technique based on the difference between the sub-upper group
averages (based on the internal consistency criterion) applied to 20 items in the trial form of
the scale, the scale scores of the individuals were ranked in descending order. According to
this ranking, 14 people constituting the first 27% of the group of 51 people were
determined as the upper group, and the last 14 people constituting the last 27% as the sub
group. For each scale item of the 27% lower-upper groups at both ends of the scale scores
distribution, the difference between the t-test for independent groups and their averages was
examined. The t-test results for the item averages of the Lower 27% and Upper 27% groups
of the scale are shown in Table 1.
Table no. 1. T-Test Results for the Item Averages of the Lower 27%
and Upper 27% Groups of the Scale
Item N Mean Std. d. df t p
Item 1 Upper Group 14 4.43 0.51
26 6.500 0.000 Lower Group 14 2.58 0.94
Item 2 Upper Group 14 4.14 0.53
26 9.774 0.000 Lower Group 14 2.29 0.47
Item 3 Upper Group 14 4.64 0.50
26 4.053 0.000 Lower Group 14 3.14 1.29
Item 4 Upper Group 14 4.07 0.92
26 3.606 0.001 Lower Group 14 2.71 1.07
Item 5 Upper Group 14 4.00 0.78
26 3.994 0.000 Lower Group 14 2.71 0.91
Item 6 Upper Group 14 4.43 0.51
26 3.045
0.005 Lower Group 14 3.50 1.02
Item 7 Upper Group 14 3.43 1.02
26 1.427 0.166 Lower Group 14 2.93 0.83
Item 8 Upper Group 14 3.14 0.86
26 2.550 0.017 Lower Group 14 2.26 0.91
Item 9 Upper Group 14 4.29 0.47
26 6.379 0.000 Lower Group 14 2.64 0.84
Item 10 Upper Group 14 4.21 0.80
26 3.823 0.001 Lower Group 14 3.00 0.88
Artificial intelligence in wholesale and retail AE
Vol. 23 • No. 56 • February 2021 73
Item N Mean Std. d. df t p
Item 11 Upper Group 14 4.21 0.58
26 4.172 0.000 Lower Group 14 2.93 1.00
Item 12 Upper Group 14 4.00 0.68
26 4.163 0.000 Lower Group 14 2.57 1.09
Item 13 Upper Group 14 4.07 0.92
26 5.712 0.000 Lower Group 14 2.29 0.73
Item 14 Upper Group 14 4.07 0.62
26 5.230 0.000 Lower Group 14 2.50 0.94
Item 15 Upper Group 14 4.43 0.65
26 3.107 0.005 Lower Group 14 3.43 1.02
Item 16 Upper Group 14 3.71 0.83
26 6.000 0.000 Lower Group 14 2.00 0.68
Item 17 Upper Group 14 3.71 0.83
26 6.000 0.000 Lower Group 14 2.00 0.68
Item 18 Upper Group 14 4.71 0.47
26 4.639 0.000 Lower Group 14 3.00 1.30
Item 19 Upper Group 14 4.57 0.51
26 4.281 0.000 Lower Group 14 3.36 0.93
Item 20 Upper Group 14 4.50 0.52
26 3.513 0.002 Lower Group 14 3.43 1.02
To have a highly related group of items, each item must be highly correlated with the
remaining items. For each item, this property can be analysed by calculating the item-total
correlation (DeVellis, 2017). Correlation is used in the reliability studies of measurement
tools. Within the scope of the reliability studies of the scale, after the Cronbach's alpha
internal consistency coefficient, the corrected item-total correlation of the scale was
calculated and the item total correlation explains the relationship between the points
obtained from the items in the measurement tool and the total score. The high level of item-
total correlation indicates that the items in the measurement tool exemplify similar
behaviors and the internal consistency of the scale is high (Büyüköztürk, 2017). In the item
analysis process of the scale, it is considered appropriate to exclude items with item-total
correlations of 0.30 or less from the scale (Geuens and Pelsmacker, 2002). Results
regarding the item analysis of the scale are shown in Table 2.
Table no. 2. Item Analysis Results of the Scale
Item Item Total Correlation* t
(Upper% 27-Lower %27)**
Item 1 0.696 6.500***
Item 2 0.616 9.774***
Item 3 0.620 4.053***
Item 4 0.450 3.606***
Item 5 0.444 3.994***
AE Consumers’ Perception of Risk Towards Artificial Intelligence Technologies Used in Trade: A Scale Development Study
74 Amfiteatru Economic
Item Item Total Correlation* t
(Upper% 27-Lower %27)**
Item 6 0.325 3.045***
Item 7 0.079 1.427
Item 8 0.134 2.550
Item 9 0.647 6.379***
Item 10 0.479 3.823***
Item 11 0.531 4.172***
Item 12 0.585 4.163***
Item 13 0.588 5.712***
Item 14 0.571 5.230***
Item 15 0.536 3.107***
Item 16 0.641 6.000***
Item 17 0.652 6.000***
Item 18 0.710 4.639***
Item 19 0.647 4.281***
Item 20 0.607 3.513***
Note: Significant values for *n = 51, **n1= n2= 14, ***p <0.05.
As can be seen in Table 1 and Table 2, as a result of the analyses performed, the 7th and 8th
items with p>0.05 and correlation coefficients of r≤0.30 for the item averages almost do not
contribute to the scale. For this reason, they were excluded from the scale.
3. Findings and Discussion
The draft scale developed after the pre-application was applied to a total of 205 consumers
in November and December and the statistical calculations of the obtained data were made.
Information on the demographic characteristics of the people who participated in the study
is given in Table 3.
Table no. 3. Demographic Characteristics of the Participants
Frequency (n) Percent (%)
Gender Female 100 48.8
Male 105 51.2
Age
18-28 35 17.1
29-39 85 41.5
40-50 63 30.7
51-61 17 8.3
61 and over 5 2.4
Educational
Status
Primary School 2 1
High School 23 11.2
University 107 52.2
Graduate 73 35.6
Artificial intelligence in wholesale and retail AE
Vol. 23 • No. 56 • February 2021 75
Frequency (n) Percent (%)
Profession
Unemployed 7 3.4
Freelancer 8 3.9
Government Official 98 47.8
Worker 6 2.9
Private sector 44 21.5
Housewife 3 1.5
Retired 9 4.4
Student 21 10.2
Others 9 4.4
Monthly
Income
1000 TL and below 24 11.7
1001-2000 13 6.3
2001-3000 21 10.2
3001-4000 29 14.1
4001-5000 43 21
5001TL and above 75 36.6
TOTAL 205 100
When the participants are examined by gender, it is seen that the number of female and male participants is remarkably close to each other. Nearly half of the participants are 50 years old or under. Also, it is seen that more than half of the participants are university graduates. When the professions of the participants are examined, it is understood that civil servants and students are predominant. These two groups constitute more than 50% of the research. When evaluated in terms of income level, it is seen that most of the participants have an income of 5001 TL or more.
In the next step, the reliability of the scale was calculated. Reliability is the ability of a measurement to produce the same results under the same conditions (Field and Hole, 2019). After the main application, item analysis was applied again for the 18-item scale. The Cronbach Alpha Reliability Coefficient of the scale consisting of 18 items was determined as (α=0.917). At the end of the main application, both item analysis based on the difference between lower and upper group averages and item analysis based on correlation were performed. When the results are examined, the t values of the scale developed in the study are significant, and item-total correlations for all items vary between 0.457 and 0.726. Hence, it can be said that all items in the scale are good, have high reliability, and are aimed at measuring similar behavior.
In determining the construct validity of AITUTRPS, first Exploratory Factor Analysis (EFA) and then Confirmatory Factor Analysis (CFA) were applied. Before the exploratory and confirmatory factor analyses were made, the statistical assumptions required for these analyses were checked, and the analyses were carried out after the relevant assumptions were met.
Exploratory factor analysis is often used at the beginning of research to gather more information about the interrelationships between a set of variables. Confirmatory factor analysis, on the other hand, is a more complex and advanced technique used in the later stages of the research to test specific theories and hypotheses about the underlying structure of a number of variables (Pallant, 2016).
AE Consumers’ Perception of Risk Towards Artificial Intelligence Technologies Used in Trade: A Scale Development Study
76 Amfiteatru Economic
The first of the values examined in the exploratory factor analysis is the Kaiser-Meyer-
Olkin Sampling Adequacy Scale (KMO) and Bartlett’s Sphericity test results, which are
used to determine whether the sample is suitable for factor analysis. If the value of the
Kaiser-Meyer-Olkin test is below 0.50, factor analysis is not continued. If the value for the
sample size is between 0.50-0.60, it is interpreted as Bad, if between 0.60-0.70, Poor, if
between 0.70-0.80, Medium, if between 0.80-0.90, Good and if it is above 0.90, Excellent.
(Leech, Barrett and Morgan, 2005).
The KMO value for AITUTRPS was calculated as 0.840. This value indicates that the
sampling efficiency of the research is at a good level for factor analysis. Also, when the
Barlett sphericity test results were examined, it was seen that the chi-square value was
significant (χ2 (91) = 2056.359; p<0.01). In this direction, it was accepted that the data
came from multivariate normal distribution. In line with these findings, exploratory factor
analysis was applied to the data.
In order to determine the factor pattern of the scale consisting of 18 variables, Principal
Component Analysis was used as the factoring method, and Varimax rotation, one of the
most frequently used rotations which gives the most sensitive distinction between factors,
was used in order to interpret the factors more meaningfully as a rotation method (Ho,
2006). As a result of the analysis, it was found that AITUTRPS had four factors and the
ratio of explaining the total variance of these four factors was 77.905%. Generally, it is
recommended that the total variance explained by all factors in exploratory factor analysis
should be at least 30% in unidimensional scales and at least 50% in multidimensional scales
(Streiner, 1994). Accordingly, the variance value explained by AITUTRPS is quite high.
When each factor is examined, the first factor explains 22,055% of the total variance, the
second factor explains 19,836% of the total variance, the third factor explains 18,129% of
the total variance, and the fourth factor explains 17,885% of the total variance. In Table 4,
four factors and factor load of each variable belonging to these factors are shown.
Table no. 4. AITUTRPS Exploratory Factor Analysis Results
Variables Components
1 2 3 4
Harm_Employees 0.862
Harm_Customers 0.843
Harm_People 0.837
Watching_Humans 0.770
Protect_Data 0.803
Hack_Accounts 0.794
Capture_Personal_Information 0.752
Use_Maliciously 0.739
Affect_Decision 0.892
Direct_To_Purchase 0.837
Click_On_Ads 0.752
Service_Sector 0.892
Manufacturing_Sector 0.851
Retail_Sector 0.835
Artificial intelligence in wholesale and retail AE
Vol. 23 • No. 56 • February 2021 77
In the exploratory factor analysis, a factor load should be at least 0.30. Also, the difference
between the factor loads given by a variable to more than one factor must be at least 0.1
(Stevens, 2002). Factor load is the correlation of an item or an observed variable with the
relevant factor. Or it is the rate at which variables contain the essence or hidden variable of
each factor. The square of the factor loading of an item within the factor shows how much
of the variance in the relevant factor explains. A low factor load value of an item indicates
that that item is insufficient to explain the relevant factor (Gürbüz and Şahin, 2018).
According to Tabachnick and Fidell (2013), the factor load of an item on a factor should be
minimum 0.32. As a result of this analysis, the factor loads of the 3rd, 9th, 10th and 13th
items were removed from the scale since the factor loads were spread over more than one
dimension and these items were located in a dimension different from the theoretically
determined ones. With the exclusion of these items from the scale, the general reliability
level of the scale consisting of 14 items, was calculated as 0.917, which shows that the
reliability of the scale is quite high. Detailed information on the factors resulting from the
analysis is given below.
Factor 1. Security Issue
Four items were given under the first factor, and these items, load values and other
statistical values are shown in Table 5. This factor was named Security Issues.
Table no. 5. Information on the Security Issues Factor
FACTOR 1. SECURITY ISSUES
Item Factor Load Common Variance
Harm_Employees 0.862 0.861
Harm_Customers 0.756 0.843
Harm_People 0.739 0.779
Watching_Humans 0.685 0.643
Cronbach’s Alpha: 0.897, Eigenvalues: 6.287, Variance Explained: 22.055
It was determined that the factor loadings of the Security Issues were between 0.685 and
0.862. Cronbach’s Alpha value of the factor was 0.897. This value shows that the factor is
highly reliable. The security issues factor explains 22,055% of the total variance.
In Factor Analysis, an item can correlate with more than one factor. Common variances are
the sum of squares of correlation with the factors with which an item is associated. The
extraction value corresponding to the common variance value of an item shows the total
variance explained by that item (Gürbüz and Şahin, 2018) When interpreting common
variance values, it is stated that 0.50 is generally required as a criterion (Thompson, 2004).
However, it is not always possible to reach high common variance values in the social
sciences. Therefore, Costello and Osborne (2005) stated that it would be more appropriate
to take a value of 0.40 for the common variance as a criterion. Tabachnick and Fidell
(2013) stated that items with a common variance of less than 0.20 indicate the
heterogeneity between items. It is possible to say that the common variance of all items is
good since the common variance values for the Security Issues factor is between 0.643 and
0.861.
AE Consumers’ Perception of Risk Towards Artificial Intelligence Technologies Used in Trade: A Scale Development Study
78 Amfiteatru Economic
Factor 2. Violation of Personal Privacy
Four items were given under the second factor, and these items, load values and other
statistical values are shown in Table 6. This factor was named Personal Privacy Violation.
Table no. 6. Information on Personal Privacy Violation Factor
FACTOR 2. VIOLATION OF PERSONAL PRIVACY
Item Factor Load Common
Variance
Protect_Data 0.803 0.705
Hack_Accounts 0.794 0.765
Capture_Personal_Information 0.752 0.773
Use_Maliciously 0.739 0.740
Cronbach’s Alpha: 0.873, Eigenvalues: 1.882, Variance Explained: 19.836
It was determined that the factor loadings of the privacy violation are between 0.739 and
0.803. Cronbach’s Alpha value of the factor is 0.873. This value shows that the factor is
highly reliable. The Personal Privacy Violation factor explains 19.836% of the total
variance. Since the common variance values for the Personal Privacy Violation factor are
between 0.705 and 0.773, it can be said that the common variance of all items is good.
Factor 3. Ethical Issues
Three items were given under the third factor, and these items, load values and other
statistical values are shown in Table 7. This factor was named Ethical Issues.
Table no. 7. Information on the Ethical Issues Factor
FACTOR 3: ETHICAL ISSUES
Item Factor Load Common Variance
Affect_Decision 0.892 0.871
Direct_To_Purchase 0.837 0.792
Click_On_Ads 0.752 0.695
Cronbach’s Alpha: 0.866, Eigenvalues: 1.671, Variance Explained: 18.129
Factor loads of Ethical Issues were determined to be between 0.752 and 0.892. Cronbach’s
Alpha value of the factor is 0.866. This value shows that the factor is highly reliable. The
Personal Privacy Violation factor explains 18.129% of the total variance. Since the
common variance values for the Ethical Issues factor are between 0.695 and 0.866, it can be
said that the common variance of all items is good.
Factor 4. The Employment Issue
Three items were given under the fourth factor, and these items, load values and other
statistical values are shown in Table 8. This factor was named Employment Issue.
Artificial intelligence in wholesale and retail AE
Vol. 23 • No. 56 • February 2021 79
Table no. 8. Information on the Employment Issue Factor
FACTOR 4: THE EMPLOYMENT ISSUE
Item Factor Load Common Variance
Service_Sector 0.892 0.862
Manufacturing_Sector 0.851 0.801
Retail_Sector 0.835 0.777
Cronbach’s Alpha: 0.866, Eigenvalues: 1.068, Variance Explained: 17.885
Factor loadings of the Employment Issue were determined to be between 0.835 and 0.892.
Cronbach’s Alpha value of the factor is 0.866. This value shows that the factor is highly
reliable. The Employment Issues factor explains 17.885% of the total variance. Since the
common variance values for the employment issues factor are between 0.777 and 0.862, it
can be said that the common variance of all items is good. In the process of developing a
scale, explanatory factor analysis is used as an explanatory first step, while confirmatory
factor analysis can be used as a second step to check the operability of a structure defined
by exploratory factor analysis (Harrington, 2009). In this direction, confirmatory factor
analysis was performed on the data.
Many fit indices are used to determine the adequacy of the model tested in the confirmatory
factor analysis. Since the fit indices have strengths and weaknesses relative to each other in
evaluating the fit between the theoretical model and the real data, different fit indices
should be used to comment on the fit of the model. Fit indices were used in the study to test
the accuracy and fit of the model determined by factor analysis.
As a result of the exploratory factor analysis, the compatibility of the remaining 14 items of
AITUTRPS to the model was tested through confirmatory factor analysis. The values of the
scale items were examined in terms of fit indices. The relevant dimensions of AITUTRPS,
which was determined to consist of four factors by applying exploratory factor analysis,
were tested with first-level multifactorial confirmatory factor analysis. For this purpose, the
model of primary level multifactorial confirmatory factor analysis was created. The latent
factors in the scale and the mutual effects between these factors were evaluated within the
framework of the model created. First Level Multi-Factor Confirmatory Factor Analysis
Diagram for AITUTRPS is shown in Figure 1.
According to CFA, it was determined that the structural equation model result of the scale
was significant at the level of p=0.000, and the 14 items and four sub-factors forming the
scale were related to the scale structure. Also, improvements were made in the model.
While making improvements, variables that decrease the fit value were determined, and
new covariance’s were created for those with high covariance among residual values. In the
renewed fit index calculations, acceptable values for fit indices are shown in Table 9.
AE Consumers’ Perception of Risk Towards Artificial Intelligence Technologies Used in Trade: A Scale Development Study
80 Amfiteatru Economic
Figure no. 1. First Level Multi-Factor Confirmatory Factor Analysis Diagram
for AITUTRPS
Table no. 9. AITUTRPS First Level Multifactor Confirmatory Factor Analysis
Fit Indices
Fit
Criteria Good Fit*
Acceptable
Fit*
Calculated
Value Before
Modification
Calculated
Value After
Modification
Fit Status
CMIN/Df 0 ≤ χ2/df ≤ 3 3 ≤ χ2/df ≤ 5 2.659 2.083 Good Fit
GFİ ≥ 0.95 ≥ 0.90 0.892 0.916 Acceptable Fit
CFI ≥ 0.95 ≥ 0.90 0.942 0.963 Good Fit
RMSEA ≤ 0.05 ≤ 0.08 0.090 0.073 Acceptable Fit
When the values in Table 9 are analysed, it is seen that the CMIN/df after the modification
is 2.083, the GFI value is 0.916, the CFI value is 0.963, and the RMSEA value is 0.073.
Since these values are within acceptable limits, the four-factor structure of AITUTRPS was
confirmed. In addition, factor loads for each factor are given in Table 10.
Artificial intelligence in wholesale and retail AE
Vol. 23 • No. 56 • February 2021 81
Table no. 10. Factor Loadings as a Result of Confirmatory Factor Analysis
for AITUTRPS
Factors and Items Factor
Loadings
Security Issues
I am afraid that an AI robot used as a cashier, aisle or warehouse attendant
in a store could spiral out of control and harm other employees. 0.96
I am afraid that an AI robot used as a cashier, aisle or warehouse attendant
in a store could spiral out of control and harm customers. 0.97
I am afraid that a product controlled by artificial intelligence technology
(e.g. a driverless vehicle) could spiral out of control and harm people. 0.73
I think that an AI product (e.g. robot, drone, smart car) can lead to
undesirable results when it learns by watching human behaviour patterns. 0.58
Violation of Personal Privacy
I think that enough measures have not been taken to protect my data
collected by using artificial intelligence technology. 0.67
When I shop from a site that uses artificial intelligence technology, I am
concerned that fraudsters may hack my accounts. 0.85
I think that my personal information can be captured because of the artificial
intelligence technologies used while shopping on the Internet. 0.84
I am concerned that the information obtained by artificial intelligence
technology can be used maliciously (acting on behalf of the person against
his will, etc.).
0.79
Ethical Issues
I think that artificial intelligence technology affects the decision-making
process of an individual by learning and monitoring the preferences on the
Internet.
0.91
With artificial intelligence technology, I think an individual is monitored on
the Internet and directed to purchase through his preferences. 0.86
I think companies develop computer-based applications on the Internet for
us to click on ads. 0.73
The Employment Issue
I think that artificial intelligence technology may unemploy people working
in the service sector (robot waiter, etc.) in the future. 0.91
I think that artificial intelligence technology may unemploy people working
in the manufacturing sector (robots, etc.) in the future. 0.84
I think that artificial intelligence technology may unemploy people working
in the retail sector (store without a cashier, etc.) in the future. 0.80
Conclusions
Artificial intelligence technologies, which are increasingly prominent today, have positive
impacts on many industries. It is a fact that these technologies are gradually getting more
and more attention from companies and investors. Artificial intelligence technology
provides advantages in various issues such as analysing the demands and needs of
customers well, offering the right product for them, making price comparison, especially in
AE Consumers’ Perception of Risk Towards Artificial Intelligence Technologies Used in Trade: A Scale Development Study
82 Amfiteatru Economic
the field of wholesale and retail trade, and it enables businesses to develop effective and
efficient sale strategies in the digital environment. Developing artificial intelligence
technologies, unfortunately, cause differences of opinion between scientists and business
people. While some defend the positive aspects of this technology, many scientists and
businessmen claim that it may cause various problems and even be dangerous if not used
carefully.
When the literature on the use of artificial intelligence in wholesale and retail trade is
examined, it is seen that there are some issues related to the artificial intelligence
technologies used in selling, such as security, privacy, employment, and ethics. When these
problems are evaluated, it is predicted that artificial intelligence technology can cause
security problems when it is out of control, it can violate personal privacy as it reaches
billions of personal information, and it may cause many job losses and increase
unemployment, especially in the retail and service sector. It is also understood that these
technologies may ignore some ethical values.
While many applications with artificial intelligence technologies, such as driverless vehicles,
facial recognition technologies, and personal assistants that offer suggestions by analysing
human face, continue to develop and bring many advantages, the issues that this technology
may create must be taken seriously. These technologies, which may become difficult to
control as they develop, must be planned and implemented carefully and cautiously. In the
study, considering the information obtained from the literature, Artificial Intelligence
Technologies Used in Trade Risk Perception Scale (AITUTRPS) was developed. The reason
for the development of this scale is that there is no specific scale in the literature to measure
the risk perceptions of consumers towards artificial intelligence technologies used in wholesale and retail trade. Thus, it is aimed to fill this gap in the literature. Expert opinions
were used in determining content validity during the scale creation phase. The scale was
piloted in the first stage, and items that reduced validity and reliability were removed from the
scale. The draft scale after this process was applied to 205 people, and the results were
subjected to explanatory and confirmatory factor analysis. In this part, it can be said that the
use of both explanatory factor analysis and confirmatory factor analysis makes the study more
reliable. As a result of the analysis, four dimensions, namely security issue, violation of
personal privacy, unemployment issues and ethical issues, occurred. The rate of explaining
the total variance of these four factors was found to be 77.905%.
Among the security issues, there are problems such that an artificial intelligence robot used
as a cashier, aisle or warehouse clerk in a store can get out of control and harm other
employees or customers. To avoid these problems, it is beneficial to make continuous
controls of this robot, and it should be monitored by an expert. The possibility of a product
such as a driverless vehicle controlled by artificial intelligence technology to get out of
control and harm people is another security-related issue. This issue should be prevented at
the production stage, and all necessary measures should be taken to prevent such a
situation. As for the violation of personal privacy, there are concerns that personal data
collected by artificial intelligence technology may be intercepted, misused and that this data
is not adequately protected. It is useful to prevent this problem with necessary software and
legal regulations and sanctions. With artificial intelligence technology, ethical issues such
as monitoring the preferences of the person on the Internet and learning their preferences
and affecting their purchasing decision process or directing them to purchase are also
encountered. Firms should not include such practices that take place against the will of the
Artificial intelligence in wholesale and retail AE
Vol. 23 • No. 56 • February 2021 83
consumers, because this may damage the image of the firm or negatively affect the attitudes
of consumers towards the firm or the brand. The fact that artificial intelligence technology
may leave people working in the production, service, and retail sectors unemployed in the
future creates employment issues. Regarding this issue, states and authorities must take
precautions and limit the number of robots working with artificial intelligence technology.
This study is significant in terms of addressing the ignored AI risks while people
continuously highlight the positive aspects of the AI in today’s world. In this direction, the
developed scale is likely to become a basis for researchers who want to work on this subject
in the future. By using this scale in the future, the subject can be addressed from various
angles and different studies can be conducted. Besides, by using this scale, the differences
in perception of risks related to artificial intelligence technologies used in wholesale and
retail trade in terms of demographic characteristics such as gender, age, income and
profession, or how these risks are perceived by people from different segments such as
trade professionals, marketing professionals, entrepreneurs, engineers, computer
programmers can be investigated. By using this research tool, multidisciplinary studies and
projects can be developed, which can be useful to the academic and business environment.
In addition, this scale can be useful for companies that want to develop different marketing
strategies using artificial intelligence technologies. Risks that may arise and how these risks
are perceived by consumers are important for the implementation of correct strategies.
References
Altan, S., 2019. Two pricing AIs collaborated and overcharged people, [online] Available
at: <https://pazarlamasyon.com/iki-fiyatlandirma-yapay-zekasi-birlik-olup-insanlari-
kazikladi/.> [Accessed 25 January 2020].
Antonio, V., 2017. Nike Goes AI with Nikeid.com, [online] Available at:
<http://www.sellingergroup.com/ai-in-sales/nike-goes-ai-with-nikeidcom.> [Accessed
10 December 2019].
Ayvaz, T., 2014. Marshall Augmented Reality Application: Visualizer, [online] Available at
<http://www.dijitalajanslar.com/marshall-artirilmis-gerceklik-uygulamasi-visualizer/>
[Accessed 26 March 2020].
BBC, 2015. Microsoft’s Bill Gates insists AI is a threat, [online] Available at:
<https://www.bbc.com/news/31047780> [Accessed 18 November 2019].
BBC, 2018. What are the dangers that artificial intelligence can bring in 10 years? [online]
Available at: <https://www.bbc.com/turkce/haberler-43144059> [Accessed 15
December 2019].
Bostrom, N., 2014. Superintelligence: Paths, Dangers, Strategies. Oxford: Oxford
University Press.
Bowling, M., Burch, N., Johanson, M. and Tammelin, O., 2015. Heads-up Limit Hold’em
Poker is Solved. Science, 347(6218), pp.145-149. https://doi.org/10.1126/
science.1259433.
Brockman, J., 2015, What to Think About Machines That Think: Today’s Leading Thinkers
on the Age of Machine Intelligence. New York: Harper Perennial.
Bryman, A. and Cramer, D., 2001. Quantative Data Analysis with SPSS Release 10 For
Windows. London: Routledge.
AE Consumers’ Perception of Risk Towards Artificial Intelligence Technologies Used in Trade: A Scale Development Study
84 Amfiteatru Economic
Büyüköztürk, Ş., 2017. Data Analysis Handbook For Social Sciences (23rd Edition).
Ankara: Pegem Academic Publishing.
CNNTURK., 2018. Amazon’s robot damaged 24 employees, [online] Available at:
<https://www.cnnturk.com/dunya/amazonun-robotu-24-calisani-hastanelik-etti.>
[Accessed 20 December 2019].
Cohen, L., Manion L. and Morrison, K., 2007. Research Methods in Education. New York:
Routledge.
Costello, A.B. and Osborne, J.W., 2005. Best Practices in Exploratory Factor Analysis:
Four Recommendations for Getting the Most from Your Analysis. Practical Assessment
Research & Evaluation, 10(7), pp.1-9.
Dedeoğlu, G., 2006. Ethical Issues in the Information Society, II. Applied Ethics Congress,
18-20 October, METU, Ankara.
DeVellis, R.F., 2017. Scale Development (Trans. Tarık Totan). Ankara: Nobel Publishing.
Dirican, C., 2015. The Effects of Technological Development and Artificial Intelligence
Studies on Marketing. Journal of Management Marketing and Logistics, 2(3),
pp. 178-190. https://doi.org/10.17261/Pressacademia.2015312948.
Econsultancy, 2018. Dream vs. Reality: The State of Consumer-First and Omnichannel
Marketing. Market Data, London: Econsultancy.com Ltd.
Euronews, 2018. Transkribus: Historical manuscripts digitalize thanks to artificial
intelligence, [online] Available at: <https://tr.euronews.com/2018/11/03/yapay-zeka-
sayesinde-el-yazisini-okuyabilen bilgisayarlar> [Accessed 17 April 2020].
Field, A. and Hole, G., 2019. How to Design and Report a Research. (Trans. Arif Ozer),
Ankara: Ani Publishing.
Geuens M. and De Pelsmacker P., 2002. Validity and reliability of scores on the reduced
Emotional Intensity Scale. Educational and Psychological Measurement. [e-jourmal] 62
(2), pp: 299 -315. https://doi.org/10.1177/0013164402062002007.
Gürbüz, S. and Şahin, F., 2018. Research Management in Social Sciences. Philosophy-
Method-Analysis. Ankara: Seçkin Publishing.
Harrington, D., 2009. Assessing Confirmatory Factor Analysis Model Fit and Model
Revision. Confirmatory Factor Analysis (1st Edition). New York: Oxford University.
Henkoğlu, T., 2019. Risk Assessment for the Use of Artificial Intelligence Systems in
Information Storage Processes. Archive World, 6(2), pp.134-147.
Ho, R., 2006. Handbook of Univariate and Multivariate Data Analysis and Interpretation
with SPSS. Florida: Chapman & Hall/CRC.
Jones, R.C., 2014. Hawking: Artificial intelligence could bring about the end of humanity,
[online] Available at: <https://www.bbc.com/turkce/haberler/2014/12/141202_
hawking_yapay_zeka> [Accessed 19 November 2019].
Kietmann, J., Paschien J. and Trieen E., 2018. Artificial Intelligence in Advertising How
Marketers Can Leverage Artificial Intelligence Along the Consumer Journey.
Advertising Research Journey, 58(3), pp.263-267. https://doi.org/10.2501/jar-2018-035.
Kline, R.B., 2005. Principles and Practice of Structural Equation Modeling (2nd Edition).
New York: The Guilford Press.
Artificial intelligence in wholesale and retail AE
Vol. 23 • No. 56 • February 2021 85
Knight, W., 2017. The Dark Secret at the Heart of AI, [online] Available at
<https://www.technologyreview.com/s/604087/the-dark-secret-at-the-heart-of-ai/>
[Accessed 10 October 2019].
Köse, U., 2018a. Artificial Intelligence: Paradoxes in Future Science. Popular Science,
25(261), pp.12-21.
Köse, U., 2018b. Artificial intelligence and the future: Should we worry? Science and
Utopia, 284(24), pp.39-44.
Kulabaş, A., 2017. Smart Hologram Technology, [online] Available at:
<https://medium.com/@AyseKulabas/akıllı-hologram-teknolojisi59faa8915f27>
[Accessed 12 September 2019].
Leech, N.L., Barrett, K.C. and Morgan, G.A., 2005. SPSS for Intermediate Statistics: Use
and Interpretation (2nd Ed). Mahwah, NJ: Lawrence Erlbaum Associates.
Levi Strauss & Co., 2017. Levi’s launches new ‘virtual stylist’ online feature, [online]
Available at: <https://www.levistrauss.com/unzipped-blog/2017/08/31/levis-launches-
new-virtual-stylist-online-feature/>[Accessed 18 September 2019].
Morris, D.Z., 2017. Nearly Half of All Retail Jobs Could be Lost to Automation within 10
Years, [online] Available at: <http://fortune.com/2017/05/21/automation-retail-job-
losses/> [Accessed 20 April 2020].
O’Shea, D., 2017. West Elm taps Pinterest images for product recommendations, [online]
Available at: <https://www.retaildive.com/news/west-elm-taps-pinterest-images-for-
product-recommendations/446563/> [Accessed 20 December 2019].
Pallant, J., 2016. SPSS User Guide – Step-by-Step Data Analysis with SPSS. (Trans. S.
Balcı & B. Ahi). Ankara: Ani Publishing.
Parry, K., Cohen, M. and Bhattacharya, S., 2016. Rise of The Machines: A Critical
Consideration of Automated Leadership Decision Making in Organizations. Group and
Organization Management, 41(5), pp.571-594. https://doi.org/10.1177/
1059601116643442.
PC World, 2011. IBM Watson Vanquishes Human Jeopardy Foes, [online] Available at:
<https://www.pcworld.com/article/219893/ibm_watson_vanquishes_human_jeopardy_
foes.html> [Accessed 19 December 2019].
Pittman, R.J., 2016. Say ‘Hello’ to eBay ShopBot Beta, [online] Available at:
<https://www.ebayinc.com/stories/news/say-hello-to-ebay-shopbot-beta/> [Accessed 18
October 2019].
Russell, S., 2015. Will they make us better people?, [online] Available at:
<http://edge.org/response-detail/26157> [Accessed 03 November 2019].
Seçer, I., 2015. Psychological test development and adaptation process: SPSS and LISREL
Aplications. Ankara: Anı Publishing.
Şentürk, B., 2018. Artificial intelligence (AI) in marketing - How UPS did It?, [online]
Available at: <https://pazarlamaturkiye.com/makale/pazarlamada-yapay-zeka-ai-ups-
bunu-nasil-yapti/> [Accessed 13 November 2019].
Shavel, M., Vanderzeil, S. and Currier, E., 2017. Retail Automation: Stranded Workers?
Opportunities and risks for labor and automation, Global Thematic Research. IRRC
Institute: Cornerstone Capital Group.
AE Consumers’ Perception of Risk Towards Artificial Intelligence Technologies Used in Trade: A Scale Development Study
86 Amfiteatru Economic
Silver, N., 2012, The Signal and the Noise: Why So Many Predictions Fail - But Some
Don’t. New York: The Penguin Press.
Statista, 2020. Revenues from the artificial intelligence (AI) software market worldwide
from 2018 to 2025 (In Billion U.S. Dollars), [online] Available at:
<https://www.statista.com/statistics/607716/worldwide-artificial-intelligence-market-
revenues/> [Accessed 11 June 2020].
Sterne, J., 2017. Artificial Intelligence for Marketing: Practical Applications. New Jersey:
John Wiley & Sons, Inc.
Stevens, J.P., 2002. Applied Multivariate Statistics for The Social Sciences (Fourth
Edition). New Jersey: Lawrance Erlbaum Association.
Streiner, D.L., 1994. Figuring Out Factors: The Use and Misuse of Factor Analysis.
Canadian Journal of Psychiatry, 39(3), pp.135-140. https://doi.org/10.1177/
070674379403900303.
Tabachnick, B.G. and Fidell, L.S., 2013. Using Multivariate Statistics. Boston: Pearson
Education.
The Associated Press, 2017. For driverless cars, a moral dilemma: who lives and who dies?,
[online] Available at: <https://www.nbcnews.com/tech/innovation/driverless-cars-
moral-dilemma-who-lives-who-dies-n708276> [Accessed 12 October 2019].
The Guardian, 2015. Elon Musk donates $10m to keep artificial intelligence good for
humanity, [online] Available at: <https://www.theguardian.com/technology/2015/jan/16/
elon-musk-donates-10m-to-artificial-intelligence-research> [Accessed 12 October 2019].
Thiraviyam, T., 2018. Artificial Intelligence Marketing. International Journal of Recent
Research Aspects, 4(Special Issue), pp.449-452.
Thompson, B., 2004. Exploratory and Confirmatory Factor Analysis: Understanding
Concepts and Applications. Washington DC: American Psychological Association.
Tjepkema, L., 2019. What Is Artificial Intelligence Marketing & Why Is It So Powerful,
[online] Available at: <https://www.emarsys.com/en/resources/blog/artificial-
intelligence-marketing-solutions/ > [Accessed 12 April 2020].
Tüfekçi, Z., 2018. We create a dystopia for people to click on ads, [online] Available at:
<https://www.ted.com/talks/zeynep_tufekci_we_re_building_a_dystopia_just_to_make
_people_click_on_ads/transcript?language=tr> [Accessed 26 March 2020].
Ulutaş Ertuğrul, T., 2019. What were the sectors pushing the limits of artificial intelligence
in 2019?, [online] Available at: <https://pazarlamasyon.com/2019da-yapay-zekanin-
sinirlarini-zorlayan-sektorler-nelerdi/>[Accessed 20 April 2020].
Wakefield, J., 2018. What are the dangers of artificial intelligence in 10 years?, [online]
Available at: <https://www.bbc.com/turkce/haberler-43144059> [Accessed 07
November 2019].
Yamak, M., 2017. The cafe of the future where robots prepare all coffees: Café X, [online]
Available at: <https://www.webtekno.com/tum-kahveleri-robotlarin-hazirladigi-
gelecegin-kafesi-cafe-x-h24913.html> [Accessed 10 December 2019].
Young, J. and Cormier, D., 2014. Can Robots Be Managers, Too? Harvard Business
Review, [online] Available at: <http://blogs.hbr.org/2014/04/can-robots-bemanagers-
too/> [Accessed 12 October 2019].
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 87
ARTIFICIAL INTELLIGENCE IN ELECTRONIC COMMERCE:
BASIC CHATBOTS AND THE CONSUMER JOURNEY
Eliza Nichifor1, Adrian Trifan2 and Elena Mihaela Nechifor3 1)2)3) Transilvania University of Brașov, Romania
Please cite this article as:
Nichifor, E., Trifan, A. and Nechifor, E.M., 2021.
Artificial Intelligence in Electronic Commerce: Basic
Chatbots and the Consumer Journey. Amfiteatru
Economic, 23(56), pp. 87-101.
DOI: 10.24818/EA/2021/56/87
Article History
Received: 30 September 2020
Revised: 10 November 2020
Accepted: 24 December 2020
Abstract
This study aims to empirically cover the impact of the use of artificial intelligence through
chatbots on online retail in terms of content implemented in the communication process. The
presented research brings a contribution to the specialized literature by analyzing the
perceived utility and demonstrating the facility, key concepts of the Technology Acceptance
Model. In this sense, ten online stores in Romania were studied, selected according to the
number of users, the research being carried out through a non-reactive method - content
analysis. The method of data collection was that of the “mysterious client” in order not to
generate a change in the behavior of the entities studied. The interpretation of the data
obtained through the content grid allowed a horizontal and vertical approach that led to a
series of results that confirmed the low level of performance of market leaders, as well as the
high potential of this type of technology applied in the field. Regarding the impact of the use
of chatbots, it has been shown that poor quality of the content displayed to users affects the
consumer's journey, the point of satisfaction not being reached in these conditions.
Keywords: chatbot, artificial intelligence in commerce, customer service, e-commerce,
buying behavior, customer engagement
JEL classification: O30, M31, M10
Corresponding author, Eliza Nichifor – e-mail: [email protected]
AE Artificial Intelligence in Electronic Commerce: Basic Chatbots and Consumer Journey
88 Amfiteatru Economic
Introduction
Creating a positive experience in providing services to consumers has become a key strategy
for gaining competitive advantage (Berry, 1995), and the ever-changing technology and the
complex nature of a new society have changed the volume and diversity of activities
(Bătăgan, Mărăşescu and Pocovnicu, 2010). The development of e-commerce has improved
online activity (Voineagu et al., 2016), and as buyer intelligence is accelerated by the use of
technology (EY, 2020), retailers are trying to keep pace with the pace of evolution to qualify
on the map of consumer perception. In the context of the contemporary market dominated by
a “deep electronic phenomenon” (Varga Apăvăloaie, 2015), its use is increasing, it is on the
side of customers who have become more and more cautious and informed about the products
and services they purchase. Thus, the artificial intelligence technologies used to involve end
customers in the retail value chain appear. They focus on interaction at all stages of the
consumer: pre-purchase, purchase and post-purchase (Rese, Ganster and Baier, 2020),
improved support services and sales support functions (Kaplan and Haenlein, 2019). Given
the relatively new nature of the technologies specific to Artificial Intelligence (hereinafter
AI) integrated into e-commerce, especially chatbots, the studies present pros and cons
regarding their acceptance. Some illustrate meeting user expectations (Chopra, 2019;
McLean and Osei-Frimpong, 2019; Chung et al., 2020), but others also argue against them
(Xueming et al., 2019; Sheehan, Jin and Gottlieb, 2020). Polarization can be considered the
ideal framework for the introduction in the research of the Technology Acceptance Model
(hereinafter TAM), first proposed by Davis (1985). Based on Rational Action Theory (TRA),
TAM presents determinants of behaviors described as consciously intentional (Rese, Ganster,
& Baier, 2020). In other words, a user's behavior is determined by the intention to use
computer systems according to the perceived utility, as well as the perceived ease of use. The
authors of this study consider that the two concepts underlie the use of AI technology and
can be analyzed from the perspective of the behavior of online commerce companies
(hereinafter retailers) in the communication process.
Research shows that chatbots are increasingly being implemented in messaging services and
are considered an integral part of future consumer services. As a result, 80% of businesses
use or are in the process of implementing chatbots to communicate with users 24/7 and solve
their customer problems. These are aspects that present real opportunities, and the approach
of empirical studies on user satisfaction with chatbots and the intention to continue the
interaction in the acquisition process, is becoming increasingly relevant (Ashfaq et al., 2020).
One can notice the lack of large-scale empirical research to reveal the consumer experience
concerning AI technologies (Ameen et al., 2020), which maintains an uncertainty about how
retailers in Romania communicate with users through chatbots. As such, it can be concluded
that it is the main limitation of the research, the authors aiming, in particular, to enrich the
literature by empirically covering the impact of AI technology by using basic chatbots in the
communication between retailers, which carries out the online activity, and users at different
stages of the process of creating a stronger consumer experience.
The controversial nature of this type of AI technology is marked by the acceptance by
consumers of chat technology, which determines that the paper addresses especially the link
between the quality of content transmitted through computer systems and user behavior.
Given the fact that 54% of users expressed their reluctance towards chatbots due to
impersonal communication, this study aims to determine the quality of the communication
process by using AI technology by retailers. Thus, the TAM model is extended, stating that
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 89
the two motivations of users, perceived usefulness and perceived ease of use, which mark the
process of accepting chatbots, are assessed using the four elements addressed by conducting
content analysis: quality, response time, relevance and performance level of basic chatbots.
Specifically, the study involves analyzing how the top ten retailers in Romania, chosen based
on the number of users (trafic.ro), react to initiatives to communicate with the public, through
instant messages from the Facebook Messenger application (hereinafter referred to as basic
chatbots). The number of users is considered by the authors a quantitative criterion, chosen
to select the most accessed online commerce websites in Romania because the action of
visiting this type of website reflects the advantage of using the Internet. Over 67% of the
subjects of a study stated that e-commerce services are accessible, easy to use, but especially
efficient (Bătăgan, Mărăşescu and Pocovnicu, 2010). The results of the analysis led to the
conclusion that the ten existing online stores in the top have a poor quality of the content
communicated to customers and to the fact that there is a development potential regarding
artificial intelligence in e-commerce in Romania.
To achieve this goal, the paper was structured into five sections. Thus, after reviewing the
literature, the research methodology was described, the section is followed by a review of the
results obtained, the penultimate section including the discussions, and, in the final part, the
conclusions and proposals are nominated.
1. Literature review
Given that it is perceived as a relatively new technology, the chatbot application has been
around for a long time, and the terms used have varied over time. This was recognized as an
automatic conversation system, virtual agent, dialogue system, or chatterbot (Ciechanowski et
al., 2019). The first notable appearance in the field is the chatbot ELIZA, developed by Joseph
Weizenbaum in 1956 (Smutny and Schreiberova, 2020), which was built using simple word
matching techniques, to simulate a psychotherapist, succeeding for the first time. given the
interaction between a person and a computer through natural text language (Rese, Ganster and
Baier, 2020). Chatbots are currently defined as software tools that interact with users on a
specific topic in the most natural way possible, and the message is conveyed in the form of text
or voice (Ashfaq et al., 2020; Rese, Ganster and Baier, 2020). To illustrate the applicability of
this technology, it is mentioned that chatbots have been introduced in the Facebook platform
(2016) with the role of accelerating and facilitating customer service processes, managed by
organizations. Since then, they are considered an “important technological trend” (Baier, Rese
and Röglinger, 2018), with natural language skills, which can be configured to “converse” with
users (Sheehan, Jin and Gottlieb, 2020) and provide information about products and services,
or place orders online in real-time (Ashfaq et al., 2020).
Technology is recognized as a smart tactic (Toorajipour et al., 2020), especially from a
commercial point of view. Chatbots can be used in the interaction between retailer and
customer (Lemon and Verhoef, 2016) both in the pre-purchase process (Forrest and Hoanca,
2015) and in the integrated marketing communication strategy, to revolutionize how
interaction with users happens (McLean and Osei-Frimpong, 2019). In both applications,
user behavior is influenced by the experience, determining the decision to continue the
interaction process or not (Mihart (Kailani), 2012). For this reason, reference is made to the
concept of customer engagement. According to the literature, it indicates the connection and
participation of customers or users with retailers (Hollebeek, 2011). This concept takes the
form of a synergy such as bilateral creation between suppliers and users and can be
AE Artificial Intelligence in Electronic Commerce: Basic Chatbots and Consumer Journey
90 Amfiteatru Economic
represented by a component of the marketing strategy to attract new customers, determine
the acquisition of purchases and consumer loyalty (Brodie et al., 2011). The introduction of
technologies that use artificial intelligence (AI) by an organization, generates the potential
for it to involve users while achieving organizational results related to efficiency, satisfaction
and commitment (Prentice and Nguyen, 2020). In the same framework is mentioned the
model of an electronic business, which for carrying out company-specific activities, the
organization has electronic resources such as websites adapted to the device, chats, blogs and
emails (Sitar-Taut et al., 2009). Therefore, with the development of e-commerce and mobile
purchases, retailers are competing for a favorable long-term market position (Souiden,
Ladhari and Chiadmi, 2019), trying to satisfy consumers who want information and real-time
responses (Reinartz, Wiegand and Imschloss, 2019). It is important to mention in this
framework, the estimates that show that globally, the growth rate of e-commerce will increase
from 10% at present, to 50% by 2027 (Platon, 2015). To achieve their goal of adapting to
evolution, retailers have begun to use AI technologies in various ways to gain a competitive
advantage (chatbots, generating relevant content for consumers, etc.). Ameen et al. (2020)
mention in their paper that the implementation of AI solutions in the retail sector can reach a
percentage of 1% of customers, which are 18 times more valuable than the average customer
in the sector, and Mindbrowser (2017) mentions the benefits retailers in the following terms:
customer service (95%), sales and marketing (55%) and order processing (48%) and the fact
that chatbots can supplement pre-purchase support activities with the help of learning
algorithms and predictive modeling. Thus, an instantaneous synchronization of a need
presented by the consumer and the accessible products that match his expectations is
achieved (Forrest and Hoanca, 2015). The applicability of the service has the effect of
reducing the costs related to customer support activities and increasing the speed of response
to user questions (IBM, 2017), but also of transport costs (Bătăgan, Mărăşescu and
Pocovnicu, 2010). Therefore, the duration of receiving a response is one of the significant
factors in obtaining their satisfaction. A customer-centric value chain is created, in which
connections are built with users who become potential customers, and retailers compete to
exceed their expectations to survive in a competitive market (SAP Industries, 2020).
Canhoto and Clear (2020) state that this can improve the efficiency of business processes,
but on the other hand, in some cases, these technologies can also destroy the value of the
business. In this regard, the situations in which end customers are reluctant when it comes to
using chatbots are highlighted. Mention is made of the case of the United States of America,
which has a 40% share of consumers who prefer to be approached by a real person
(Mindbrowser, 2017; CGS, 2018), and the results of a global sample show that only 34% of
subjects they feel comfortable to proactively receive personalized recommendations in the
pre-purchase phase via a chatbot (Rese, Ganster and Baier, 2020). A higher percentage is
represented by every second online shopper (52%), who expressed antipathy towards
chatbots due to an impersonal approach, an immature technology, the lack of a recognized
benefit, or the feeling of following through this tool (Smutny and Schreiberova, 2020). Those
with a much higher interest in chatbots and their use are representatives of Generation Z and
Millennials: at the level of a global sample aged 18 to 34, 25% of subjects choose
personalized shopping through a chatbot (Chatbots Magazine, 2018).
Favorable results are obtained by traders' attention to the quality delivered, i.e., “extreme
personalization” and “increased involvement based on contextual and behavioral data” (Forbes,
2017). The quality of this type of service is defined by Parasuraman, Zeithaml and Berry (1994),
as the difference between the expected and the perceived service, being evaluated concerning
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 91
the way consumers perceive the services offered by the retailer. Thus, artificial intelligence
transforms the way retailers operate (Pillai, Sivathanu and Dwivedi, 2020), and the quality of
implementation of this technology must increase the level of customer commitment to the brand,
to reach the level of satisfaction (Prentice and Nguyen, 2020). A chatbot demonstrates a
“reliable” performance when it is technically competent to provide the requested information in
the form of answers, without encountering problems (Aoki, 2020).
The context described above favors the introduction of a concept regarding the customers’
journey along the marketing funnel, which is marked by the four stages: awareness,
consideration, purchase intention and satisfaction (Colicev, Kumar and O’Connor, 2019).
Chatbots, through the forces of AI technology act in the middle area of the marketing funnel,
in the consideration stage, so that, in the interaction of users with instant messages set by
retailers, they show a behavior close to the desire to purchase. At this point, reference is made
to the two motivations inherent in the TAM model and the extension of the theory by
associating with the qualitative nature of the chatbots content: perceived utility and perceived
ease of use being two concepts influenced by the user's experience about the chatbot
implemented by retailer. If chatbots provide high-quality information, positive effects are
expected, even loyalty to them. Otherwise, consumers do not move to the next stage and do not
reach the desired level of satisfaction. The completion of the process of going through all the
stages of the funnel that the user goes through may depend on the experience with the chatbot.
Companies recognize the high potential of chatbots, which are becoming increasingly used
in messaging services being described by Koumaras et al. (2018), as an integral part of the
future in customer service. It is mentioned that access to technology accelerates consumer
intelligence with a speed that would be impossible to conceive for our ancestors (Young,
2014), therefore, consumers' buying behavior in the online environment presents real
challenges for companies. The exploratory marketing research conducted by Stoica, Vegheş
and Orzan (2015), shows the percentage distribution of Romania according to the consumer
behavior that divides the researched population into digital informants (62%), digital buyers
(26%) and hyperstackers (12%). According to the descriptions of the three categories, their
common denominator is the high degree of consumer involvement, how they have become
increasingly cautious over time regarding the purchases they make, and the quantity and
quality. the information they need to reach the final stage of the funnel plays a particularly
important role.
Murtarelli, Gregory and Romenti (2020), mention that future research could empirically test
aspects from the conversational perspective, which the authors of this paper have done with
high interest. As chatbots have recently been widely used, their acceptance is beginning to
be increasingly researched (Rietz, Benke and Maedche, 2019) due to the high importance of
accepting new technology. According to Fernandes and Oliveira (2020), this is one of the
essential steps to success in the future.
2. Research methodology
The behavioral bipolarity mentioned in the literature, regarding the use of chatbots, favored
the identification of the gap of the research. This refers to how retailers interact with potential
customers and are represented by the analysis of the quality of the content of the messages in
the interaction initiated by users. In this study, this fact was approached scientifically, by
analyzing text chats, integrated with the instant messaging application, Facebook Messenger.
AE Artificial Intelligence in Electronic Commerce: Basic Chatbots and Consumer Journey
92 Amfiteatru Economic
This application was chosen because it offers a myriad of B2C and B2B links in the sale-
purchase process (Xu et al., 2020).
Given that the objective of the empirical study was to analyze the quality of AI solutions
implemented by retailers, it was intended to maintain the behavior regarding the positioning
of online stores to this type of technology. In order not to influence to some extent the results,
it was considered that for the studied context, it is necessary to apply the non-reactive
research method - content analysis. According to Krippendorff (2004), this type of analysis
is a discrete technique, applied when researchers consider avoiding reactive situations for
two reasons: distorting the data, jeopardizing the validity of the study and manipulating the
data by the sources or subjects analyzed. By approaching from this perspective, the study
was successful: no reaction was generated from online stores and there were no changes in
the settings of instant messaging in Facebook Messenger, which led to greater veracity of the
data obtained and accuracy of results.
To delimit the existing phenomenon, information was gathered through structured and direct
observation, with the help of which ten retailers were carried out from the position of a
“mysterious customer”, carrying out their retailing activity in the Romanian online market.
This method was approached to identify a new solution for profile websites, embodied in the
opportunity to perform a pre-test of a survey before launch, in case of starting a subsequent
quantitative study (Columbia Public Health, 2020). The ten retailers were selected based on
the public information available on www.trafic.ro, the data provider with the help of which
the secondary data related to the ranking of online stores were extracted. The rest of the
information that contributed to this study was collected using the Facebook Messenger
application and the Facebook platform (for analysis of response time).
The content analysis focused on Romanian retailers that are performing in terms of the
number of users. The choice of the Romanian market for conducting the study was based on
the development of electronic commerce and the potential for multiplying the transactions
made through electronic commerce developed according to the Western model (Bătăgan,
Mărăşescu and Pocovnicu, 2010). According to the Romanian Association of Online Stores
in Romania, the e-commerce sector showed an increase of 20-22% in 2019, compared to
2018, and the average shopping cart value increased from 204 lei in 2018 to 273 lei in 2019,
for desktop purchases and an increase from 170 lei in 2018 to 208 lei in 2019, for purchases
made with the help of smartphones (GPec, 2020), which must be mentioned considering the
impact they have on the Gross Domestic Product of the country (Pantelimon, Georgescu and
Posedaru, 2020). Estimates from the iSense Solution study for the GPec 2020 Summit show
that the e-commerce market is reaching six billion euros and that in the context of the
COVID-19 pandemic, the number of people who shop online has increased by 13% compared
to 2019. According to Romania Insider (2020), the average value of online shopping
increased by 40%.
The selection of websites was made considering the number of users who visited online stores
in August 2020. The choice of criteria is justified by the fact that, in Romania, out of 19.94
million people (INS, 2014), 70% are Internet users (Ecommerce News, 2016). Given this,
the authors considered the variable to be a relevant criterion for illustrating the scientific
research framework. With the analysis grid, the results obtained provide an overview of the
use of basic chatbots, their quality and performance, response time, and the relevance of
responses to consumers. The coding of the data (Table no. 1) was performed so that there
was a homogeneous analysis of the results.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 93
Table no. 1. Coding of answers in content analysis
Topic Coding
Using Chatbots
1. Don't use any chatbots
2. Instant replies
3. Chatbot in Facebook Messenger
(Full Set Version)
Response time
1. Not publicly visible
2. Very prompt
3. In a few hours
4. Within a day
5. More than a day
Instant Message Quality
- Information on personalized
recommendations
1. Yes
2. No
- Product cost information 1. Yes
2. No
- Information on personalized assistance 1. Yes
2. No
- Information on popular personalized
products
1. Yes
2. No
- Information on the availability
of products in stock 1. Yes
2. No
- Using a custom addressing formula
1. First name only
2. Last name only
3. First name and last name
4. No first name, no last name
Relevance of replies
1. High
2. Medium
3. Low
Performance 1. Basic sending of automatic messages
2. Proactively recommend content
The data obtained were interpreted based on the results inserted in the horizontal analysis
grid (in which the number of occurrences of the codes was determined) and the vertical
analysis.
3. Results and discussions
The results of the research led to the main actions that retailers could include in their strategy
to become electronic commerce businesses. To understand their relevance, importance and
efficiency, the authors present details below regarding content analysis results on the use,
quality and performance of chatbots, the response time and relevance replies provided to
users. The section is followed by related discussions, namely the integration of research
conducted in the literature and aims to provide an overview of the actual context and future
perspectives for the entities studied.
AE Artificial Intelligence in Electronic Commerce: Basic Chatbots and Consumer Journey
94 Amfiteatru Economic
3.1. Content analysis results
The research results provide an image of the support services in the pre-purchase phase from
the consideration stage of the consumer's journey from the perspective of the marketing
funnel, in which the user intends to interact with the provider through Facebook Messenger.
The horizontal and vertical analysis approach highlights the argument that 100% of the top
retailers (Table no. 2) do not use any full version of a chatbot.
Table no. 2: The Rank of Online Stores in Romania (August)
Place Website Users Visits Views
1 www.pretzmic.ro 51,904 72,308 269,250
2 www.baterii-lux.ro 43,128 59,857 158,099
3 www.uscatorrufe.ro 25,992 31,429 67,343
4 www.wainertools.ro 25,958 31,763 69,993
5 www.magazin-unelte.ro 19,658 24,002 90,991
6 www.vintagetime.ro 19,408 23,954 64,382
7 www.parfumas.ro 12,367 25,350 248,482
8 www.fabricadelenjerii.ro 11,874 18,720 76,668
9 www.lexservice.ro 11,774 15,343 86,643
10 www.gamestore.ro 11,259 16,279 65,688
Source: Trafic, 2020
The quality of the content was studied from the perspective of the conversations initiated to
determine the impact that the content of chatbots has on the buying behavior of final
consumers. To present the results as concisely as possible, the analysis of each category
integrated in the content analysis was used, as follows:
Using chatbots. Although no online store among the most visited according to the
number of users (Figure no. 1) has implemented a full version of a chatbot, the scenario is
outlined by the existence of the basic form of a chatbot in all ten cases studied. This form is
represented by the automatic messages that retailers have set up and through which the
interest is exercised to initiate conversations in the online environment with users using this
form of AI technology.
Response time. Regarding this topic, there are significant differences concerning the
performance of the studied online stores. To have a homogeneous coding, the following
options regarding the response time have been chosen, which users can consult on the public
page of online stores on the Facebook platform: “not visible”, “respond very promptly”,
“respond in a few hours”, “responds within a day or more”. The analysis reveals that the
website with the highest number of users (ranked first in the top) responds in a longer time
frame (one day) than the websites ranked seventh, eighth and nine in the ranking, which has
a response time of only a few hours. The most common situation is when for five of the ten
online stores, the response rate is not visible to users; this is also the case of the second-placed
retailer. Although it ranks first in terms of the number of users, the website www.pretzmic.ro
shows a response in a long time for consumer satisfaction in the purchase process (one day).
The quality of chatbots was analyzed from the perspective of information provided to
users regarding the availability of products in stock, the cost of personalized products,
personalized popular products, personalized recommendations, personalized assistance and
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 95
information on the use of the first name. In this sense, it emerged that only in one of the cases,
was a chance for the user to find out information about the availability of products in stock,
represented by the retailer www.fabricadelenjerii.ro, located on the eighth place. IT
deficiencies were also encountered in the case of product cost details (aspect set only in six
out of ten situations), in the case of “trendy” products (six entities displayed the option), and
in the case of information on personalized recommendations (where four entities did not post
questions to start a conversation (including the brand in the first place.) The most common
option among the ten entities studied is personalized assistance, set by eight of them, which
is an argument for the fact that the providers know the importance of personalized assistance
functions and want to implement the service. Another evaluation of the quality of automatic
messages was made from the perspective of using the first name in the answers provided by
retailers. The situation where only the first name was used was not encountered in the ten
cases, and in four of them, it was not used any form of custom addressing. The situation of
the brand in the first place is noticed, in which case there is a personalized setting for
displaying the first name and last name in the automatic answers, but this is completely
missing in the case of the other nine situations. Therefore, the lack of attention from
companies strengthens a barrier in the retailer-end consumer relationship and has a negative
impact during the consumer to the time of purchase, due to the impersonal approach.
The relevance of the replies provided automatically after the selection was coded as
low or non-existent. From this perspective, the eighth-ranked retailer was the only one that
reacted after the user's interaction through instant messaging, in less than an hour, with a
response from a person responsible for the customer relationship. In five cases, the replies
were received instantly, with low relevance, and in four cases, the chatbot was not set to
answer automatically after choosing a standard question.
The performance was coded in the content analysis through two variants: “sending
basic messages” and “recommending content”. In no case studied out of the ten, the situation
in which content is recommended to the user was not encountered. All the top retailers sent
basic messages, which determines the assignment of basic performance by using the related
“basic chatbots”.
Figure no. 1. Ranking of Online Stores in Romania according to the number of users
(August 2020)
Source: Trafic, 2020
AE Artificial Intelligence in Electronic Commerce: Basic Chatbots and Consumer Journey
96 Amfiteatru Economic
3.2. Discussions
Obtaining competitive advantages is a process that can be easily achieved in the context in
which the concepts presented in this paper are known. Through their analysis and the
extension of the TAM model, the development of e-commerce can continue to improve the
online activity referred to by Voineagu et al. (2016) in his paper. The research confirms the
presence of a “deep electronic phenomenon” (Varga Apăvăloaie, 2015) in the customer-
supplier relationship and conceptually develops the approach for the development of
communication through artificial intelligence. Considering that technology is a smart tactic
(Toorajipour et al., 2020), it is highlighted that chatbots and their acceptance strongly
influence retailers' performance, user satisfaction and commitment. The originality of the
paper is marked by the analysis of the quality of the content of the message transmitted in the
communication process with the help of chatbots. It is believed that the approach may
influence the reduction of the degree of polarization that current studies highlight (Chopra,
2019; McLean and Osei-Frimpong, 2019; Xueming Luo et al., 2019; Chung et al., 2020;
Sheehan, Jin and Gottlieb, 2020) and that through the results obtained, at least the retailers
included in the study, can improve the relationship developed in the digital environment they
have with users. The mention made by Murtarelli, Gregory and Romenti (2020) regarding
the future research related to the empirical testing of the conversational perspective, is in
perfect synchrony with the purpose of the present study. The authors are concerned about the
quality of interaction between retailers and users, expressed through issues related to the use
and quality of chatbots, response time, relevance of responses, and performance that can be
used to develop sales support chatbots implemented in marketing, sales, technical support or
customer service activities, using the Facebook and Facebook Messenger platform. In
essence, this is the contribution to the enrichment of the literature.
Consumers become better informed when they seek to make a purchase, being influenced by
their experience (Mihart (Kailani), 2012). The study confirms that retailers must meet the
requirement of users to communicate efficiently and in real-time. To benefit from the use of AI
technology to obtain the development opportunities outlined above (Reinartz, Wiegand and
Imschloss, 2019; Souiden, Ladhari and Chiadmi, 2019; Prentice and Nguyen, 2020), as well as
to become an efficient online business (Sitar -Taut et al., 2009), an organization has the
opportunity to consider the results of the study and to implement a series of measures to
eliminate specific barriers (de Bellis and Venkataramani Johar, 2020). Therefore, instant
messages in the Facebook Messenger application must take the form of interaction as
personalized as possible, by using the first name as a form of address at the beginning of the
message sent. This action has a positive impact on the population who have expressed their
antipathy towards this technology, due to the impersonal approach. Another measure that
companies can implement is to increase the quality of instant messages sent to users who have
shown a desire to communicate. To this end, the standard set of questions should include
information on personalized recommendations, the cost of products, their availability in stock,
information on popular products and personalized assistance. Also, organizations that want to
develop their relationships with consumers must make the settings of Facebook pages so that
the response time is publicly visible. In this way, users who want to initiate a dialogue can
quickly recognize the performance of the provider in this regard. A short response time
encourages users to communicate with the provider, knowing that they will receive a response
as soon as possible. Therefore, the link between the quality of chatbots, the information
provided in customer support and sales support services, but also the response time influences
the consumer's behavior, in the sense that, in case of satisfaction, he decides to continue the
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 97
purchase process. Otherwise, if the user stops in the consideration phase, there is a risk that the
brand promoted by the retailer will be removed from the map of consumer perception.
Conclusions
The article contributes to the specialized literature through the detailed information obtained
regarding the activity of retailers practicing retail in the online environment in Romania. By
addressing the Technology Acceptance Model (Davis, 1985), the paper presents a new
perspective by extending the two inherent elements, perceived utility and perceived ease of
use, by applying measures to achieve a high degree of interaction in the process.
communication via chatbots. The poor quality of the content displayed through the basic
chatbots generates consumer dissatisfaction, especially in the pre-purchase stage, when users
- potential customers - are interested in communicating in real-time with retailers. The article
warns that chatbots are in an early stage of development (Smutny and Schreiberova, 2020),
with a lack of personalization of messages, inappropriate timing of responses and less
relevant texts to request customers in different stages related to the marketing funnel.
Following the results, the authors consider that although implemented at a basic level,
artificial intelligence can help improve personalized support functions, if chatbots are
technically competent to continue the conversation at the request of users and if the response
time is publicly visible on retailers' Facebook pages, which increases users' commitment to
them. The academic environment can play a particularly important role by encouraging
entrepreneurial education for students, from the perspective of using new technologies for a
competent and sustainable entrepreneurial ecosystem of the future.
It can be concluded that the managerial implications are represented by the development
actions that retailers must initiate within the organization. The authors recommend to the
organizations from the economic environment the monitoring of the stages in which the target
public is and the analysis of the level of acceptance of the use of the new technologies within
the course undertaken by the potential clients. From this point of view, the general conclusion
of the paper is outlined around the scenario in which online stores in Romania must apply AI
technology solutions through basic chatbots, to deliver information on the following issues:
availability and cost of products in stock, products personalized, personalized assistance and
popular products, transmitted by approaching a form of personalized addressing, by using
the first name.
The main limitation of the research is the analysis of retailers in terms of the number of users,
which led to the inclusion in the analysis of a small number of subjects, depending on the
classification made by the data provider. It can be deduced that certain retailers operating
exclusively online were not included in the study, as they are not part of the top mentioned.
This is also the case of online stores such as Emag.ro, Elefant.ro, Fashiondays.ro, Evomag.ro
or Epiesa.ro, which do not appear to be the most visited according to trafic.ro, but have a
significant dimension (PinBud, 2020). Another limitation of the research is that the study
shows the approach of the perspective of the “mysterious client” and not consumers
belonging to the public. Future research can be developed starting from considering the
nominated limits, first, in the form of a quantitative study, to analyze on a national scale the
opinion and attitude of consumers towards AI technology. Second, the results of the research
can be capitalized by using them to start a survey, based on the variables mentioned in the
paper, and finally, the research can be extended by taking into account the views of market
leaders, which present a figure of significant business for the industry.
AE Artificial Intelligence in Electronic Commerce: Basic Chatbots and Consumer Journey
98 Amfiteatru Economic
References
Ameen, N., Tarhini, A., Reppel, A. and Anand, A., 2020. Customer experiences in the age
of artificial intelligence. Computers in Human Behavior, [e-journal] 114, 106548.
https://doi.org/10.1016/j.chb.2020.106548.
Aoki, N., 2020. An experimental study of public trust in AI chatbots in the public sector.
Government Information Quarterly, [e-journal] 37(4), 101490. https://doi.org/10.1016/
j.giq.2020.101490.
Ashfaq, M., Yun, J., Yu, S. and Loureiro, S., 2020. I, Chatbot: Modeling the determinants of
users’ satisfaction and continuance intention of AI-powered service agents. Telematics
and Informatics, [e-journal] 54, 101473. https://doi.org/10.1016/j.tele.2020.101473.
Baier, D., Rese, A. and Roeglinger, M., 2018. Conversational user interfaces for online
shops? A categorization of use cases. [pdf] Available at: <https://www.fim-
rc.de/Paperbibliothek/Veroeffentlicht/822/wi-822.pdf> [Accessed 4 December 2020].
Bătăgan, L., Mărăşescu, A. and Pocovnicu, A., 2010. Consumer rights in digital economy.
Case study of Romanian e-commerce usage. Theoretical and Applied Economics,
9(9(550)), pp. 79-96.
Berry, L.L., 1995. Relationship marketing of services: growing interest, emerging
perspectives. Journal of the Academy of Marketing Science, 23(1), pp. 236-245.
Brodie, R.J., Hollebeek, L.D., Jurić, B. and Ilić, A., 2011. Customer engagement: Conceptual
domain, fundamental propositions, and implications for research. Journal of Service
Research, 14(3), pp. 252-271.
Canhoto, A.I. and Clear, F., 2020. Artificial intelligence and machine learning as business
tools: A framework for diagnosing value destruction potential. Business Horizons, 63(2),
pp. 183-193.
CGS, 2018. Chatbots deliver speed, but consumers still want humans. Are we moving too
quickly to automation? [pdf] Available at <https://www.cgsinc.com/sites/default/
files/media/resources/pdf/CGS_Consumer%2BCustServ%2Binfographic%2B2018.pdf>
[Accessed 30 November 2020].
Chatbots Magazine, 2019. Chatbot report 2018: Global trends and analysis. [online]
Available at: <https://chatbotsmagazine.com/chatbot-report-2019-global-trends-and-
analysis-a487afec05b> [Accessed 30 November 2020].
Chopra, K., 2019. Indian shopper motivation to use artificial intelligence. International
Journal of Retail & Distribution Management, 47(3), pp. 331-347.
Chung, M., Ko, E., Joung, H. and Kim, S.J., 2020. Chatbot e-service and customer
satisfaction regarding luxury brands. Journal of Business Research, 117(1), pp. 587-595.
Ciechanowski, L., Przegalinska, A., Magnuski, M. and Gloor, P., 2019. In the shades of the
uncanny valley: An experimental study of human–chatbot interaction. Future Generation
Computer Systems, 92(1), pp. 539-548.
Colicev, A., Kumar, A. and O’Connor, P., 2019. Modeling the relationship between firm and
user-generated content and the stages of the marketing funnel. International Journal of
Research in Marketing, 36(1), pp. 100-116.
Columbia Public Health, 2020. Content analysis method and examples. [online] Available
at: <https://www.publichealth.columbia.edu/research/population-health-methods/content-
analysis> [Accessed 28 November 2020].
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 99
Davis, F.D., 1985. A technology acceptance model for empirically testing new end-user
information systems: Theory and results. Ph.D. Massachusetts Institute of Technology.
de Bellis, E. and Venkataramani Johar, G., 2020. Autonomous shopping systems: identifying
and overcoming barriers to consumer adoption. Journal of Retailing, 96(1), pp. 74-87.
Ecommerce News, 2016. Ecommerce in Romania. [online] Available at:
<https://ecommercenews.eu/ecommerce-in-europe/ecommerce-romania/> [Accessed 30
November 2020].
EY, 2020. How to accelerate online direct to consumer strategies beyond COVID-19. [pdf]
Available at: <https://www.ey.com/en_gl/consumer-products-retail/how-to-accelerate-
online-direct-to-consumer-strategies-beyond-covid-19> [Accessed 30 November 2020].
Fernandes, T. and Oliveira, E., 2020. Understanding consumers’ acceptance of automated
technologies in service encounters: Drivers of digital voice assistants adoption. Journal
of Business Research, 122(1), pp. 180-191.
Forbes, 2017. Extreme personalization is the new personalization: How to use AI to
personalize consumer engagement. [online] Available at: <https://www.forbes.com/
sites/briansolis/2017/11/30/extreme-personalization-is-the-new-personalization-how-to-
use-ai-to-personalize-consumer-engagement/> [Accessed 30 November 2020].
Forrest, E. and Hoanca, B., 2015. Trends and Innovations in Marketing Information Systems.
Greece: Alexander Technological Educational Institute of Thessaloniki.
GPec, 2020. Raport GPeC E-Commerce România 2019: Cumpărături online de peste 4,3
miliarde de euro, în creștere cu 20% față de 2018. [online] Available at:
<https://www.gpec.ro/blog/raport-gpec-e-commerce-romania-2019> [Accessed 30
November 2020].
Hollebeek, L.D., 2011. Demystifying customer brand engagement: Exploring the loyalty
nexus. Journal of Marketing Management, 27(7-8), pp. 785-807.
IBM, 2017. How chatbots can help reduce customer service costs by 30%. [online] Available
at: <https://www.ibm.com/blogs/watson/2017/10/how-chatbots-reduce-customer-service-
costs-by-30-percent/> [Accessed 30 November 2020].
INS, 2014. Comunicat de presă nr. 155 din 30 iunie 2014: Populația rezidentă. [pdf]
Available at: <https://insse.ro/cms/ro/content/popula%C5%A3ia-rezident%C4%83-la-1-
ianuarie-2019-%C5%9Fi-migra%C5%A3ia-interna%C5%A3ional%C4%83-
%C3%AEn-anul-2018> [Accessed 20 November 2020].
Kaplan, A. and Haenlein, M., 2019. Siri, Siri, in my hand: Who’s the fairest in the land? On
the interpretations, illustrations, and implications of artificial intelligence. Business
Horizons, 62(1), pp. 15-25.
Koumaras, V., Foteas, A., Fotaes, A., Kapari, M., Sakkas, C. and Koumaras, H., 2018. 5G
Performance Testing of Mobile Chatbot Applications. In: IEEE, 23rd International
Workshop on Computer-Aided Modeling and Design of Communication Links and
Networks (CAMAD). Barcelona, Spain, 17-19 September 2018. Spain: Institute of
Electrical and Electronics Engineer.
Krippendorff, K., 2004. Content Analysis: An Introduction to Its Methodology. 2nd ed.
London: SAGE.
Lemon, K.N. and Verhoef, P.C., 2016. Understanding customer experience throughout the
customer journey. Journal of Marketing, 80(6), pp. 69-96.
AE Artificial Intelligence in Electronic Commerce: Basic Chatbots and Consumer Journey
100 Amfiteatru Economic
McLean, G. and Osei-Frimpong, K., 2019. Chat now… Examining the variables influencing the
use of online live chat. Technological Forecasting and Social Change, 146(1), pp. 55-67.
Mihart (Kailani), C., 2012. Modeling the influence of integrated marketing Communication
on consumer behaviour: an approach based on the hierarchy of effects concept. Procedia
- Social and Behavioral Sciences, 62(1), pp. 975-980.
Mindbrowser, 2017. 5 learnings from our 'Chatbot Survey-2017'. Chatbots Journal, [online]
Available at <https://chatbotsjournal.com/5-learnings-from-our-chatbot-survey-2017-
72a6a4fc209c> [Accessed 21 November 2020].
Murtarelli, G., Gregory, A. and Romenti, S., 2020. A conversation-based perspective for
shaping ethical human-machine interactions: The particular challenge of chatbots.
Journal of Business Research, [e-journal] https://doi.org/10.1016/j.jbusres.2020.09.018.
Pantelimon, F.-V., Georgescu, T.-M. and Posedaru, B.-Ş., 2020. The Impact of Mobile
e-Commerce on GDP: A Comparative Analysis between Romania and Germany and how
Covid-19 Influences the e-Commerce Activity Worldwide. Informatica Economica,
24(2), pp. 27-41.
Parasuraman, A., Zeithaml, V.A. and Berry, L.L., 1994. Reassessment of Expectations as a
Comparison Standard in Measuring Service Quality: Implications for Further Research.
Journal of Marketing, 58(1), pp. 111-124.
Pillai, R., Sivathanu, B. and Dwivedi, Y.K., 2020. Shopping intention at AI-powered
automated retail stores (AIPARS). Journal of Retailing and Consumer Services, [e-
journal] https://doi.org/10.1016/j.jretconser.2020.102207.
PinBud, 2020. Statistici comerț online România 2017. [online] Available at:
<https://www.ksd.ro/cat-e-de-profitabil-comertul-online-din-romania/> [Accessed 1
December 2020].
Platon, O.-E., 2015. The evolution of e-commerce in Romania. Procedia Economics and
Finance, 23(1), pp. 1446-1450.
Prentice, C. and Nguyen, M., 2020. Engaging and retaining customers with AI and employee
service. Journal of Retailing and Consumer Services, [e-journal] https://doi.org/
10.1016/j.jretconser.2020.102186.
Reinartz, W., Wiegand, N. and Imschloss, M., 2019. The impact of digital transformation on
the retailing value chain. International Journal of Research in Marketing, [e-journal]
36(3), pp. 350-366. https://doi.org/10.1016/j.ijresmar.2018.12.002.
Rese, A., Ganster, L. and Baier, D., 2020. Chatbots in retailers’ customer communication:
How to measure their acceptance? Journal of Retailing and Consumer Services,
[e-journal] https://doi.org/10.1016/j.jretconser.2020.102176.
Rietz, T., Benke, I. and Maedche, A., 2019. The impact of anthropomorphic and functional
chatbot design features in enterprise collaboration systems on user acceptance. [pdf]
Available at: <https://chatbotresearch.com/wp-content/uploads/2019/02/wi2019_2.pdf>
[Accessed 4 December 2020].
Romania Insider, 2020. Study: Romania’s e-commerce market up 40% to EUR 6 bln in 2020.
[online] Available at: <https://www.romania-insider.com/ro-ecommerce-isense-nov-
2020> [Accessed 28 November 2020].
SAP Industries, 2020. The intelligent enterprise for the automotive industry. [online]
Available at: <https://www.sap.com/documents/2016/03/2e5a77d0-627c-0010-82c7-
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 101
eda71af511fa.html> [Accessed 30 November 2020].
Sheehan, B., Jin, H.S. and Gottlieb, U., 2020. Customer service chatbots: Anthropomorphism
and adoption. Journal of Business Research, 115(1), pp. 14-24.
Sitar-Taut, D.A., Stanca, L.M., Buchmann, R. and Lacurezeanu, R., 2009. A case study on
usability metrics applied in Romanian E-commerce environment. WSEAS Transactions
on Information Science and Applications, 6(10), pp. 1697-1706.
Smutny, P. and Schreiberova, P., 2020. Chatbots for learning: A review of educational
chatbots for Facebook Messenger. Computers and Education, [e-journal]
https://doi.org/10.1016/j.compedu.2020.103862.
Souiden, N., Ladhari, R. and Chiadmi, N.E., 2019. New trends in retailing and services.
Journal of Retailing and Consumer Services, 50(1), pp. 286-288.
Stoica, I., Vegheş, C. and Orzan, M., 2015. Statistical exploratory marketing research on
Romanian consumer’s behavior regarding smartphones. Procedia Economics and Finance,
[e-journal] 32(1), pp. 923-931. https://doi.org/10.1016/S2212-5671(15)01549-X.
Toorajipour, R., Sohrabpour, V., Nazarpour, A., Oghazi, P. and Fischl, M., 2020. Artificial
intelligence in supply chain management: A systematic literature review. Journal of
Business Research, 122(1), pp. 502-517.
Trafic, 2020. Top siteuri general. [online] Available at: <www.trafic.ro> [Accessed 26
September 2020].
Varga Apăvăloaie, E., 2015. France regarding e-commerce. Acta Electrotehnica, 56(1-2),
pp. 83-89.
Voineagu, V., Vasilache, S.N., Şerban, D., Cristache, S.E. and Begu, L.S., 2016. An analysis
of the Romanian e-commerce trade trends in European perspective, economic
computation and economic cybernetics studies and research. Faculty of Economic
Cybernetics, Statistics and Informatics, 50(1), pp. 235-252.
Xu, Y., Shieh, C.H., van Esch, P. and Ling, I.L., 2020. AI customer service: Task complexity,
problem-solving ability, and usage intention. Australasian Marketing Journal, [e-journal]
24(4), pp. 189-199. https://doi.org/10.1016/j.ausmj.2020.03.005.
Xueming, L., Tong, S., Fang, Z. and Qu, Z., 2019. Machines versus humans: The impact of
AI chatbot disclosure on customer purchases. Marketing Science, 38(6), pp. 937-947.
Young, E., 2014. How to copilot the multichannel journey key global contacts. [online]
Available at: <https://www.slideshare.net/SandraThevenaz/eyconsumersonboard-
49320356> [Accessed 3 December 2020].
AE Risks of Observable and Unobservable Biases in Artificial Intelligence Predicting Consumer Choice
102 Amfiteatru Economic
RISKS OF OBSERVABLE AND UNOBSERVABLE BIASES IN ARTIFICIAL
INTELLIGENCE USED FOR PREDICTING CONSUMER CHOICE
Florian Teleaba1*, Sorin Popescu2, Marieta Olaru3 and Diana Pitic4 1), 2) Technical University of Cluj-Napoca, Cluj-Napoca, Romania.
3) University of Economic Studies, Bucharest, Romania. 4) Babes-Bolyai University, Cluj-Napoca, Romania.
Please cite this article as:
Teleaba, F., Popescu, S., Olaru, M. and Pitic, D., 2021.
Risks of Observable and Unobservable Biases in
Artificial Intelligence Predicting Consumer Choice.
Amfiteatru Economic, 23(56), pp. 102-119.
DOI: 10.24818/EA/2021/56/102
Article History
Received: 6 August 2020
Revised: 28 October 2020
Accepted: 2 December 2020
Abstract
Companies are increasingly adopting Artificial Intelligence (AI) today. Recently however
debates started over the risk of human cognitive biases being replicated (and scaled) by AI.
Research on biases in AI predicting consumer choice is incipient and focuses on observable
biases. We provide a short synthesis of cognitive biases and their potential risk of being
replicated in AI-based choice prediction. We also discuss for the first time the risk of
unobservable biases, which affect choice indirectly, through other biases. We exemplify
this by looking at looking at three prevalent, most frequently investigated biases in
consumer behaviour: extremeness aversion, regret aversion and cognitive regulatory focus
(prevention- versus promotion-focus). Based on a sample of 1747 respondents, through
partial least squares structural equation modelling and significance testing, we show that
regret aversion (unobservable bias) significantly reduces extremeness aversion (observable
bias) and mediates the influence of cognitive regulatory focus (unobservable bias).
Keywords: cognitive bias, artificial intelligence, choice prediction, consumer choice
behaviour, regret aversion, extremeness aversion, regulatory focus.
JEL Classification: D91, D80, D01
* Corresponding author, Florian Teleaba – e-mail: [email protected]
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 103
Introduction
AI is increasingly being adopted today in business and commerce. Hopes that it can help
predict better consumer choice, to the benefit of both companies and consumers, are high.
There is however an increasing debate on whether human cognitive biases may be adopted
and replicated in AI, and even scaled by AI, leading to suboptimal predictions and
outcomes for both stakeholders. The current research literature is still incipient on how
cognitive biases can be reflected in data and further copied by machine learning – based AI
models, and how this risk should be managed. Moreover, there is currently little to no
research on whether human cognitive biases can be traced through (big) data, and on what
happens when some biases influence choice differently through other biases than on their
own – multiplicative effects. In such cases, the risk of AI replicating biases, and being
unaware of this, or wrongly estimating the effect on each bias on choice, could potentially
increase to unknown levels.
We have two goals in this paper. First, we provide a synthesis of examples of cognitive
biases and their potential risks, should they be undesirably replicated and scaled by AI. Our
second goal is to test the relationships between observable and unobservable biases and
show whether unobservable biases can, beyond simply influencing choice, mediate the
effect of other observable or unobservable biases. As such, we separate for the first time in
research these cognitive biases into two categories, observable and unobservable, and we
discuss how unobservable biases pose a double threat to AI: on the one hand, they cannot
be detected; on the other hand, they could in fact mediate (or be mediated by) other biases,
observable or unobservable in their own right. This could have tremendous implications for
research on biases in AI and in fact could represent a new paradigm of research in the field.
To study these relationships, we chose three prevalent, most frequently investigated biases
in consumer behaviour: extremeness aversion, regret aversion and regulatory focus.
Extremeness aversion (the tendency to avoid extremes and choose the middle option) is one
of the most prevalent biases in choice behaviour, an outcome of purchasing behaviour, and
thus observable through off-the-shelf (big) data (like historical purchasing data). Regret
aversion is one of the strongest and most prevalent anticipated emotions in consumer
buying; thus, it is a driver, and very likely unobservable through off-the-shelf data.
Cognitive regulatory focus is twofold. On the one hand, promotion- or a prevention-focused
cognition can be a driver of behaviour and therefore unobservable. On the other hand,
cognitive focus leads to either regret minimisation or utility maximisation behaviour, but
this is likely unobservable as well, as it is well related to regret aversion and in fact an
underlying motivation of the behaviour.
Our hypotheses for the quantitative study are as follows:
H1. Regret aversion reduces extremeness aversion in consumer choice, in the consumer’s
attempt to reduce the potential regret after choosing.
H2. Prevention regulatory focus reduces regret aversion, unlike promotion regulatory
focus.
H3. Regret aversion mediates the relationship between regulatory focus and extremeness
aversion; in other others, prevention regulatory focus combined with regret aversion
reduces extremeness aversion.
AE Risks of Observable and Unobservable Biases in Artificial Intelligence Predicting Consumer Choice
104 Amfiteatru Economic
To test these hypotheses, we use the results from a comprehensive survey of 1747
respondents, which we analyse through partial least squares structural equation modelling
(PLS-SEM) and statistical significance testing.
The paper is structured as follows. We first offer a background on the state of AI in the
consumer context today, biases in AI, and the further analysed biases: extremeness
aversion, regret aversion and regulatory focus. We then describe our research hypotheses
and framework, the research method and survey questions used, as well as methodological
aspects of data analysis in this context. Finally, we show the results, focused on (a) how
choice from a set of low-medium-high alternatives is influenced by regret aversion, (b) how
choice is influenced by consumer cognitive regulatory focus (prevention-focused or
promotion-focused cognition), and (c) the mediation effect of regret aversion on the
relationship between cognitive regulatory focus and choice and extremeness aversion.
1. Background and research opportunity
1.1. The state of artificial intelligence today in consumer context
Kearney, recently stated that companies must “embrace AI to survive” (Kearney, n.d.). AI
is therefore a must for survival in business today, no longer just a nice-to-have on the
CEO’s agenda, something a company can afford to leave aside. While the first AI
algorithms appeared in the 1960s with pre-programmed and rule-based learning (if-then
reasoning), AI has moved along the spectrum of intelligence into various kinds of
supervised and unsupervised learning – figure 1 below provides a summary of where AI
has been used until today and expectations of its use over the next decade, along the four
main domains of AI and machine learning: natural language processing, computer vision,
pattern recognition, and reasoning and optimisation.
Figure no. 1. The use of artificial intelligence today and beyond
Source: Kearney, n.d.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 105
AI today is expected to dramatically affect how different corporate functions in business,
and commerce in particular, accomplish their goals, especially in marketing (Davenport, et
al., 2020), marketing personalisation (Kumar, et al., 2019) and advertising (Kietzmann,
Paschen and Treen, 2018), as well as back-end functions like product development,
sourcing, supply chain management and manufacturing. AI applications in commerce are
technically focused on pattern recognition. Practically, beyond offering insights on the
target customer audience and revealing complex patterns in consumer choice data, the
focus is on predicting consumer choice, from analysing hidden customer preferences to
making product recommendations. The benefits should be positive and clear: customers
will spend more, will become more loyal, will trust brands more, or will adopt behaviours
(e.g., shifting to channels) that optimise costs for companies.
1.2. Biases in AI
Recently, however, there has been significant debate over the risk of human cognitive
biases being replicated (and scaled) in machine learning – based AI models, as such biases
are reflected in the (big) data that AI models learn from, in applications ranging from courts
and law enforcement to medicine to business. As Manyika, Silberg and Presten (2019)
show: “Over the past few years, society has started to wrestle with just how much these
human biases can make their way into artificial intelligence systems — with harmful
results”. One great recent example is Apple’s being accused of sexism in 2019 because the
company’s new credit card seemed to offer men more credit than women (BBC, 2019).
Another is the case of COMPAS, the computer program used to calculate the likelihood of
prisoners reoffending, coming under serious scrutiny because it was found to be biased
against African-American defendants (Dressel and Farid, 2018). IBM research states that
“within five years, the number of biased AI systems and algorithms will increase” (IBM
Research, 2018). Tackling bias in AI is therefore one of the priorities on the AI research
frontier (Silberg and Manyika, 2019). But while much of the focus (and concern) of
research related to biases in AI is on building fairness and equity in machine learning (and
in particular areas like medicine, on the correctness of prediction), in prediction of
consumer choice in commerce, the risk of misprediction can have negative consequences
such as providing consumers with products or services that do not bring them the required
or needed value. That can have as a spill-over effect negative consequences on companies’
revenues and profitability due to decreasing customer satisfaction and loyalty. Especially
since one of the expected next uses of AI is mimicking intuition and creative connecting of
dots, as well as beating human forecasting in several domains (see Figure 1), managing
cognitive biases and avoiding their replication and scaling in AI models are critical.
We list below in Table 1 some of the most notable cognitive biases researched in
psychology and behavioural economics and explain the potential risk they pose should they
be replicated (and scaled) in AI predicting consumer choice. We also classify them into two
categories: observable and unobservable. To understand the difference, take the example of
a retailer that would like to understand its customers’ biases by analysing large datasets of
customers’ purchases over the last 3 years; these datasets contain information about which
products consumers bought, product attributes, product availability, even competitive
intelligence – observable data, in other words. There are cognitive biases that are expected
to be easily identified in AI/machine learning through analysing only the mentioned
datasets and employing the right analysis techniques. An example could be extremeness
AE Risks of Observable and Unobservable Biases in Artificial Intelligence Predicting Consumer Choice
106 Amfiteatru Economic
aversion in a choice set (defined below). By analysing the choice patterns and controlling
for factors such as price and product attributes, such a bias can likely be easily isolated.
There are, however, biases or attitudes that are not observable in this way, i.e., nearly
impossible to detect solely based on analysis of purchase (big) data. To detect regret
aversion (defined below), for example, in consumer purchase data, additional customer
research would likely be required on specific samples, and then extrapolation to the entire
data through specific methods. We do not imply this is impossible, yet it would be much
more difficult than identifying biases directly observable in actual purchase data.
Table no. 1. Cognitive biases and potential risks in being replicated (and scaled)
in AI predicting consumer choice Cognitive bias
(and alike) Potential risk in being replicated (and scaled) in AI predicting consumer choice
Observable/
unobservable
Different utility
types (transaction
utility, procedural
utility, etc.)
While people experience pleasure from finding a good ‘deal’, or from the fairness of the
transaction, this could be picked up by AI as a key driver of consumer choice,
underestimating the importance of other factors and putting too much focus on
promotion/discounting. This could be detrimental to both the company (margin erosion) and
consumers (making them buy products of less value to them just because they have a good
price).
Observable
Satisficing
behaviour (and
alike)
Any brand choice behaviour that deviates from the utility maximisation theory does so
because consumers do not have the cognitive power, patience or access to all relevant
information. More than 25 distinct models of brand choice behaviour deviating from
traditional utility theory exist. AI could ‘learn’ to replicate this behaviour as being optimal
and start predicting choices or recommending products that are not really optimal to
consumers.
Observable
Heuristics
thinking
All choices made based on heuristics such as anchoring, availability, representativeness,
recognition, are not, in most cases, optimal for the consumer in long run. As above, AI
however could ‘learn’ to replicate this behaviour start predicting or recommending
suboptimal choices.
Observable
Pain of paying
Some consumers are tightwads, not liking to spend money, often buying cheaper than is
optimal. Spendthrifts however like to spend money and often spend more than is optimal. An
AI model which does not isolate this ‘pain of paying’ might simply propose the wrong price
level to the wrong segment.
Observable
Variety seeking,
try-new bias
The fact that consumers might sometimes have a tendency to switch products or brands
simply because of a ‘try-new’ bias might not be in their best economical/utilitarian interest
but only satisfy an impulsive emotional need. AI however could ‘learn’ this as a beneficial
rule and predict choices that are not optimal for consumers instead of helping reduce that bias.
Observable
Choice
architecture
The way choices are designed and framed leads to various biases (extremeness aversion,
attraction/decoy effect choice overload, evaluability hypothesis, distinction bias, less is better
effect, etc.). An AI model however needs to be actively taught to ‘detect’ these as such, not as
a direct preference.
Observable
Information
avoidance
The fact that people might avoid some information (consciously or not) is, again, because of
limited cognitive power or patience, not because it’s in their best interest. AI could however
‘learn’ that this is desirable behaviour.
Unobservable
Diversification
bias
The fact that people prefer more choices or features that are not needed today, ‘just in case’
they are needed in the future, is a serious psychological aspect. AI might nonetheless not
detect if a preference for diversity or some features is triggered by such a bias or by actual in-
the-moment preference and again may lead to predictions that are inaccurate or not optimal
for the consumer.
Unobservable
Hyperbolic
discounting,
planning fallacy,
dual-self model
When consumers deal with intertemporal choices, they tend to focus on the short rather than
the long term and discount the value of a later utility/benefit by a factor that increases with the
length of the delay. That is however suboptimal for them, yet AI could interpret it as a
rational, beneficial choice.
Unobservable
Social norms,
proof, herd
behaviour, reason-
based choice
When choice is influenced by social aspects (how others see it, what others do, how to justify
it to others, how to justify it to yourself and against your values/beliefs), choice deviates from
expected utility theory. AI could ‘learn’ to interpret this as rational, even if it may not be
beneficial to the consumer.
Unobservable
Emotions,
feelings, deeper
motivations
Anticipated emotions such as regret aversion, or even unconscious feelings or motivations of
consumer behaviour, influence consumer choice yet may ‘fool’ AI into attributing the choice
to the observable product and context attributes (price, product specifications, environment,
etc.).
Unobservable
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 107
Unobservable cognitive biases (which cannot be directly observed through off-the-shelf data) pose another, more serious risk to AI/machine learning, which has been undiscussed and unexplored until now: it is not only their direct effect on consumer choice which goes unnoticed, but they can also mediate the effects of other biases, whether observable or not on their own.
1.3. Extremeness aversion in dealing with multiple choices
Research on multiple choices and choice in context has been at the centre of behavioural economics. Among other things, it shows that people tend to avoid choosing the extreme options in a choice set. Simonson and Tversky introduced the term “extremeness aversion” in their famous 1992 paper (Simonson and Tversky, 1992). Extremeness aversion states that the attractiveness of an option is enhanced if it is an intermediate option in a choice set and diminished if it is an extreme option. Extremeness aversion is highest when the middle alternative is exactly at the centre in the choice set (equal distances from the extremes) (Padamwar, Dawra and Kalakbandi, 2018). A recent meta-analysis (Neumann, Bockenholt and Sinha, 2016) showed that extremeness aversion is one of the most robust phenomena in consumer behaviour, and that its strength varies: it is weaker when employing price-quality trade-offs, nondurable categories or binary-trinary choice-set comparisons, and stronger when using a large number of trade-off dimensions, non-numeric attributes, high-quality extensions, or utilitarian products. Simonson, Sela and Sood (2017) showed that people are in general unaware of their tendency to avoid extremes, and when made aware of it, they may deny it and find counter-examples or other explanations – avoiding extreme options is thus likely considered as a weakness and inconsistent with a person’s self- and other-image as a decisionmaker.
1.4. Regret aversion or fear of a better option in making product and brand choices
Regret theory in economics was initially formulated, simultaneously, by Loomes and Sugden (1982), Bell (1982), and Fishburn (1982), and has been researched since then. Regret aversion, or fear of a better option in more ‘commercial’ terms, is generally defined as a negative cognitively determined emotion appearing when comparing an obtained decision outcome to outcomes that might have been, had one chosen differently (Rosenzweig and Gilovich, 2012; Van Dijk and Zeelenberg, 2005). In economic rather than psychological terms, regret is the “disutility an individual experiences from the value gap between an actual outcome and the best possible outcome that one could have attained” (Braun and Muermann, 2004). In regret per se, literature differentiates between experienced regret and anticipated regret. Experienced regret (or rejoice) can lead to risk-averse behaviour (Creyer and Ross, 1999) and may affect the anticipation of regret (Coricelli, et al., 2005; Cooke, Meyvis and Schwartz, 2001; Creyer and Ross, 1999). It can lead consumers to switch to a different product (Zeelenberg and Pieters, 2004) or brand (Zeelenberg and Rik, 2007), and it can (directly and indirectly) negatively influence customer satisfaction and repurchase intentions (Tsiros and Mittal, 2000). Anticipated regret on the other hand is a stronger emotion than experienced regret. Research shows people do not necessarily want to avoid making the same mistake twice as much as they do not want to experience the same negative emotion twice (Raeva, Mittone and Schwarzbach, 2010). Anticipated regret therefore tends to be overestimated compared to actual experienced regret (Gilbert, et al., 2004; Sevdalis and Harvey, 2007), and it strongly affects purchase or choice-making behaviour. Anticipated regret can lead for example to purchase of a currently available item on sale rather than waiting for a better sale, or to preference for
AE Risks of Observable and Unobservable Biases in Artificial Intelligence Predicting Consumer Choice
108 Amfiteatru Economic
a higher-priced, well-known brand over a less expensive, lesser-known brand (Simonson, 1992). In some instances, it can even lead to complete inertia (or status quo option selection) and ambiguity-driven indecisiveness (Sautua, 2017). Anticipation of regret can even augment other cognitive biases, such as herding behaviour as a rational response to regret aversion (Arlen and Tontrup, 2015) (i.e., observing the choices of many others or professionals, who are perceived as less likely to be biased by regret). Anticipated regret can also induce an endowment effect; for example, owners resist selling houses or increase selling prices because they experience more anticipated regret over selling in error than over failing to make a deal when they should have (Thaler, 1980). As a note, though often confused, regret aversion and risk or loss aversion are different cognitive and emotional responses. Regret aversion has been shown to be even stronger than risk aversion; experiments show people choosing regret-minimising gambles over risk-minimising gambles in both gain and loss contexts and in both relatively high-risk and low-risk pairs of gambles (Zeelenberg, et al., 1996). Regret aversion has been demonstrated to be a more powerful predictor of behaviour than many other types of anticipated negative emotions or risks (Brewer, et al., 2016). In fact, most purchase decisions are frequently a source of regret for most consumers (Rosenzweig and Gilovich, 2012); they often compare their purchase outcomes with what they could have bought differently (Abendroth and Diehl, 2006) and often experience second thoughts and anxiety (Inman and Zeelenberg, 2002).
1.5. Cognitive regulatory focus
According to recent research, consumers may have different appetites for minimising regret or maximising the utility of their choice from a set of alternatives, also depending on their regulatory focus typology: chronically prevention-focused consumers are more likely to be regret minimisers, and chronically promotion-focused consumers are more likely to be utility maximisers (Lim and Hahn, 2019). Moreover, research has shown that people with an ingrained preference for moderation in their lives also tend to have extremeness aversion in experiments with sets of three alternatives (Simonson, Sela and Sood, 2017). It is therefore evident that studying the relationship between regret aversion and choice and extremeness aversion needs to be done, accounting also for consumers’ regulatory focus (prevention- versus promotion-focus) and their regret minimisation versus utility maximisation goals as well.
1.6. Missing link, and risk of unobservable risks (biases)
According to Connolly and Butler (2006), numerous studies have investigated how choice can be influenced by anticipated regret or disappointment with the choice. However, the effect of fear of a better option or anticipated regret on extremeness aversion in making a choice from a set of “low-medium-high” alternatives has not been researched so far, let alone with respect to cognitive regulatory focus types. In other words, do consumers who generally have a fear of a better option when making choices between products or brands tend to have a reduced extremeness aversion (compared to those that do not typically think of better options or potential regret of their choice)? If so, in the hypothetical retailer example we discussed previously, the retailer might detect extremeness aversion from its off-the-shelf customer purchase data, and potentially even promotion- versus prevention-focused cognition (potentially, however unlikely, as this is determined through customer research, not based on purchase data); however, we highly doubt off-the-shelf data can reveal consumers’ (anticipated) emotions like regret aversion.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 109
2. Research framework and methodology
2.1. Research framework and hypotheses
Based on the discussed gap between regret aversion and extremeness aversion, we
formulate our first hypothesis:
H1: Consumers who typically exhibit regret aversion, or fear of a better option, will tend to
have a lower extremeness aversion and select the high product alternative in a set of three
product alternatives in an attempt to reduce their potential regret of the choice to be made.
Based on the previously discussed understanding that consumers exhibit different
behaviours of minimising their regret or maximising the utility of their choice depending on
their cognitive regulatory focus typology (prevention-focused or promotion-focused), we
formulate the second and third hypotheses of our research:
H2: Consumers that are prevention-focused tend to be regret minimisers and thus have a
lower extremeness aversion than those who are promotion-focused (who will tend to select
the middle option).
Finally, we are able to formulate our third hypothesis and explore the effects of
unobservable biases on observable biases and their influence on choice. As such:
H3: The relationship between cognitive regulatory focus typology and extremeness
aversion is mediated by regret aversion. Consumers who are typically prevention-focused
and exhibit regret aversion will tend to have lower extremeness aversion and select the high
product alternative in a set of three product alternatives more often than those that typically
have lower fear of a better option.
Figure no. 2. Research framework and hypotheses
2.2. Research method, survey design, raw data transformation – methodological aspects
To test these hypotheses and effects, we use a comprehensive quantitative survey deployed
online through a specific professional online survey instrument and based on a random
sampling approach. The survey was deployed as part of a professional consulting project by
one of the authors in the retail banking industry in Romania. Data collection was monitored
AE Risks of Observable and Unobservable Biases in Artificial Intelligence Predicting Consumer Choice
110 Amfiteatru Economic
frequently based on age and income quotas to ensure representativity of the sample for the
population. 1747 complete responses were collected, as further described, and the results
are statistically representative of the adult (24+) population owning a current account in a
bank in the respective country.
The set of product alternatives tested through the survey is depicted in the following figure 3.
Prices construction (the middle option being placed at equal distances from the extremes in
terms of price) and the additional remarks provided to respondents in terms of potential
savings through those prices were designed to ensure sufficient attractiveness of each
option on its own. The answers to this question are coded as 1, 2 or 3, depending on which
package was chosen (basic, standard or premium, respectively).
Figure no. 3. Set of alternatives tested
To quantify respondents’ regret aversion, or fear of a better option, we asked them choose
between never, sometimes and always to the question: After you have chosen a product or
service and purchased it, how often do you think or feel you might have found a better offer
or option, in terms of quality and/or price, if you had looked more? – see figure 4 below.
The way the question is formulated is highly relevant for the purpose of this study, better
than directly asking a question like Do you typically regret your choices or Do you typically
fear a better option could exist? because it offers a reference point regarding when the
feeling under question occurs, i.e., after you make your choice and purchase, it offers a
reference point regarding how the feeling exactly is experienced, i.e., if you had looked
more, and finally it offers a simple scale to measure the feeling without any confusion.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 111
Figure no. 4. Question for determining regret aversion (fear of a better option)
To segment the respondents between prevention-focused and promotion-focused cognitive
regulatory focus types (i.e., regret minimisation versus utility maximisation behaviour), we
asked them to evaluate each of the five key statements as shown in the below figure 5.
Figure no. 5. Question to determine cognitive regulatory focus
We classify between promotion- and prevention-focused consumers as such:
Respondents with a score of 4 on both statements 1 and 3 and without a score of 4 on
statement 5, as a clearly promotion-focused segment (coded with 3)
Respondents with a score below 4 on both statements 1 and 3 and without a score of 1
on statements 2 and 4, as a clearly prevention-focused segment (coded with 1)
The rest of respondents as a ‘balanced’ segment (coded with 2)
This approach to classifying promotion and prevention cognitive regulatory focus is similar
in principle to the original method of Higgins, et al. (2001) and further utilised by Lim and
Hahn (2019) to study its influence on regret minimisation behaviour, but has been adapted
for the purposes of the original consulting project in which the survey was deployed and to
tailor it toward purchasing, not general instances.
We believe the above indirect way of constructing the questions and statements to be
evaluated is a better method for isolating promotion-focused and prevention-focused
consumers than simply asking a direct choice question (i.e., “Do you consider yourself this
type or this type? Choose.”), as often in research, consumers are not cognitively able to
specify their preferences correctly or may be unwilling to disclose real motivations or
preferences (fear of appearing superficial or influenced by social norms). Likewise, as
aiming to understand whether specific consumer typologies influence another variable is
not a statistically optimal way to segment consumers based on characteristics or outcomes,
this approach is also preferred, instead of approaches like cluster analysis or PLS (partial
least squares) prediction-oriented segmentation.
We used several ex-ante mechanisms to reduce the risk of common method bias or variance
(CMV). The online survey tool provided big visible indications at the beginning of the
survey about anonymity and confidentiality and that responses should be as honest as
AE Risks of Observable and Unobservable Biases in Artificial Intelligence Predicting Consumer Choice
112 Amfiteatru Economic
possible. We ordered the questions in such a way that no question could have a priming
effect on a subsequent question. We formulated the questions in a simple way so
respondents wouldn’t have to expend too much cognitive effort in answering them.
Commonality in scale endpoints (or anchor effects) is not very likely to affect responses, as
there are only a few questions using same scale. As an ex-post check, we ran Harman’s
single factor test across all variables (except those for basic profiling like age) using SPSS’s
factor analysis procedure, with principal axis factoring as the extraction method: one single
fixed factor could not be extracted; therefore we have no indication that CMV might exist
(only if a single factor can be extracted and it explains above 50% of variance in the sample
is there an indication that CMV might exist; see more in Podsakoff, et al., 2003, or Chang,
van Witteloostuijn and Eden, 2010).
3. Data analysis – methodological aspects
The relationships presented in our framework (figure 1) can naturally be analysed through
regression analysis; therefore they can be described using the following regression
equations and remarks.
In the following equations below, the main outcome variable “Choice” is measured on a
scale from 1 to 3 (as per figure 3, where 1 denotes choice of the Basic package, 2, the
Standard package, and 3, the Premium package). Variable FOBO_seg denotes the level of
fear of a better option (regret aversion) and is measured on a scale from 1 to 3 (see remarks
after figure 4).
For H1: Choice = B10 + B1FOBO_seg + e
In the next equation for H2, Cognitive_seg denotes the cognitive typology and is measured
on a scale from 1 to 3 (as per the remarks after figure 5).
For H2: Choice = B20 + B2Cognitive_seg + e
Analysing the mediation effect of ‘fear of a better option’ on the relationship between
‘cognitive typology’ and ‘extremeness aversion’ requires coefficients from two linear
regressions:
For H3 – part 1 of 2: FOBO_seg = B3a0 + B3aCognitive_seg+ e, and
For H3 – part 2 of 2: Choice = B3b0 + B1-2FOBO_seg + B3bCognitive_seg + e
More specifically, we will utilise the so-called Sobel product of coefficients approach
(Sobel, 1982), where coefficients B3a and B1-2 are multiplied to obtain the indirect effect of
FOBO, i.e., the mediation effect.
Bindirect = B3a * B1-2
In case B3a * B1-2 is significant and B2 (from equation for H2) is not, the indirect effect is
considered as full mediation. In case B2 is significant, the indirect effect is considered only
partial mediation (complementary if B3a * B1-2 * B2 is positive, or competitive if B3a * B1-2 *
B2 is negative).
Advanced procedures (and software) exist to model efficiently and correctly equations like
above, such as partial least squares structural equation modelling – PLS-SEM. We use
SmartPLS 3.2.9 software for this purpose, which enables both path modelling for the
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 113
structural equation model (all our hypotheses combined, as our framework depicts in figure
1) and bootstrapping for testing whether path coefficients are statistically significant. In this
case, the path coefficients are interpreted as standardised regression coefficients (i.e., the
effect of one standard deviation increase in a predictor variable on the outcome).
In this analysis, because our dependent variable is ordinal (it receives the values of 1, 2 and
3), a significant positive coefficient of any independent variable implies nothing about
extremeness aversion. It can explain an effect on the overall choice pattern, i.e., if a ‘larger’
account package is selected in this case, which we are still interested in studying, but not on
the tendency to choose the middle ‘standard’ package.
Therefore, to truly understand the influence on extremeness aversion (beyond the influence
on choice), we use standard significance testing of the differences between various groups
of respondents (grouped by their defining characteristics: fear of a better option, cognitive
segment) in terms of extremeness aversion: for example, the difference in extremeness
aversion between respondents with high fear of a better option, medium fear of a better
option, and low fear of a better option.
For this approach, we use two tests. We use the standard Student’s t-test to determine if
each individual choice percentage is statistically significant. Each group is compared to the
opposite one based on its characteristics. For example, if we analyse the choice pattern of
those with high FOBO, we compare it with the choice patterns of those with low and
medium FOBO. We also use the chi-square test as a non-parametric test for determining if
the actual distribution of choice percentages is statistically significant. For example, if we
analyse the choice pattern of those with high FOBO, we compare it with the choice patterns
of those with low and medium FOBO.
4. Results
Table 2 below shows the distribution of choice percentages in the set of alternatives. As is
easily seen, extremeness aversion appears to exist, on average, across all respondents, with
more than half of them choosing the middle option (Standard package). The table also
summarises respondents’ patterns of fear of a better option and their cognitive segment.
Table no. 2. Choice patterns and respondents’ characteristics
(% of total, # of respondents)
Distribution of
choices
Distribution of respondents by
fear of a better option segment
Distribution of respondents
by cognitive segment
Choice 3
(Premium
package)
22%
(384) FOBO_seg – low
44%
(762)
Cognitive_seg –
prevention-focused
16%
(271)
Choice 2
(Standard
package)
57%
(996) FOBO_seg – medium
42%
(741)
Cognitive_seg –
balanced
60%
(1055)
Choice 1
(Basic
package)
21%
(367) FOBO_seg – high
14%
(244)
Cognitive_seg –
promotion-focused
24%
(421)
AE Risks of Observable and Unobservable Biases in Artificial Intelligence Predicting Consumer Choice
114 Amfiteatru Economic
Figure 6 below represents the actual output from SmartPLS (organised visually as per our
research framework in figure 1), which shows the path coefficients for each relationship
hypothesised and their p-values in parentheses (derived from running bootstrapping on
5000 samples). Fear of a better option has a statistically significant effect on the choice
made from the set of product alternatives, whereas Cognitive_seg does not (at least at the
5% level, although p-value indicates an effect at the 10% level). Cognitive_seg appears to
have a positive influence on fear of a better option on the other hand. Moreover, fear of a
better option appears to have a full mediation effect on the relationship between cognitive
typology and choice (as the coefficient of Cognitive_seg on choice is not significant, at
least at the same significance level) – this mediation (indirect) effect is 0.002 (or using the
Sobel product of coefficients, 0.044 * 0.047). Overall, effects’ sizes still appear small,
however, measured here in standard deviations.
Figure no. 6. Smart-PLS output results (consistent PLS algorithm and bootstrapping)
4.1. Standard significance testing across groups of respondents and their choice
patterns
Below, table 3 shows the distribution of choice percentages for different groups of
respondents. Groups are organised by lines, and their defining characteristics mapped on
the first columns with tick-marks. For example, line 1 shows all respondents (no filters or
tick-marks); line 2 shows only respondents with a low fear of a better option; line 10 only
those with low fear of a better option and prevention-focused cognition. The following
columns show the percentages of respondents selecting choice 1 (basic package), choice 2
(standard) and choice 3 (premium), and in parentheses, the calculated t-statistics, whereas
the last column shows the calculated χ2 values. Percentages statistically different from the
comparison groups are marked with an asterisk (*) – see table legend – and bolded for
easier visualisation. Where two test statistics appear one below the other, the first (top
value) is the test statistic (t-stat or χ2) value calculated when compared to the group one line
above (the same where only one test statistic appears), while the second (bottom value) is
the test statistic value calculated when compared to the group two lines above.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 115
Table no. 3. Distributions of choice percentages and standard significance testing
results
Line
no.
Fear of better option
(FOBO)
Cognitive typology
(Cognitive_seg)
Distribution of choice percentages
(t-stat values)
χ2 values
Low Mediu
m
High Prevention
-focused
Balanced
cognition
Promotion
-focused
% choice 1 % choice 2 % choice 3
1 21% 57% 22%
2 X 19% 63% 18%
3 X 20%
(0.47) 56%
(-3.71**)
23%
(4.22**)
2.69
4 X 26%
(2.23**)
(2.53***)
42%
(-4.47***)
(-6.72***)
32%
(3.13***)
(5.90***)
8.26**
20.72***
5 X 22% 58% 20%
6 X 18%
(-2.81***)
59%
(0.42)
23%
(2.36***)
1.02
7 X 26%
(4.09***)
(2.05**)
54%
(-2.04**)
(-1.77**)
20%
(1.36*)
(0.07)
4.00
1.09
8 X X 22% 63% 15%
9 X X 16%
(-3.04***)
66%
(1.01)
18%
(2.18**)
2.38
10 X X 27%
(3.68***)
(1.53*)
55%
(-2.77***)
(-2.14**)
18%
(-0.12)
(1.12)
8.60**
2.82
11 X X 20% 57% 23%
12 X X 19%
(-0.81)
57%
(0.01)
25%
(0.75)
0.23
13 X X 23%
(1.54*) (0.94)
56%
(-0.15)
(-0.14)
21%
(-1.22)
(-0.74)
1.60
0.59
14 X X 29% 42% 29%
16 X X 23%
(-1.71**)
42%
(0.04) 35%
(1.67**)
2.64
16 X X 31%
(1.65**)
(0.39)
43%
(0.24)
(0.27)
25%
(-1.70**)
(-0.68)
5.97*
0.72
Note: ***significant at 1% level or below, **significant at 5% level or below, *significant
at 10% level or below
Some of these results confirm what previous PLS-SEM modelling revealed. For example,
the results on line 4 are highly statistically significant, showing that consumers with a high
fear of a better option have a different choice behaviour, and choice pattern, from the set of
product alternatives – they avoid the middle option more frequently, in favour of either of
the extremes, and therefore display a significantly lower extremeness aversion. Even those
with a medium fear of a better option tend to avoid the middle option in favour of at least
the highest alternative (line 3 results). These results show a major influence of fear of a
better option on the choice behaviour and extremeness aversion, influence which previous
PLS-SEM modelling indicated to be statistically significant but rather low.
Cognitive typology, however (lines 5-7), appears to play a role unconfirmed by the
previous PLS-SEM modelling (at least not at the 5% level in the latter case). Those who are
AE Risks of Observable and Unobservable Biases in Artificial Intelligence Predicting Consumer Choice
116 Amfiteatru Economic
promotion-focused tend to favour the basic alternative more often, and the middle option
less often, compared to those who are prevention-focused or balanced; therefore, they show
lower extremeness aversion in favour of only one extreme (the lowest one).
These behaviours seem to be different between those with high and medium fear of a better
option and those with low fear of a better option. With low fear of a better option,
promotion-focused consumers display lower extremeness aversion compared to balanced or
prevention-focused consumers, and a higher inclination to favour the cheapest alternative
compared to prevention-focused consumers. With medium or high fear of a better option,
extremeness aversion stays similar between different cognitive typologies (and the
preference for either of both extremes does not change significantly). It seems therefore that
once fear of a better option settles in and is especially high, cognitive typology no longer
plays a major role in influencing consumers’ extremeness aversion or the attractiveness of
either of the extremes. This indicates that this fear of a better option (or regret aversion) is
stronger than consumers’ cognitive regulatory focus.
As a closing remark, there is one intriguing behaviour specific to those with a balanced
cognition, in other words, not prevention- or promotion-focused. Even if they seem to have
the same extremeness aversion level compared to prevention-focused segment (percentage
of choice 2), they clearly favour more the expensive alternative over the cheapest one
compared to both other segments. Even more, this happens with different levels of fear of a
better option, but again, when fear of a better option is high, the effect is bigger. It appears
that cognitive indecisiveness (having no tendency to either minimise regret or maximise the
utility/value of choice) makes consumers more prone to favour the higher/highest product
alternatives, and this effect is also augmented by regret aversion.
Conclusions
Regret aversion, or fear of a better option, has a major influence on consumers’ choice
patterns from a set of low-medium-high alternatives and their extremeness aversion.
Hypothesis H1 is therefore confirmed. This influence has been unexplored or unexplained
until now, yet with important implications for brands and retailers. Regret aversion
significantly reduces extremeness aversion. It doesn’t lead only to the high alternative
being chosen, however, as initially expected. Extremeness aversion is reduced in favour of
either one of the extremes; therefore, other factors may influence which extreme is chosen,
when and why, and they should be researched further.
Cognitive regulatory focus in purchasing (being prevention- or promotion-focused, or
balanced) also plays a role in influencing consumers’ choice and extremeness aversion.
Hypothesis H2 is therefore confirmed. Promotion-focused consumers tend to have a lower
aversion to extremes, especially in favour of the cheapest option, yet this is somewhat
expected, as they are more likely to perceive a low (or zero) priced option as a bargain.
When regret aversion is higher, cognitive regulatory focus typology no longer significantly
influences consumers’ extremeness aversion or their choice pattern. A strong mediation
effect of regret aversion is visible. Therefore, Hypothesis H3 is also confirmed. Regret
aversion, as an anticipated emotion, appears overall to be stronger in influencing choice and
extremeness aversion than consumer cognitive regulatory focus on its own, and to mediate
its effect.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 117
These insights bring important implications for how to study biases in AI and establish an
important research agenda for differentiating and anticipating unobservable biases in
addition to observable ones, as well as their multiplicative effects. In our example case,
through understanding what influences consumers’ choice patterns and their extremeness
aversion, an AI model could design a better choice architecture with improved business
results (sales and profitability); product alternatives could be better designed through
different features and price levels and better displayed in store or online to shift consumers
towards more premium alternatives. However, an AI model would need to differentiate
very well between (a) the ‘first-order’ effect of extremeness aversion, which is observable,
(b) unobservable effects like those of regret aversion and cognitive regulatory focus, and (c)
the ‘second-order’ indirect effects (multiplicative effects) like the one regret aversion has in
fully mediating the role of regulatory focus.
An additional point on the research agenda should be drawn beyond only predicting what
consumers will choose: predicting/understanding how they choose, i.e., which brand choice
behaviour models consumers adopt, when, why and how. This is also not currently
explored. Existence of unobservable biases and of their multiplicative effects on observable
biases should be studied in relation also to how complex the adopted brand choice
behaviour model is (i.e., from the simplest heuristics-based choice models, to attribute-
based sequential elimination models like elimination-by-aspects, to conjunctive, disjunctive
or lexicographic rules, up to the fully fledged utility maximisation model from neoclassical
economics).
References
Abendroth, L. and Diehl, K., 2006. Now or Never: Effects of Limited Purchase
Opportunities on Patterns of Regret over Time. Journal of Consumer Research, 33(3),
pp.342-351.
Arlen, J. and Tontrup, S., 2015. Strategic Bias Shifting: Herding as a Behaviorally Rational
Response to Regret Aversion. Journal of Legal Analysis, 7(2), pp.517-560.
BBC, 2019. Apple's 'sexist' credit card investigated by US regulator, [online] Available at:
<https://www.bbc.com/news/business-50365609> [Accessed 4 December 2020].
Bell, D.E., 1982. Regret in decision making under uncertainty. Operations Research, 30(5),
pp.961-981.
Braun, M. and Muermann, A., 2004. The Impact of Regret on the Demand for Insurance.
Journal of Risk and Insurance, 71(4), pp.737-761.
Brewer, N.T., DeFrank, J.T. and Gilkey, M.B., 2016. Anticipated Regret and Health
Behavior: A Meta-Analysis. Health Psychology, 35(11), pp.1264-1275.
Chang, S.-J., van Witteloostuijn, A. and Eden, L., 2010. Common method variance in
international business research. Journal of International Business Studies, 41, pp.178-184.
Connolly, T. and Butler, D., 2006. Regret in economic and psychological theories of
choice. Journal of Behavioral Decision Making, 19(2), pp.139-154.
Cooke, A., Meyvis, T. and Schwartz, A., 2001. Avoiding future regret in purchase timing
decisions. Journal of Consumer Research, 27, pp.447-459.
AE Risks of Observable and Unobservable Biases in Artificial Intelligence Predicting Consumer Choice
118 Amfiteatru Economic
Coricelli, G., Critchley, H.D., Joffily, M., O’Doherty, J.P., Sirigu, A. and Dolan, R.J., 2005.
Regret and its avoidance: a neuroimaging study of choice behavior. Nature
Neuroscience, 8(9), pp.1255-1262. DOI: 10.1038/nn1514.
Creyer, E.H. and Ross, W.T., 1999. The development and use of a regret experience
measure to examine the effects of outcome feedback on regret and subsequent choice.
Marketing Letters, 10(4), pp.373-386.
Davenport, T., Guha, A., Grewal, D. and Bressgott, T., 2020. How artificial intelligence
will change the future of marketing. Journal of the Academy of Marketing Science, 48,
pp.24-42.
Dressel, J. and Farid, H., 2018. The accuracy, fairness, and limits of predicting recidivism.
Science Advances, 4(1), pp.eaao5580. doi: 10.1126/sciadv.aao5580.
Fishburn, P.C., 1982. The Foundations of Expected Utility. Theory and Decision Library,
31, pp.176.
Gilbert, D. T., Morewedge, C. K., Risen, J. L. and Wilson, T. D., 2004. Looking Forward to
Looking Backward The Misprediction of Regret. Psychological Science, 15(5),
pp.346-350.
Higgins, T.E., Friedman, R.S., Harlow, R.E., Idson L.C., Ayduk, O.N. and Taylor, A.,
2001. Achievement orientations from subjective histories of success: Promotion pride
versus prevention pride. European Journal of Social Psychology, 31(1), pp. 3-23.
IBM Research, 2018. AI bias will explode. But only the unbiased AI will survive, [online]
Available at: <https://newsroom.ibm.com/IBM-research?item=30305> [Accessed 4
December 2020].
Inman, J.J. and Zeelenberg, M., 2002. Regret in repeat purchase versus switching decisions:
The attenuating role of decision justifiability. Journal of Consumer Research, 29(1),
pp. 116-128.
Kearney, n.d. Embrace AI to survive. [online] Available at: <https://www.kearney.com/
operations-performance-transformation/article/?/a/will-you-embrace-ai-fast-enough>
[Accessed 4 December 2020].
Kietzmann, J., Paschen, J. and Treen, E., 2018. Artifcial Intelligence in Advertising: How
Marketers Can Leverage Artifcial Intelligence Along the Consumer Journey. Journal of
Advertising Research, 58(3), pp. 263-267.
Kumar, V., Rajan, B., Venkatesan, R. and Lecinski, J., 2019. Understanding the Role of
Artificial Intelligence in Personalized Engagement Marketing. California Management
Review, 61(4), pp. 135-155.
Lim, J. and Hahn, M., 2019. Regulatory focus and decision rules: Are prevention-focused
consumers regret minimizers?. Journal of Business Research, 120(C), pp.343-350. DOI: 10.1016/j.jbusres.2019.11.066.
Loomes, G. and Sugden, R., 1982. Regret theory: An alternative theory of rational choice
under uncertainty. Economic Journal, 92(4), pp.805-824.
Manyika, J., Silberg, J. and Presten, B., 2019. What Do We Do About the Biases in AI?,
[online] HBR. Available at: <https://hbr.org/2019/10/what-do-we-do-about-the-biases-
in-ai> [Accessed 4 December 2020].
Neumann, N., Bockenholt, U. and Sinha, A., 2016. A meta-analysis of extremeness
aversion. Journal of Consumer Psychology, 26(2), pp.193-212.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 119
Padamwar, P., Dawra, J. and Kalakbandi, V., 2018. Range effect on extremeness aversion.
Decision, 45(4), pp.345-355. https://doi.org/10.1007/s40622-018-0197-5.
Podsakoff, P.M., MacKenzie, S.B., Lee, J.-Y. and Podsakoff, N.P., 2003. Common Method
Biases in Behavioral Research: A Critical Review of the Literature and Recommended
Remedies. Journal of Applied Psychology, 88(5), pp. 879-903.
Raeva, D., Mittone, L. and Schwarzbach, J., 2010. Regret now, take it now: On the role of
experienced regret on intertemporal choice. Journal of Economic Psychology, 31(4),
pp.634-642.
Rosenzweig, E. and Gilovich, T., 2012. Buyer's Remorse or Missed Opportunity?
Differential Regrets for Material and Experiential Purchases. Journal of Personality and
Social Psychology, 102(2), pp.215-223.
Sautua, S.I., 2017. Does uncertainty cause inertia in decision making? An experimental
study of the role of regret aversion and indecisiveness. Journal of Economic Behavior
and Organization, 136, pp.1-14.
Sevdalis, N. and Harvey, N., 2007. Biased Forecasting of Postdecisional Affect.
Psychological Science, 18(8), pp.678-681.
Silberg, J. and Manyika, J., 2019. Notes from the AI frontier: Tackling bias in AI (and in
humans). [online] McKinsey Global Institute, Available at:
<https://www.mckinsey.com/featured-insights/artificial-intelligence/tackling-bias-in-
artificial-intelligence-and-in-humans> [Accessed 4 December 2020].
Simonson, I., 1992. The influence of anticipating regret and responsibility on purchase
decisions. Journal of Consumer Research, 19(1), pp.105-118.
Simonson, I. and Tversky, A., 1992. Choice in context: Tradeoff contrast and extremeness
aversion. Journal of Marketing Research, 29(3), pp.281-295.
Simonson, I., Sela, A. and Sood, S., 2017. Preference-construction habits: The case of
extremeness aversion. Journal of the Association for Consumer Research, 2(3),
pp.322-332.
Thaler, R.H., 1980. Toward a Positive Theory of Consumer Choice. Journal of Economic
Behavior and Organization, 1(1), pp.39-60.
Tsiros, M. and Mittal, V., 2000. Regret: A Model of Its Antecedents and Consequences in
Consumer Decision Making. Journal of Consumer Research, 26(4), pp. 401-417.
Van Dijk, E. and Zeelenberg, M., 2005. On the psychology of “if only”: Regret and the
comparison between factual and counterfactual outcomes. Organizational Behavior and
Human Decision Processes, 97(2), pp.152-160.
Zeelenberg, M., Beattie, J., van der Pligt, J. and de Vries, N.K., 1996. Consequences of
Regret Aversion: Effects of Expected Feedback on Risky Decision Making.
Organizational Behavior and Human Decision Processes, 65(2), pp. 148-158.
Zeelenberg, M. and Pieters, R., 2004. Beyond valence in customer dissatisfaction A review
and new findings on behavioral responses to regret and disappointment in failed
services. Journal of Business Research, 57, pp.445-455.
Zeelenberg, M. and Rik, P., 2007. A theory of regret regulation. Journal of Consumer
Psychology, 17(1), pp.3-18. https://doi.org/10.1207/s15327663jcp1701_3.
AE The Integration of Artificial Intelligence in Retail: Benefits, Challenges and a Dedicated Conceptual Framework
120 Amfiteatru Economic
THE INTEGRATION OF ARTIFICIAL INTELLIGENCE IN RETAIL: BENEFITS,
CHALLENGES AND A DEDICATED CONCEPTUAL FRAMEWORK
Ionuț Anica-Popa1*, Liana Anica-Popa 2, Cristina Rădulescu3
and Marinela Vrîncianu4
1)2)3)4) University of Economic Studies, Bucharest, Romania.
Please cite this article as:
Anica-Popa, I., Anica-Popa, L., Rădulescu, C. and
Vrîncianu, M., 2021. The Integration of Artificial
Intelligence in Retail: Benefits, Challenges and a
Dedicated Conceptual Framework. Amfiteatru
Economic, 23(56), pp. 120-136.
DOI: 10.24818/EA/2021/56/120
Article History
Received: 30 September 2020
Revised: 9 November 2020
Accepted: 24 December 2020
Abstract
The aim of this study is to identify the practical benefits and associated risks generated by the
implementation of artificial intelligence (AI) in retail and capitalize on the results by developing
a conceptual framework for integrating AI technologies in the information systems of retail
companies. To this end, a systematic study of recent literature was conducted by carefully
examining the topic of AI implementations. The main results of the documentation were used
to substantiate the conceptual framework introduced by the paper. The research revealed a
variety of advanced solutions, benefits, but also risks that AI generates in retail, in different
segments of the value chain, abbreviated CECoR, from improving the customer experience
(Customer Experience, CE) with the help of virtual agents (chatbots, virtual assistants, etc.), to
cost reductions (Cost, Co) by using smart shelves, and to increasing revenues (Revenue, R) due
to product recommendations and personalized offers or discounts. The proposed conceptual
framework is focused on customer profiles and includes recommendations on AI
implementations in a retail company, from the perspective of CECoR drivers. The results of the
research can be capitalized by practitioners and researchers in the field, who are presented with
concrete examples of benefits, challenges, and risks generated by AI technologies. The CECoR
framework could be a useful tool for both retail and AI specialists, providing common and clear
guidelines for initiating and overseeing projects for integrating AI in a company’s information
systems. Literature-based CECoR analysis dimensions have allowed the restriction of the
research area, which is particularly wide, at the confluence of retail and AI. The originality of
the article lies in the CECoR orientation of the research and the conceptual framework focused
on customer profiling.
Keywords: artificial intelligence, retail, customer experience, cost reduction, revenue
increase, CECoR conceptual framework.
JEL Classification: F43, M15, N70, O33
* Corresponding author, Ionuț Anica-Popa – e-mail: [email protected]
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 121
Introduction
In today's dynamic, super-connected business environment, organizations are forced to use
systems, mechanisms and tools that allow them to obtain a significant to a significant
competitive advantage. With a wide variety of applications, artificial intelligence (AI) is
considered disruptive and revolutionary because it allows “the simulation of human
intelligence, which replaces human beings in complex tasks” (Yang, 2020). Research efforts
target aspects such as natural language recognition and processing, image recognition, object
manipulation, and there are various categories of AI tools: analytical, human-inspired, and
humanized (Kaplan and Haenlein, 2019).
Retail is one of the industries where the number of successful implementations of AI
technologies is constantly increasing. Other new technologies – process automation (Robotic
Process Automation, RPA), Internet of Things (IoT), virtual / augmented reality (VR / AR),
robotic and autonomous vehicles, etc. – also have a significant impact on retail, which is
expected to grow further.
On the other hand, the costs of AI, the issues of reorganizing human resources in the
integration of such technologies, but also the perception that a company's customers and
public opinion have on the specific way in which machines and algorithms manage the huge
amount of personal information collected by companies, represent major challenges for any
organization from retail and require an appropriate treatment in the context of risk
management associated with AI.
To ensure a competitive advantage by adopting emerging technologies, retailers must
consider three key elements: (1) improving the consumer experience, (2) reducing costs, and
(3) increasing revenues and business profitability (Hetu, 2020). The present study aims to
identify and highlight the main benefits and challenges of implementing AI technologies in
retail along the three mentioned axes. The cognitive acquis thus obtained was capitalized by
developing a conceptual framework for integrating AI techniques and algorithms in the
information systems of companies in the retail sector. To this end, efforts have been focused
on the work areas delimited by the following two research questions: (1) “What are, from the
CECoR perspective, the benefits and risks reported by retailers, generated by the AI
implementation?”; (2) “How could the initiation and management of an AI technology
integration project be supported through this research?”.
The research results consist of the practical benefits and risks identified in connection with
the use of AI in retail companies. Conducting the research along the three axes of the CECoR
analysis and automated management of customer profiles at general, individual, and
contextual levels are original approaches, which led to the development of a conceptual
framework that could provide useful guidelines for implementation teams initiating and
running projects aimed at integration of AI with the information systems supporting retail
activities.
The article has three main sections. The first part focuses on the benefits created using AI in
retail, analysed from the perspective of CECoR drivers. The same CECoR axes were used to
investigate the challenges and risks associated with AI implementations. The second section
describes the research methodology, and the third section is dedicated to the presentation of
the research results: CECoR conceptual framework, a scenario illustrating how the CECoR
integration architecture is actually used, along with aspects concerning risks and practical
implementation issues.
AE The Integration of Artificial Intelligence in Retail: Benefits, Challenges and a Dedicated Conceptual Framework
122 Amfiteatru Economic
1. Literature review
1.1. The benefits of Artificial Intelligence in retail
Defined as the ability of a system to acquire and interpret data, learn, and then apply the new
knowledge to achieve certain results or execute a task through adaptive behaviour, AI
includes many subdomains. Among them we can mention Machine Learning (ML), with
supervised, unsupervised, and semi-supervised algorithms for training a software agent, and
Deep Learning (DL), based on artificial neural network techniques that can perform complex
learning tasks (Lee and Shin, 2020; Madurai Elavarasan and Pugazhendhi, 2020). Learning
algorithms (artificial neural networks, Bayesian networks, genetic algorithms, nearest k
neighbours, vector support machines, etc.) use advanced processing capabilities to make
associations, classifications, groupings, and regressions, by analysing large volumes of data.
(Kartal, et al., 2016). Deep Neural Networks (DNNs) combine many machine learning tasks
and leverage other advanced technologies, such as cloud computing, the Internet of Things,
or big data, enabling general purpose machine learning algorithms (GPML) to manage
various data (video, audio, text) and to improve the accuracy of product demand forecasting
by analysing customer behaviour. Moreover, by capitalizing on GPML technologies and
other digital platform features, small retail firms have managed to increase their visibility and
expand their business globally (Meltzer, 2018).
1.1.1. Improving customer experience through Artificial Intelligence technologies
The development of e-commerce and the exponential growth of customer data have
generalized the efforts of companies to identify solutions for predicting customer behaviour
and improving consumer experience (CE). According to Maghraoui and Belghith (2019), the
data shared on the internet in the last decade exceeds those of the entire history of mankind.
As a result, the company that owns, analyses, and harnesses this data properly will gain
differentiation advantage.
According to a study by BearingPoint (2019) there are two broad categories of emerging AI
technologies that help improve the CE: (1) technologies facilitating direct interactions with
customers; (2) technologies allowing for a better treatment of customer demands and
expectations.
Semantic recognition technologies or so-called chatbots improve the CE by providing 24/7
services while offering the great advantage of reducing the volume of low added value
contacts involving human workers. These technologies are currently enjoying increasing
popularity due to messaging applications. Retail companies use chatbots to align online and
offline experiences. For example, in 2016 H&M was ahead of its competitors when it
launched a chatbot on the Canadian messaging application Kik (Prokopiško, 2019), allowing
customers to view, share and purchase products from the H&M catalogue. A personal stylist
service was also provided by the chatbot, which uses photo options and asks questions about
the buyer’s style, creating a style profile for the customer. The supermarket chain Lidl
delivered another example: the conversational chatbot Margot on Facebook Messenger, able
to understand natural language and help buyers get the best product in the range of wines. It
also provides users with tips on combining food with wines and tests their knowledge,
generating a better customer experience.
Voice recognition technologies involve the use of virtual assistants, able to perform various
tasks (taking phone orders, searching for information, sending recommendations to
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 123
customers, etc.). By transforming the voice into text, the customer's voice command is
transmitted to the order taking systems, automatically or via an email. A notable experience
is provided by McDonald’s Corp., which has created an automated and highly complex,
multilingual, and multi-accent ordering tool (Diakantonis, 2019).
Visual recognition technologies are based on virtual assistants able to identify shapes or
people, track the status of product shipments, open accounts remotely, and detect internet
users’ preferences for certain brands. For example, using facial recognition technology, a
retailer could identify a frequent shopper or a loyalty card holder, as soon as they enter the
store. Digital signage is not new in retail, but when combined with facial recognition and Big
Data analysis, it can directly target a specific customer based on his previous buying
behaviour. This means that labels and shelf screens could display customer-oriented
messages or advertisements, drawing attention to certain items or offers that may be of
interest. As far as payment is concerned, biometric and object recognition POS systems allow
for a much-improved CE, the POS solution identifies both the buyer, and the products placed
on the counter (Worley, 2017). Amazon introduced visual search in its main iOS app in 2014,
giving its users the ability to search for a specific product using their smartphone cameras.
Technologies that use autonomous robots allow execution of actions or tasks by adapting
behaviour to the environment. One of the ways retailers use robots in stores is to help
customers find the products they need. In 2016, the LoweBot robot of the retailer Lowe’s
was used in its stores in San Francisco (Underwood, 2020). LoweBot gathers information as
it works, with the goal of identifying the buying patterns for that location to better understand
what goods are moving faster and in what seasons or days of the week. The customers'
curiosity to see them in action allows retail robots to become an important lever in adopting
a brand and improving the CE.
Predictive analytics technologies enable large companies to anticipate future customer
behaviours using past or current behavioural patterns and thus substantiate their strategic
decisions. Predictive analytics can be used for reducing the churn rate of the brand (by
identifying dissatisfied customers) and for detecting risk situations. Relevant examples
include Urban Outfitters, Sephora, and Under Armor, which use an advanced machine
learning engine provided by Dynamic Yield to segment their customers.
According to the results published by Capgemini (2019), 73% of the organizations that
implemented AI technologies noticed an increase of about 10% in customer satisfaction. The
Capgemini investigation shows that 72% of the surveyed organizations experienced a
decrease in the number of customer complaints, and 66% noticed a reduction in the churn
rate of the brand. Another research conducted by Futurum Research (2019) anticipates an
increase in the usage of AI technologies for purchasing experiences. Thus, 65% of consumers
expect to have contact with a chatbot for customer support by 2025, while 81% expect this
to happen by 2030. If 47% of surveyed consumers consider that using an AI assistant, such
as Alexa or Siri, can be a good way to interact in assisting customers, 38% admit that it is not
easy for them to use or adapt to the technology used by merchants and 53% admit that facial
recognition technologies cause them discomfort. According to Servion Global Solutions, a
company cited by Microsoft Corp. (2017), by 2025, 95% of customer interactions will be
made through AI-assisted channels, and chatbot applications will be key components in
assisting the next generation of consumers.
AE The Integration of Artificial Intelligence in Retail: Benefits, Challenges and a Dedicated Conceptual Framework
124 Amfiteatru Economic
1.1.2. Cost-driven savings
Complementary to the other two business drivers, customer experience improvement and
revenue increase, cost reduction should be carefully considered when assessing the impact
of emerging technologies for retail companies. There are some generators for AI-driven cost
savings: effectively reaching the target consumers (Grewal, Roggeveen and Nordfält, 2017),
human workforce reduction (Holmqvist, Van Vaerenbergh and Gronroos, 2017; Inman and
Nikolova, 2017; van Doorn, et al., 2017), and inventory optimization.
Reaching targeted consumers with lower costs. Timing is crucial in retail: delivering
the right message for the right customer, at the right time, could determine a significant
increase in sales. Applying big data technologies, such as predictive analytics, retailers are
able to “estimate” the consumers’ behaviour and adjust their offerings accordingly.
According to Bradlow et al. (2017), the five dimensions of big data for retail are time,
customer, product, location (geo-spatial), and sales channel. Terabytes of new data (e.g., in
case of Walmart or Amazon) are integrated with historical data about millions of customers
and billions of journeys of sold products. AI-backed tools can process this ever-increasing
volume of data with fewer technical requirements, incomparably less time and money than
humans or pre-existing computer systems, while running without errors or interruption,
consequently generating significant cost savings.
Human workforce reduction. Sensors, mobile, and AI technologies provide new
possibilities for cutting down on in-store staff accomplishing “algorithmic” task execution
(Olsen and Tomlin, 2020). Smart shelves, encapsulating meshes of strain sensors,
photodetectors, microphones, and spillage sensor, collect data on product status and send
notifications to the store staff when product quantities on the shelves is below a predefined
value; additionally, real-time inventory AI-based management enables grocery stores to
apply multiple automated price updates for all products that expire on the current date
(Quante, Meyr and Fleischmann, 2009; Inman and Nikolova, 2017). Hence, as smart shelves
are self-managed, there is no need for the store staff to periodically check the stock of
products on shelves, and then to summarize collected data and send it to the person in charge.
Amazon’s approach to intelligent stores involves Amazon Go app, which allows users to
enter a physical store, buy products without scanning them manually, and then leave the store
without staying in line to pay (Amazon, 2020). In a very short description, this process
consists of three steps: (1) when entering the store, the customer must register using the
Amazon Go app from the smartphone; (2) when a product is picked from a shelf (or put
back), shelf sensors detect the move and the software updates the customer’s virtual cart; (3)
when the customer leaves the store, a receipt is generated automatically and the Amazon
account is charged. None of these tasks involves the store staff.
Sensors, automation, robotics, and artificial intelligence help reduce staff costs generated not
only by stores, but also by the warehouses managed by retail companies. Erik Nieves, CEO
of Plus One Robotics, the leader in vision-guided robotics for logistics, maintains that before
it reaches its destination, a package gets touched by an average of 21 peoples. If supported
by an “intelligent” warehouse, the remodelled delivery process will include the following
steps: (1) based on order details, a box is selected and a “route” of shelves is configured; (2)
when the box reaches a shelf that contains some of the products on the order, a robot picks
the products from the shelf and put them into the box; (3) when all products are in the box, it
goes to packing, and then is shipped to the customer.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 125
Inventory optimization. Minimizing the inventory costs – direct costs, represented by
storage cost, and indirect costs, generated by lost sales – is one of the most critical
optimization problems for the retail sector. The quantities of ordered products and the time
of ordering for stock replenishment are highly sensitive decisions to make, as they have a
direct impact on inventory costs and, consequently, on profit maximization (Miller and John,
2010; Mousavi, et al., 2016). This is why inventory optimization emerges as a major use case
for AI implementation by retail companies; for example, machine learning algorithms for
classification (Bayes classifiers, artificial neural networks, and support vector machines) can
be used to predict the ABC classification of inventory (where A is the class with the most,
and C with the least frequently sold items), with high accuracy (Kartal, et al., 2016). Using
machine learning algorithms, Priyadarshi, et al. (2019) also attained the best forecasting
models of the weekly seasonality trend, with least possible errors. In practical terms,
predictive analytics helps retailers determine the optimum daily amount of fresh products to
be supplied, reduce the inventory of perishable food items, minimize the waste by optimally
defining the appropriate amount of produce at required locations and intelligently
synchronize downstream and upstream supply levels.
1.1.3. AI-enhanced revenue growth
Technological trends have led the rapid diffusion of AI applications across the retail field,
with positive results for business revenues, profitability, and efficiency. In this section, we
analyse the impact on revenue drivers of various retailing activities that integrate AI
technologies.
Discussing the integration of AI and ML tools in the sales process specific to the ongoing
fourth industrial revolution, Syam and Sharma (2018) labelled this retail reshaping as “sales
renaissance”. They reported many use cases of AI and ML in sales reinforcement. Among
them, a Harley-Davidson dealership in New York have attained 40 instead of 1 qualified lead
per day and an increase of 2930% in its total number of qualified leads, through AI-based
lead generation algorithms applied for three months. With the help of a ML solution that
leverages dashboards containing pricing variables, qualified leads, IP addresses and other
customers’ data, sales representatives have determined, in real time, the best price for
different segments of their customer base; also, in the post-order stage, a Gainsight system
integrated sales features, customer service and questionnaire results, alerting sales teams
when to invoice and to suggest upsell and cross-sell products etc.
AI technologies may help retailers to consolidate the sales strategy by leveraging existing
stores features (Feng and Fay, 2020). Prices optimization and sales maximization objectives
have led to an increasing use of AI-enhanced big data technologies, detecting correlations
between independent variables such as promoted price, display location, assortment
expansions, and dependent variables like store sales and profitability, brand switching, etc.
(Grewal, Roggeveen and Nordfält, 2017). It was demonstrated that survey-based indicators
like purchase intentions or positive evaluations must be considered to stimulate customers’
engagement and increase revenues. Kumar, Anand and Song (2017) cited by (Grewal,
Roggeveen and Nordfält, 2017) also highlighted the strong relationship between analytics
and retail profitability.
Analysing the impact of AI adoption on online returns policies, Yang, Ji, and Tan (2020)
found that an enhanced personalized virtual experience could lower or even eliminate product
fit uncertainty and improve after–sales services, by deeming consumers’ returns as
AE The Integration of Artificial Intelligence in Retail: Benefits, Challenges and a Dedicated Conceptual Framework
126 Amfiteatru Economic
opportunities to facilitate their continuous searches of exchanges. European retail companies
have begun using AI-based fraud prevention to address the vulnerabilities of both manned
and unmanned checkouts and regain the corresponding revenues (Fujitsu, 2019). Going
further, a smart unstaffed retail shop (SURS) architecture supporting consumer identification
and commodity recognition (Xu, et al., 2020) combines AI and IoT technologies,
engendering customer flow and transactions’ volume increase by 21.7% and 26.8%
respectively, after a month of observation.
Acknowledging the recent development of e-commerce, accelerated by at least five years due
to the current pandemic, specialists have forewarned that retailers failing to leverage new AI
technologies like Augmented Reality/Virtual Reality, IoT, mobility will not be able to
improve overall business performance by addressing the expanding demands of both
consumers and own employees (Durbin, 2020).
1.2. The challenges and risks of using Artificial Intelligence in retail
This section takes a risk management perspective on the three categories of AI benefits
analysed previously, by examining potential challenges in the way of positive outcomes.
AI technological issues. A key AI challenge is ensuring a consistently high quality of
data, considering that data volumes and diversity (web, social media, mobile devices, sensors
and IoT) have increased significantly. For example, in case of ML algorithms, if a population
is underrepresented in the data used to train the model, that model has a built-in bias, a
vulnerability often attributed to AI (Shneiderman, 2016). Moreover, due to algorithm opacity,
ML models are perceived as “black boxes” and it can be difficult to explain how they arrive
at particular predictions or recommendations (Adadi and Berrada, 2018, Miller, 2019).
Building generalized learning techniques represents another important challenge, as AI
systems are still not able to understand contextual details of a business situation and derive
the right meaning from it (Lake, et al., 2017). As they become more complex, AI learning
models create new vulnerabilities, AI systems being prone to unexpected errors and
undetectable attacks. Furthermore, seemingly non-sensitive marketing data feeding AI
systems could also be exploited with potentially disastrous effects, e.g., reputational,
financial, and regulatory consequences for the target companies.
Challenges to AI-driven cost reduction. The financial implications of the research and
development for building AI systems expose such technologies as not easily accessible.
Hence, a first-order challenge for any retailer is setting priorities when deciding where in the
business (e.g. distribution, marketing, customer services, etc.) to deploy a potentially highly
costly AI infrastructure. Since AI shortcomings relative to learning generalization and
contextualization still represent important barriers to the continuous drive to automation, a
more realistic “human in the loop” approach seems to emerge for AI-based processes. As
humans are able to manage situations that require empathy, creativity, and thinking beyond
algorithms, at present and in the near future AI is focused on enhancing human capability,
not on displacing it (Afza and Kumar, 2018). However, AI leads to important shifts in the
nature of the jobs (Huang and Rust, 2018; Makridakis, 2018), which means that important
financial resources must be directed to employee training or retraining and upskilling.
Furthermore, while low and middle-waged jobs are mainly represented by automatable
activities which are likely to decline, high-wage technical jobs related to AI implementation
are expected to grow, adding up to the total AI operating costs.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 127
Challenges to AI-driven revenue increase and enhanced customer experience. The potential increase in retail revenues due to AI implementation is inherently dependent on the quality and quantity of personal information that companies collect from their customers. However, this leads to ethical considerations relative to the right balance between personalization benefits and privacy risks (Inman and Nikolova, 2017). In order to enjoy the advantages of an AI-powered shopping experience, customers are subjected to an increasing pressure to share private information, so they could easily become mistrustful of AI and privacy issues. Collecting and using personal information for marketing purposes has always raised privacy concerns, but these concerns are growing in AI contexts, as the consumer data is being extracted and processed by machines and algorithms. Knowing and forecasting consumers’ demand across interconnected supply chains and then customizing their shopping experience are central AI drivers. However, the expected positive effect on sales and revenues could be moderated by AI technical limitations related to generalized learning (Lake, et al., 2017). For example, basing sales forecasting on past consumer behaviours may perpetuate a bias, e.g. represent a past or random concern no longer relevant to a customer's current buying needs.
AI recommendation systems may lead to sales teams not feeling in control of retail processes and worrying about opportunities that automated systems cannot detect. Furthermore, since AI learning algorithms have difficulties in transferring their experiences from one context to another, companies must allocate considerable resources to train new models for highly similar use cases; for example, in case of a virtual assistant, a customer’s preferences in one area (such as music) should be, ideally, extended to other related domains (movies, books, etc.).
Ethical concerns as moderators for AI benefits. The rapid pace of AI development increases the concerns on how companies deal with ethical issues and creates challenges connected to responsible exploitation of AI systems. Therefore, ethical guidance, adequate policies, and a legal framework to prevent the misuse of AI must be developed and enacted by regulators (Bryson and Winfield, 2017). However, as complex AI platforms often aggregate AI subsystems in the form of software-as-a-service (SaaS) offerings, all risks drivers reviewed here may well originate from third-party solutions, further complicating AI risk management for the beneficiary company; for example, a global retailer consuming a software service in the form of an AI engine must consider regulatory and compliance risks for multiple geographical regions. While regulatory aspects concern legal risks, which are difficult to ignore or circumvent due to potentially serious consequences enforced by law, ethical issues are no less important. Ethical issues are major reputational risk drivers, so it’s important to carefully manage IA aspects in relation to which each retailer is publicly perceived as paying attention not only to business success but also to customer’s and society’s well-being. When not properly managed, reputational risk could lead to dramatic drop in sales and significant increase in costs for restoring customer trust and brand image. An AI concern with ethical implications involves ML algorithms, which are prone to incorporate the biases of their human creators. This could lead to consequences such as discriminatory algorithms and racist or sexist chatbots. For example, the vast majority of digital assistants are portrayed as young Caucasian women that support the stereotyped perception of women in secretarial roles secretară (Spencer, Poggi și Gheerawo, 2018). It is therefore obvious that responsible development and training of AI algorithms is a must for avoiding costly errors. Deployment of algorithms to tailor pricing and promotions increases the need for ethical constructs, as retailers could use AI as a social discrimination vehicle when targeting customers with different messages and prices, based on profiles created about them (Gerlick and Liozu, 2020). From this perspective, AI emerges not only as an effective marketing tool, but also as a powerful manipulation tool. On the other hand, AI-enabled aggressive strategies pose a significant reputational risk that could generate major financial and operational predicaments.
AE The Integration of Artificial Intelligence in Retail: Benefits, Challenges and a Dedicated Conceptual Framework
128 Amfiteatru Economic
2. Research methodology
The aim of this study is to identify practical benefits and associated risks generated by the
implementation of artificial intelligence (AI) in retail and capitalize on the results by
developing a conceptual framework for integrating AI technologies with the information
systems of retail companies.
In the first step, the eligibility conditions of the sources to be studied were defined. The
inclusion criteria set were represented by the year of publication of the research materials
(after 2010), the language of publication (English or French) and the field: artificial
intelligence or retail. Another criterion was represented by the international databases were
the selected papers were indexed: Web of Science, Scopus, Scientific Information Database
or EconLit. Also, to capture the concrete results of the implementation of AI solutions in
retail, the research area included case studies and reports published by representative actors
in this field. Each of the authors independently analysed the identified sources to establish
their eligibility as credible and relevant references.
After selecting the research materials, they were studied to identify a set of elements that
could generate a competitive advantage for retail organizations. The benefits and challenges
of projects for the integration of artificial intelligence in retail have been analysed from the
perspective of this set of elements, considered as pillars of the proposed conceptual
framework. A flowchart of the steps of the research methodology is depicted in Figure no. 1.
Figure no. 1. Research methodology steps
3. Results and discussions
The current section of the paper introduces a conceptual, customer profile-centred
framework, which could be used by retail organizations, to integrate AI techniques and
algorithms with their information systems. The goal of this approach to AI adoption consists
in generating highly and accurately personalized offers for each customer. On a primary
level, the analysis of the advantages enabled by this conceptual framework must take into
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 129
account the same business drivers that were used to investigate AI benefits and risks specific
to the retail sector: improved customer experience (CE), cost reduction (Co), and increase in
sales and revenues (R). The main components of the CECoR framework are described below.
General AI integration architecture. The main structural and functional aspects of this
integration architecture that should be managed by the AI implementation team of the retailer
are briefly described in the paragraphs below, starting with the key subsystems.
Aggregated customer data subsystem manages sales data and internet data. The sales data is
supplied by a data warehouse collecting and organizing data on analysis dimensions which
are typical for retail (e.g. time, customer, product, and location). The internet data is
aggregated from external sources such as social networks and public forums, using web
mining technologies, or it is acquired from various providers; this data must be further
integrated with internet data coming from the organization’s own forum.
Individual customer data subsystem manages historical detailed transactional data on each
customer (completed sales, cancelled orders, updated orders, etc.). This subsystem also
integrates “intelligently” gathered and rapidly processed customer data, resulting from
interactions with agents (chatbots, virtual assistants, digital assistants, conversational agents)
or from feedback and forum opinions.
The third subsystem of the CECoR integration architecture manages inventory data and
information on current promotions for the products on sale.
Figure no. 2 offers a schematic representation of a high-level architecture supporting the AI
integration framework.
Figure no. 2. CECoR framework: high-level AI-based integration architecture
AI-enabled customer profile management. The processing logic of the integration
architecture introduced here relies on the customer profile construct, with the following
specializations: generic customer profile, individual customer profile, and contextual
customer profile. This section elaborates on the rationale of this profile typology and its role
in an AI-enabled retail information system.
AE The Integration of Artificial Intelligence in Retail: Benefits, Challenges and a Dedicated Conceptual Framework
130 Amfiteatru Economic
Generic customer profiles. Despite the aspects that individualize each client, there are certain transactional and behavioural patterns that, when identified and combined with different descriptive elements (gender, age, location, etc.), allow segmentation of company's customers into relatively homogeneous groups. Each of the resulting clusters corresponds to a generic profile that covers the defining elements for a subset of customers, allowing a uniform treatment of interactions with them.
Individual customer profiles. While generic profiles abstract common features of a certain cluster of a company's customers, individual profiles are the result of the reverse process, allowing a generic profile to be adapted to each customer’s particularities. Individual profiles define preferences and buying patterns that cannot be adequately managed in general terms, so they are directed at distinct treatment of interactions with customers sharing the same generic profile. Therefore, an individual profile is obtained by refining and extending the transactional and behavioural attributes of a generic profile, with elements determined by each customer’s identity and transactions.
Contextual customer profiles. They represent particular views on individual profiles, conveying the perception of a specific customer relative to certain elements that the company is planning to use in order to influence the current operational context. For example, contextual profiles could be used to identify the customers most likely to be interested in a new product to be offered for sale, or the customers interested in out of stock or understocked products, to be presented with alternative options. The resulting profiles are temporary, as their relevance is limited to a particular business context, with a specific time frame (e.g. a certain inventory situation, new product launches, social media campaigns for brand support, etc.).
Figure no. 3 shows the AI-enabled sequence of profiles as a gradual transition from generic to particular in relation to a company’s customers, i.e. the derivation of fine-grained profiles out of coarse-grained profiles. The key functions of an AI-enabled customer profile management include: (1) automatic update or reconfiguration of generic and individual profiles, enabled by a continuous flow of new data from internal and external sources; (2) generation of contextual profiles well adapted to situations that justify their use (for example, better configured and targeted promotional campaigns).
Figure no. 3. Customer profile layers
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 131
Main expected AI integration outcomes, by stages and CECoR business drivers. The
integration logic involving the subsystems described above and the corresponding data flows
could be represented as a three-step process. In the first step, AI learning algorithms are
used on sales data, adjusted with internet data, to produce generic customer profiles.
Subsequently, individual customer data is used to fine tune the profile of a specific person.
Once available, the individual profile of a customer is used by the offers’ subsystem to match
it with data on inventory and current promotions; in this final phase, AI techniques and
algorithms are used, once more, to generate contextual customer profiles and corresponding
personalized offers (e.g., product recommendations, discounts, etc.) for each customer.
The business case for the AI integration solution using a layered approach to customer profile
derives from positive effects that could be easily mapped to the same business drivers that
were used to investigate AI impact on the retail: improved customer experience – each
customer feels important and valued, as offers are tailored specifically for his/her needs and
tastes, which are being continuously monitored thanks to dynamically generated customer
data; increased sales revenues – this is a direct consequence of personalized product offerings
and enhanced customer experience; reduced costs – offers rely on real-time inventory data,
which supports cost reduction for both expired products and lost sales.
An explanatory scenario. For a better understanding of the CECoR integration
architecture introduced above, we are going to illustrate its underlying functional logic with
a descriptive example of a hypothetical shopping experience.
On each Wednesday, a customer (John) buys two milk cans from a specific producer (Classic
Milk). In the past few months, customers with similar characteristics as John have given, on
various internet platforms, many positive ratings and feedbacks for milk products of other
two brands (New Milk and Best Milk). Today is Wednesday, and the inventories of milk
from New Milk and Best Milk that will expire tomorrow are quite high, while the milk from
Classic Milk is out of stock. According to the aggregate data from company’s sales and
internet sources, the customer profile corresponding to John indicates preference for Best
Milk (first option) and New Milk (second option). On the other hand, the individual data of
John collected from the company’s forum reveals that his posts on Best Milk contain words
showing he does not like Best Milk. In this case, even if the customer profile based on
aggregate data will place John in the category that prefers Best Milk, the customer profile
adjusted with individual data will include him in the category that prefers New Milk; hence
the system will inform John that he could buy milk from New Milk, with a discount. This is
how the retailer prevents lost sales for milk and avoids having a large quantity of milk that
can no longer be sold past the expiration date. Similarly, from the customer’s perspective,
this is a no less rewarding experience: the customer feels special because the retailer “knows”
that he needs milk, while also getting a discount for a product of a brand he has been tempted
to buy. Moreover, the retailer no longer risks potential negative consequences due to
unpleasant customer experience; for example, if product recommendations will only be based
on aggregated data, as it happens with retail systems currently in use, the offer could be either
ignored by the target customer, or he might very well think “I don’t like Best Milk, why did
they send me this offer?”.
Practical and technical issues. The proposed CECoR AI integration model introduced
by this paper is a highly generic solution that could be used in multiple business contexts.
Naturally, this means that the architecture described here must be further developed to
accommodate the information systems currently in use and the IA technologies of a specific
AE The Integration of Artificial Intelligence in Retail: Benefits, Challenges and a Dedicated Conceptual Framework
132 Amfiteatru Economic
retail company. For example, in case of traditional physical stores, the customer needs to
“register” using the phone when he enters the store; alternatively, he could be automatically
registered by the system, provided that face recognition technology is available in the store.
Moreover, the system must be able to “associate” distinct customers: when a couple goes
shopping, it is necessary that the persons in question are treated as a single customer entity,
with specific characteristics.
Implementation of AI-enriched solutions and the adoption of the CECoR integration
framework presented here and pose multiple technical challenges, as this involves a
technology mix which is not easily accessible to all retail companies: cloud computing, big
data, deep learning, machine learning, neuro-linguistic programming, etc. However,
continuous progress in AI research and development, as well as already proven business
benefits of AI adoption are likely to pave the way for more affordable AI technologies in the
years to come, with significant impact in all industries, including retail.
A brief review of risks. As AI implementations have implications that move beyond the
technical realm of learning algorithms and big data processing, the CECoR conceptual
framework also involves consistent risk management. This requires a top-down view on the
AI context specific to each retailer in order to decide on key principles relative to IT, but also
to business and ethical issues, e.g. what is acceptable or not in terms of use of learning
algorithms for customer profiling, customer privacy, etc. In fact, all ethical concerns that
were reviewed in section 1.2 are of immediate interest for any retail system using AI-driven
customer profiling. As companies collect, track and analyse so much about their customers,
they also have the means to use the profiles to exploit customers’ likes and dislikes and to
manipulate their buying decisions to the extreme extents allowed by formal regulation.
Though apparently justifiable from a business perspective, such practices pose a significant
reputational risk that could jeopardize the very business benefits promised by AI
technologies. In fact, the key business drivers of AI integration investigated by this paper –
improved customer experience, increased sales revenues, cost reduction – could also be
perceived as indicators of effective (or ineffective) AI risk management.
Conclusions
While the fourth industrial revolution is in full swing, the huge wave of technological changes
is pushing companies to adapt quickly to remain competitive. The current research
contributes to the support of retail organizations nowadays when artificial intelligence seems
to become a pervasive and slowly inserted enhancement in almost every commercial activity.
The authors took a cross-disciplinary approach, bringing into the research field a triadic
contribution.
Firstly, in order to help retail specialists looking to adopt AI in their organisations have real
insights, an analysis of AI benefits identified in current practice was performed, structured
on the business drivers provided by Hetu (2020): customer experience (CE) enhancement,
cost (Co) decrease and revenue (R) growth. Among other findings, we have observed that
there are many AI emerging tools with positive influences on not only one, but two or even
all three CECoR drivers. For example, the query-based AI systems, like Macy’s On Call,
Alexa on Amazon’s Echo, Cortana on Microsoft, or Siri on the Apple phone, in both online
and physical stores, answer customers’ questions about specific goods, provide suggestions
on possible combination with complementary products, or offer directions about where to
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 133
find the goods within a store (Grewal, Roggeveen and Nordfält, 2017), Thus, the business
impact of those AI systems is significant, manifesting itself simultaneously on all three
CECoR levels: customer experience improvement, sales volume increase and to labour costs
cut. On smart shelves with integrated digital tags, it was noted it can perform remote price
updates almost instantly (Inman and Nikolova, 2017), not needing human intervention; as a
result, both staff and stock keeping costs savings and increased customer satisfaction are
possible.
Secondly, in relation to the CECoR pillars, the risks that practitioners associate with the AI
implementation in retail were revealed. Based on the performed analysis, it can be stated that
the positive impact of AI implementation depends on an efficient management of risks
associated with this type of technologies. Particular attention should be paid to ethical issues,
such as the possibility of manipulating customers through AI. The abusive exploitation of
data can cause contrary effects to those initially expected, at the time of implementation of
AI solutions for capturing and processing this data. For example, one of the risks that can
cause major negative effects is reputational risk, especially important in the context of the
specific way the CECoR framework approached customer profiles definition.
The third contribution of the paper is the CECoR conceptual framework, designed to enable
implementation teams to align AI initiatives with business priorities aiming at fully
leveraging these improvement opportunities and obtaining a meaningful impact and
significant competitive advantages for their companies. The elaborated framework was
substantiated by capitalizing on the CECoR cognitive acquis, its application having two
important purposes: refining customer profiles and optimizing the personalization of offers.
This dichotomous vision – CECoR oriented and customer profiles-focused – differentiates
this article from previous studies in the same area of research.
Though it could be seen a research limitation, confining the set of analysed resources by
using methodological criteria to filter scientifically validated and recent papers makes this
study relevant and up-to-date, while guiding the research efforts towards the conceptual
framework introduced here.
The pragmatic analysis of the digital change generated by AI in retail and the elaborated
conceptual framework can be documentary resources for futures studies on possible
sustainable implementations of AI in retail. Further developments in this research could also
aim at deepening and expanding the application scenarios of the CECoR framework, as well
as analysing trends in the maturation of artificial intelligence technologies and the effects on
retail and other areas of activity.
References
Adadi, A. and Berrada, M., 2018. Peeking inside the black-box: A survey on explainable
artificial intelligence (XAI). IEEE Access, 6. [e-journal] pp.52138-52160.
10.1109/ACCESS.2018.2870052.
Amazon, 2020. Got questions? We have answers. [online] Available at:
<https://www.amazon.com/b?ie=UTF8&node=16008589011> [Accessed 10 September
2020].
BearingPoint, 2019. L’expérience client à l’ère de l'intelligence artificielle. [online]
Available at: <https://www.bearingpoint.com/files/L_exp%C3% A9rience_client_
AE The Integration of Artificial Intelligence in Retail: Benefits, Challenges and a Dedicated Conceptual Framework
134 Amfiteatru Economic
%C3%A0_l_%C3%A8re_de_l_intelligence_artificielle.pdf?download=0&itemId=5715
34> [Accessed 10 September 2020].
Bradlow, E.T., Gangwar, M., Kopalle, P. and Voleti, S., 2017. The Role of Big Data and
Predictive Analytics in Retailing. Journal of Retailing, 93(1), pp.79-95.
Bryson, J. and Winfield, A., 2017. Standardizing ethical design for artificial intelligence and
autonomous systems. Computer, [e-journal] 50(5), pp.116-119. 10.1109/MC.2017.154.
Capgemini, 2019. Impact of AI for Customer Experience (CX) AI: re-humanizing digital
customer experience. [online] Available at: <https://www.capgemini.com/fr-fr/wp-
content/uploads/sites/2/2019/06/Point-of-view_Impact-of-AI-for-CX_Final-1.pdf>
[Accessed 27 November 2020].
Diakantonis, D., 2019. Retail Tech M&A #5: Voice recognition gives retailers more ways
to communicate. Mergers&Acquisitions Magazine, [online] Available at:
<https://www.themiddlemarket.com/news/retail-tech-m-a-5-voice-recognition-gives-
retailers-more-ways-to-communicate> [Accessed 19 September 2020].
Durbin, E., 2020. Top emerging technologies transforming the retail experience. Retail
Customer Experience, [online] Available at: <https://www.retailcustomerexperience.
com/resources/top-emerging-technologies-transforming-the-retail-experience-2/>
[Accessed 10 September 2020].
Feng, C. and Fay, S., 2020. Store Closings and Retailer Profitability: A Contingency
Perspective. Journal of Retailing, [e-journal] 96(3), pp.411-433.
10.1016/ j.jretai.2020.01.002.
Fujitsu, 2019. Customer Case Study. [online] Available at: <https://www.fujitsu.com/
global/imagesgig5/CS_2019Mar_Supermarket-Retailer_Eng_v1.pdf> [Accessed 18
September 2020].
Futurum Research, 2019. Experience 2030: The Future of Customer Experience EMEA.
[online] Available at: <https://www.sas.com/content/dam/SAS/documents/marketing-
whitepapers-ebooks/third-party-whitepapers/en/futurum-experience-2030-emea-
110977.pdf> [Accessed 29 November 2020].
Gerlick, J.A. and Liozu, S.M., 2020. Ethical and legal considerations of artificial intelligence
and algorithmic decision-making in personalized pricing. Journal of Revenue and Pricing
Management, [e-journal] 19(2), pp.85-98. 10.1057/s41272-019-00225-2.
Grewal, D., Roggeveen, A.L. and Nordfält, J., 2017. The Future of Retailing, Journal of
Retailing, [e-journal] 93(1), pp.1-6. 10.1016/j.jretai.2016.12.008.
Hetu, R., 2020. Retail Digital Transformation and Innovation Primer for 2020. [online]
Gartner. Available at: <https://emtemp.gcom.cloud/ngw/globalassets/en/
informationtechnology/documents/insights/retail_digital_trans_2020_primer.pdf>
[Accessed 10 September 2020].
Holmqvist, J., Van Vaerenbergh, Y. and Gronroos, C., 2017. Language use in services:
Recent advances and directions for future research. Journal of Business Research,
72(March 2017), pp.114-118.
Huang, M.H. and Rust, R.T., 2018. Artificial Intelligence in Service. Journal of Service
Research, 21(2), pp.155-172.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 135
Inman, J.J. and Nikolova, H., 2017. Shopper-Facing Retail Technology: A Retailer Adoption
Decision Framework Incorporating Shopper Attitudes and Privacy Concerns. Journal of
Retailing, [e-journal] 93(1), pp.7-28. 10.1016/j.jretai.2016.12.006.
Kaplan, A. and Haenlein, M., 2019. Siri, Siri, in my hand: Who’s the fairest in the land? On
the interpretations, illustrations, and implications of artificial intelligence. Business
Horizons, [e-journal] 62(1), pp.15-25. 10.1016/j.bushor.2018.08.004.
Kartal, H., Oztekin, A., Gunasekaran, A. and Cebi, F., 2016. An integrated decision analytic
framework of machine learning with multi-criteria decision making for multi-attribute
inventory classification. Computers & Industrial Engineering, 101, pp.599-613.
Kumar, V., Anand, A. and Song, H., 2017. Future of Retailer Profitability: An Organizing
Framework. Journal of Retailing, 93(1), pp.97-120.
Lake, B.M., Ullman, T.D., Tenenbaum, J.B. and Gershman, S.J., 2017. Building machines
that learn and think like people. Behavioral and Brain Sciences, [e-journal] 40, pp.e253.
10.1017/S0140525X16001837.
Lee, I. and Shin, Y.J., 2020. Machine learning for enterprises: Applications, algorithm
selection, and challenges. Business Horizons, [e-journal] 63(2), pp.157-170.
10.1016/j.bushor.2019.10.005.
Madurai Elavarasan, R. and Pugazhendhi R., 2020. Restructured society and environment:
A review on potential technological strategies to control the COVID-19 pandemic. The
Science of the Total Environment, [e-journal]. 10.1016/j.scitotenv.2020.138858.
Maghraoui, S. and Belghith, E., 2019. L’expérience-client: quels apports des technologies
de l’Intelligence Artificielle. International Journal of Economics & Strategic
Management of Business Process (ESMB), 15, pp.7-14.
Makridakis, S., 2018. Forecasting the impact of artificial intelligence, Part 3 of 4: The
potential effects of IA on businesses, manufacturing, and commerce. Foresight: The
International Journal of Applied Forecasting, issue 49, pp.18-27.
Meltzer, J., 2018. The impact of artificial intelligence on international trade. [online]
Brookings Institution. Available at: <https://www.brookings.edu/research/the-impact-of-
artificial-intelligence-on-international-trade/> [Accessed 2 September 2020].
Microsoft Corp., 2017. Powering the Future of the Customer Experience.
[online] Available at: <https://news.microsoft.com/europe/features/ai-powering-
customer-experience/> [Accessed 28 November 2020].
Miller, S. and John, R., 2010. An Interval Type-2 Fuzzy multiple echelon supply chain
model. Knowledge-Based Systems, 23(4), pp.363-368.
Miller, T., 2019. Explanation in artificial intelligence: Insights from the social sciences.
Artificial Intelligence, [e-journal] 267(February 2019), pp.1-38.
10.1016/j.artint.2018.07.007.
Mousavi, S.M., Sadeghi, J., Niaki, S.T.A. and Tavana, M., 2016. A bi-objective inventory
optimization model under inflation and discount using tuned Pareto-based algorithms:
NSGA-II, NRGA, and MOPSO. Applied Soft Computing, 43(6), pp.57-72.
Olsen, T.L. and Tomlin, B., 2020. Industry 4.0: Opportunities and Challenges for Operations
Management. Manufacturing & Service Operations Management, 22(1), pp.113-122.
AE The Integration of Artificial Intelligence in Retail: Benefits, Challenges and a Dedicated Conceptual Framework
136 Amfiteatru Economic
Pegasystems Inc., 2017. Artificial Intelligence and Improving the Customer Experience.
[online] Available at: <https://www.pega.com/system/files/resources/2019-09/ai-and-
improving-cx-en.pdf> [Accessed 28 November 2020].
Priyadarshi, R., Panigrahi, A., Routroy, S. and Garg, G.K., 2019. Demand forecasting at retail
stage for selected vegetables: a performance analysis. Journal of Modelling in
Management, [e-journal] 14(4), pp.1042-1063. 10.1108/JM2-11-2018-0192.
Prokopiško, A., 2019. How Chatbots are transforming retail in 2019. Chatbots life, [online]
Available at: <https://chatbotslife.com/how-chatbots-are-transorming-retail-in-2019-
d98ac 2f13b73> [Accessed 11 August 2020].
Quante, R., Meyr, H. and Fleischmann, M., 2009. Revenue management and demand
fulfillment: matching applications, models, and software. OR Spectrum, 31(1), pp.31-62.
Shneiderman, B., 2016. Opinion: The dangers of faulty, biased, or malicious algorithms
requires independent oversight. Proceedings of the National Academy of Sciences, [e-
journal] 113(48), pp.13538-13540. 10.1073/pnas.1618211113.
Spencer, J., Poggi., J. and Gheerawo, R., 2018. Designing out stereotypes in artificial
intelligence: Involving users in the personality design of a digital assistant. Proceedings
of the 4th EAI international conference on smart objects and technologies for social good,
pp.130-135.
Syam, N. and Sharma, A., 2018. Waiting for a sales renaissance in the fourth industrial
revolution: Machine learning and artificial intelligence in sales research and practice.
Industrial Marketing Management, 69(February 2018), pp.135-146.
Underwood, C., 2020. Robots in Retail – Examples of Real Industry Applications. [online]
Available at: <https://emerj.com/ai-sector-overviews/robots-in-retail-examples/>
[Accessed 13 September 2020].
van Doorn, J., Mende, M., Noble, S.M., Hulland, J., Ostrom, A.L., Grewal, D. and Petersen,
A.J., 2017. Domo Arigato Mr. Roboto: The Emergence of Automated Social Presence in
Customers’ Service Experiences. Journal of Services Research, 20(1), pp.43-58.
Worley, M., 2017. NRF 2018: Five Trends We Think Will Transform Brick and Mortar
Again. [online] Available at: <https://nectoday.com/tag/retail-customer-experience/>
[Accessed 2 September 2020].
Xu, J., Hu, Z., Zou, Z., Zou, J. and Hu, X., 2020. A Design of Smart Unstaffed Retail Shop
Based on IoT and Artificial Intelligence. IEEE Access, [e-journal] 8(August 2020),
pp.147728-147737. 10.1109/ACCESS.2020.3014047.
Yang, G., Ji, G. and Tan, K., 2020. Impact of artificial intelligence adoption on online returns
policies. Annals of Operations Research, [e-journal] April 2020. 10.1007/s10479-020-
03602-y.
Yang, L.B., 2020. Application of artificial intelligence in electrical automation control,
Procedia Computer Science, [e-journal] 166, pp.292-295. 10.1016/j.procs. 2020.02.097.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 137
THE IMPACT OF ARTIFICIAL INTELLIGENCE USE
ON E-COMMERCE IN ROMANIA
Adrian Micu1, Angela-Eliza Micu2, Marius Geru3, Alexandru Căpățînă4
and Mihaela-Carmen Muntean5 1) 4) 5) Dunarea de Jos University, Galati, România
2)Ovidius University, Constanța, România 3)Transilvania University, Brașov, România
Please cite this article as:
Micu, A., Micu, A.E., Geru, M., Căpățînă, A. and
Muntean, M.C., 2021. The Impact of Artificial
Intelligence Use on the E-Commerce in Romania.
Amfiteatru Economic, 23(56), pp. 137-154.
DOI: 10.24818/EA/2021/56/137
Article History
Received: 30 September 2020
Revised: 3 November 2020
Accepted: 21 December 2020
Abstract This study aims at identifying the tools used in e-commerce, able to optimize marketing
campaigns. Managerial and marketing processes have been identified in the relevant body
of knowledge that can be optimized using artificial intelligence; thus, a questionnaire has
been designed within a quantitative research. The sample used in the research consists of
201 persons having managerial positions, who are involved in e-commerce, their
companies’ have been their company active in 2020, with at least one employee. The article
highlights the managerial tools used in promoting products in the online environment and
business processes that they want to optimize using artificial intelligence. At the same time,
for the quantitative study, three hypotheses have been tested to identify the motivation to
buy online, as well as the methods used by online store managers in the communication
process. The limitations of this study are determined by the fact that only the managerial
perspective is analysed, without considering the perception of the final consumer, which
could have ethical implications. Optimizing the flow of stocks and logistics processes will
be the subject of future research considering that it is the main challenge for management,
as the quantitative research proved.
Keywords: e-commerce, m-commerce, artificial intelligence, marketing automation.
JEL Classification: M31, M15, O33
Corresponding Author, Alexandru Căpățînă – e-mail: [email protected]
AE The Impact of Artificial Intelligence Use on the E-Commerce in Romania
138 Amfiteatru Economic
Introduction
In the context of accelerated digitization, the role of online marketing in a company's
strategy is expanding significantly, as proved by the increase in companies' investments
within e-commerce activities. About a quarter of total marketing budgets (26%) are used by
companies whose only business model is online commerce (Alvarez, 2013). In addition to
the profitability and changes in customer behaviour, investments in digitization are argued
primarily by their results, but also by the fact that they are easier to measure compared to
those of traditional marketing (Pickton, 2005).
As customers interact more and more with companies through digital channels and social
networks, marketers have recognized the need to track these interactions and measure their
performances (Chaffey and Patron, 2012). To this end, companies need to adopt web
analytics tools that help collecting, measuring, analyzing and reporting data to web visitors,
in order to understand and optimize the use of e-commerce platforms, also called Big Data.
Today, online marketing is an essential branch of e-commerce and includes different ways
of promoting a company, such as email marketing, content marketing, social media,
affiliate marketing and other marketing strategies. The diversity of content sharing channels
on the Internet and the way it is presented, requires marketers to consider the location and
how their customers communicate. Thus, new opportunities are noticed in Mobile
Commerce (m-commerce), which becomes a valuable solution for improving e-commerce,
which allows users to interact with merchants, anywhere and anytime. It is found that over
a third of buyers have made at least one purchase through a smart mobile device in the last
6 months (Alvarez, 2013). Kang et al. (2015) highlighted the increased use of mobile
communications, and the fast growth of mobile shopping technologies has considerable
financial potential, especially for retail transactions between businesses and consumers.
The main objective of this research is to highlight the role of management teams from e-
commerce companies to automate processes and streamline data flows through predictive
analytical platforms based on artificial intelligence algorithms. The specific objectives aim
at testing the correlations between the intention of managers to automate certain marketing
processes through artificial intelligence algorithms and the intent to identify customer
satisfaction, respectively the use of a Customer Relationship Management (CRM)
application.
Starting from the general concepts related to e-commerce outlined in the literature review
and presented in the first section of this article, Section 2 addresses the transition to e-
commerce on mobile devices (m-commerce). Based on the trends of e-commerce in
Romania discussed in Section 3, a quantitative research is conducted to analyse the
perceptions of online store managers on artificial intelligence algorithms in Sections 4 and
5, the findings being presented in Section 6. Discussions of the results are highlighted in
Section 7, and the last section illustrates the research conclusions.
1. Digital Commerce - general concept
E-commerce is increasingly used by traditional companies looking to expand in the online
environment, as well as by companies that were established due to the benefits of the
Internet, especially for B2C digital commerce (Chaparro-Peláez, Agudo-Peregrina and
Pascual-Miguel, 2016). Taylor (2019) estimates that the growth rate of online commerce
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 139
adoption will reach an average of about 25% by 2026. The pillars for these predictions are
given by the growth rates of over 30% found in 2015 and 2018 (Eurostat, 2018; Monnot, et
al., 2019), being estimated similar values taking into account the context of the COVID-19
pandemic caused by coronavirus.
Although e-commerce has been promoted in developed countries from Western Europe and
America, the developing regions of Eastern Europe and Asia have seen an exceptional
expansion of the digital commerce industry in the last ten years. Countries like India, China
and Singapore have developed a real culture for online business. Of these, China is the
largest power in the digital market, with the largest volume of online sales in the world.
China's progress in this area is largely due to Alibaba's digital commerce company, which
monopolizes the local market (26.6%) and provides opportunities for local business
development by marketing products online worldwide (Zhang, et al., 2020). Although the
digital market offers many benefits, the lack of consumer confidence in the relationship
with online merchants has been a major inhibitor to access the global market (Srinivasan
and Barker, 2012). Buyers' trust in online companies is influenced by factors such as
integrity, transparency, quality, location, ease of use of the online trading platform,
confirmed by McKnight, Choudhury and Kacmar (2002). These factors become even more
evident in the context in which consumer attention is divided between more and more
screens of different sizes. One of the most researched areas of online marketing is the
buying behaviour of users depending on the device used, areas of interest, searches and
pages visited to make online purchases. Many companies have focused their efforts on
identifying the probability that a visitor will complete a transaction, turning these tools into
business models (Bucklin and Sismeiro, 2009).
By the end of 2017, Google and Facebook became the two largest online companies
occupying over 60% of the online advertising market in the US. E-commerce giants, such
as Amazon and Alibaba, are experiencing exponential growth using smart promotional
tools that exploit consumer behaviour on social media and search engines. The more and
more varied these companies offer products, the more likely they are to buy for visitors and
buyers, resulting in a faster flow of stock. Moreover, platforms such as Amazon have begun
to develop these mechanisms for understanding the behaviour of internal users, better
optimizing this flow and integrating it into their own systems such as Kindle, Amazon
Prime Video and Alexa (Ritala, Golnam and Wegmann, 2014). Thus, in the digital trade, an
economy has developed that favours the largest companies to the detriment of the small
ones that have too little data to make relevant recommendations to customers. Other similar
examples are Uber, the largest company in the taxi industry, eBay, which is the leader in
online auctions C2C and Airbnb, which dominates the hotel market (Tadelis, 2015).
Despite this obstacle that small businesses face, the low cost of entering the digital market
causes independent traders to operate in a niche where competition is low and which they can
effectively exploit (Moriset, 2020). An alternative for entrepreneurs is to use new
communication channels, which allow a similar analysis of consumer behaviour, which is
currently expanding. The progress of mobile devices has resulted in the emergence of mobile
commerce (m-commerce) (Zheng, et al., 2019). Unlike digital commerce, mobile commerce
has advantages such as instantaneity, ubiquity, location, customization and identification
(Wang, Ngamsiriudom and Hsieh, 2015). However, mobile digital transactions face the refusal
of users to become customers due to fears and anxieties regarding the use of the smartphones in
placing online orders (Jaradat, Moustafa and Al-Mashaqba, 2018).
AE The Impact of Artificial Intelligence Use on the E-Commerce in Romania
140 Amfiteatru Economic
A recently published study (Barnes, 2020) shows that during the spring 2020 lockdown due
to the COVID-19 pandemic, online shopping increased by 207% in April alone, as
consumers sought to buy products through available online channels. At the same time,
there was a huge pressure at the level of online retailers globally, subject to massive
delivery requests in the shortest possible deadlines (Bhatti, et al., 2020).
2. Online commerce on mobile devices (m-commerce)
Mobile targeting can increase the stake perceived by the consumer in online shopping,
which influences consumers' perceptions and purchasing intentions. The hedonic value of
shopping reflects the values that consumers receive from the multisensory, fantastic, and
emotional aspects of their shopping experiences, such as entertainment and pleasure
(Hirschman and Holbrook, 1982). With a mobile targeting feature, consumers can access
multiple channels for similar products and services, which can usually lead to more
informed decision-making and more attractive offers. As smartphones’ penetration rate, app
adoption and mobile browsing continue to grow, companies are increasingly leveraging the
power of mobile devices to drive online sales.
As the use of applications continues to grow, m-commerce will make a major contribution
to increasing sales, especially with the help of young people called Millennials and
Generation Z who have significant purchasing power. These technologically equipped
consumers are endowed with the ability to increase the sales volume because they are more
likely to shop on their smartphones.
3. The influence of electronic commerce on the Romanian economy
In 2019, in Romania, the e-commerce sector reached a share of 4.3 billion euros, compared
to 3.6 billion euros in the previous year, 2018, with an increase of 22%. On average, in
2019, the average value of the shopping cart made from desktop was 273 lei, compared to
204 lei in 2018. The average value per basket made from mobile increased from 170 lei in
2018 to 208 lei in next, to all these values being added VAT (Radu, 2020).
More empirical research on the Romanian e-commerce market is very necessary because
the e-commerce industry has grown amazingly. The main advantages brought by the e-
commerce activities to the Romanian companies indicate a higher productivity and an
increased efficiency, followed by the superior positioning compared to the competitors.
According to the study conducted by Kulcsar and Teglas (2017), web stores in Romania
must formulate marketing policies and strategies depending on the development regions,
because there is a significant relationship between ordered products, respectively the
average value of orders and online customer residence.
The development of technology, artificial intelligence, augmented reality, and virtual reality
will help consumers meet their needs in a simpler way (Enache, 2018). Since they have less
time to shop in a physical store, they will prefer to buy products available online.
Consumers will choose to buy from online retailers who sell various types of products and
who ensure the transparency of product information, such as price and distribution, in order
to be satisfied with the purchase made (Table no. 1).
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 141
Table no. 1. The evolution of electronic commerce in Romania between 2015-2018 Electronic commerce issues 2015 2016 2017 2018
Romania's population (millions of people) 19.8 19.7 19.6 19.5
Internet users (millions of people) 11 11.2 11.2 11.7
Internet penetration rate (%) 56 58 58 70
Smartphone penetration rate (%) 31.6 38.8 46 52.5
Orders placed online from mobile devices (%) 25-30 35-40 45 54
The value of online shopping (e-tail) (billion €) 1.4 1.8 2.8 3.5
Average value of online shopping / day (€ million) 3.8 4.9 7.6 9.8
Average number of transactions / day 8.2 8.4 8.7 9
Black Friday sales (€ million) 100 130 200 250
Online payment by card (million €) 514 745 980 1300
Source: Data processed by authors based on reports available on www.gpec.ro
and www.statista.com
Even if approximately 13.6 million bank cards are active in Romania (Dospinescu,
Anastasiei and Dospinescu, 2019), only 17% chose to use them for online payments, the
others preferring cash on delivery (Pavel, 2019). In 2018, e-commerce covered 8% of the
total value of the Romanian retailing market (Pavel, 2019), but for the next period it is
estimated that these values will increase taking into account the degree of adoption of smart
mobile devices. Despite limitations related to small screen sizes and processing power, the
growth of commerce on mobile devices is more accelerated reaching 3.79 billion lei in
2019 (Pantelimon, Georgescu and Posedaru, 2020). International e-commerce retailers,
with subsidiaries in Romania, have recognized the importance and necessity of their online
presence on various social platforms, developing in this sense a sustained activity in the
cyberspace. Online retailers should always try to innovate and change the features and
performance of products and services so that they can best meet customer expectations and
preferences (Dabija, Bejan and Tipi, 2018).
4. Artificial intelligence algorithms in e-commerce
The increased attention paid by researchers to the field of e-commerce proves the
awareness of digital transformation importance (Rogers, 2016) and the impact on
sustainable development, being obvious the continuous quest to identify new methods to
optimize processes using artificial intelligence algorithms. These two emerging paradigms,
e-commerce and artificial intelligence, can have an impact on social standards of interaction
between consumers and retailers, but also on public policies that govern and regulate the
legal framework in which these actors operate (Vanneschi, et al., 2018).
The fast growth of e-commerce has stimulated applications for extracting and
understanding the data generated by customers' financial transactions. Currently, the
exploitation of financial data has been one of the most important research topics in the data
mining community, which has resulted in a considerable workload for the academic
community (Akter and Wamba, 2016).
E-commerce communities have spawned a substantial amount of consumer-generated
information, including online product or vendor reviews, online transaction assessments,
and industry-specific metrics using crowdsourcing applications. Most of the time, this
AE The Impact of Artificial Intelligence Use on the E-Commerce in Romania
142 Amfiteatru Economic
information is used for a continuous development of e-commerce from a technical
perspective, abandoning the marketing-centric approach. The flows of user-generated
content reflect full acceptance of Web 2.0 key features, such as user-centric design and
information sharing. User-generated content, usually product reviews, can be exploited
through econometric models and data extraction analysis, to better understand consumer
behaviour (Zoghbi, Vulić and Moens, 2016). Marketers need reliable tools to interact with
real-time data to obtain actionable information. Most of the time, these software tools have
only a one-way approach, being used as support for the end customer, without focusing on
what data is relevant or what marketing analyses can be leveraged in customer interaction
(Kakatkar and Spann, 2019).
Developments in deep learning algorithms and accelerated innovation in digital image
recognition (computer vision) have the potential to endow managers and marketers with the
ability to improve the performance of e-commerce and digital marketing campaigns. In
practice, however, the application of deep learning and digital image recognition as a subset
of artificial intelligence has been limited due to technical challenges (e.g. accuracy and
reliability) to be used in the managerial decision process. These challenges are a result of
the dynamic and complex nature of marketing and the challenges related to acquiring data
on consumer behavior. Before making any business decision, Business Intelligence is a
necessary and essential tool for marketing managers. Business Intelligence (BI) is a set of
techniques for analyzing big data and presenting information in a way that is actionable to
management teams (Massaro, et al., 2019).
The techniques based on artificial intelligence, such as decision tree (DT), vector support
machine (SVM), neural network (NN) and deep learning, are used to make decisions about e-
commerce campaigns. They are frequently used by online retailers’ engines that train
algorithms in systems based on artificial intelligence (Lu, et al., 2018). As the effectiveness of
value chains for e-commerce depends on the transparency and use of data across IT systems,
researchers are encouraged to continue to use big data in the cloud to develop tools that help
practitioners make better and faster decisions in B2B and B2C environments, through artificial
intelligence applications and innovations, data mining, machine learning tools, as well as the
integration of blockchain technology, especially in the areas of e-commerce order management
process and predicting online shopping behaviour (Leung, et al., 2019).
The marketing manager has access to a wide range of digital tools, which allows him to
better understand marketing trends and gather specific information on how he can improve
new products. However, compared to data collected from customer marketing surveys, Big
Data (BD) obtained from data mining process within consumer behaviour in the online
environment has contrasting characteristics. For example, a large volume of online reviews
is displayed on e-commerce sites, such as Amazon.com or emag.ro, without a specific
aggregate meaning. On these websites, customers are encouraged to share their views on
previously purchased products only for human validation and content indexing in search
engines. E-commerce platforms are not the only places where consumers' preferences can
be observed, data can also be found on social networking sites such as Twitter.com, on
review sites such as tripadvisor.com, on media sites such as Cnet.com. While researchers'
attention has been focused on using deep learning and computer vision to monitor
consumer behaviour in physical stores, there is a wealth of data available, as online text
content can be turned into intelligence actionable by marketers (O'Mahony, et al., 2019). In
addition, as new tools and devices allow customers to create more versatile content in the
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 143
form of images, text, audio and video, search options increase even more, as long as there
are appropriate tools for interpreting all this information.
In the field of computer science, machine learning, which is a subset of artificial
intelligence (Jordan and Mitchell, 2015), has been widely applied in areas such as natural
language processing (NLP), speech recognition (Agarwalla and Sarma, 2016) and computer
vision. Conventional machine learning approaches are limited in their ability to process
data in their raw form (LeCun, Bengio and Hinton, 2015). The inability to process data is
due to the fact that a considerable amount of knowledge in programming and fields is
required for the design of a feature extractor (LeCun, Bengio and Hinton, 2015). But in
2018, in a project funded by Amazon, Shin and Choi (2015) managed to create a
nomenclature that automatically extracts and indexes their specific products and attributes.
Machine learning refers to the techniques needed to work intelligently with a large amount
of data, by developing useful algorithms in synthesizing, classifying or sorting this data. It
would be impossible for a human user to perform search activities in a short time, as the
Google search engine succeeds. This is where machine learning comes into play, an integral
part of artificial intelligence (Ballestar, Grau-Carles and Sainz, 2019).
Deep learning is a representation method that can be used to automatically extract
sophisticated features at high levels of abstraction. This method can also learn from data
with multiple levels of end-to-end representations (LeCun, Bengio and Hinton, 2015). By
combining deep learning methods (neural networks) with computer visualization, specific
elements or products from images can be extracted and used for managerial decisions or
marketing strategies. A particular type of deep learning method that has been widely used is
CNN (Convolutional Neural Network), which has managed to outperform other neural
networks in terms of image classification (Krizhevsky, Nair and Hinton, 2020), object
detection, and their targeting. In-depth learning has allowed the development of
applications based on computer vision, for example, in the case of autonomous vehicles and
in the automatic diagnosis of some forms of cancer (Mehta and Shah, 2016), but also for e-
commerce operations (Koehn, Lessmann and Schaal, 2020).
5. Research method
The analysis was based on 201 computer-assisted telephone interviews, representing
approximately 9% of all managers and people with managerial position of online stores in
Romania (2329). Initially, a campaign to complete the questionnaire by e-mail was
launched, but the success rate was extremely low. The contact information database for
online stores was collected from two sources:
contact details from the website listafirme.ro - Companies have been identified that
have the NACE code 4791 - retail through the order houses or through the Internet.
Because the contact details of the companies are public data, they are not subject to GDPR
according to regulation no. 679 of April 27, 2016.
contact details obtained from the contact pages of online stores - A database was
created with online stores indexed in price comparators, websites that manage affiliate
marketing campaigns and events dedicated to e-commerce.
This analysis has identified the functions of artificial intelligence-based e-commerce
platforms that management teams considers compulsory and intend to develop in the future
AE The Impact of Artificial Intelligence Use on the E-Commerce in Romania
144 Amfiteatru Economic
in e-commerce businesses. The differences they encounter in the behaviour of customers on
mobile devices compared to desktop ones were also taken into account, thus formulating
the following hypotheses:
H01 (null hypothesis) – There is no correlation between management's intention to
automate certain marketing processes and the desire to identify customer satisfaction.
Ha1 (alternative hypothesis) – There is a correlation between management's intention to
automate certain marketing processes and the desire to identify customer satisfaction.
H02 (null hypothesis) – There is no statistical correlation between the managerial intention
to automate flows in marketing processes and the use of a customer relationship
management application
Ha2 (alternative hypothesis) – The managerial intention to automate flows in marketing
processes is statistically correlated with owning a CRM application.
H03 – There is no statistical correlation between the use of artificial intelligence algorithms
in online store management processes and the position held by the respondent in the
company
Ha3 – There is a statistical correlation between the use of artificial intelligence algorithms in
online store management processes and the position held by the respondent in the company.
The Chi-Square test is used in the statistical analysis due to the fact that categorical data were
collected through the interview. For this research, a significance level of 95% (p <0.05) was
selected, the results below this level being accepted in the academic field as a valid scientific
result or implicit truth. However, Amrhein, Greenland and McShane (2019) proposed a more
specific way of communication through which all statistical reporting should implicitly
include the p-value - the statistical probability used in the test. The hypotheses testing in this
empirical study was performed using the SPSS software, version 21.0.14.
6. Findings
The study reveals that many digital online store managers use artificial intelligence
algorithms in tools provided by third-party companies to promote their products. Table 2
shows how social media promotion campaigns are the most used (169 out of 201) by
Romanian online store managers and less than a quarter of them (n=40) use artificial
intelligence algorithms available on the platform.
Table no. 2. Promotional tools that use artificial intelligence in e-commerce
Digital campaigns types Used Use the AI option of the platform
Google Adwords campaigns 141 33
Search engine optimization 83 11
Social media campaigns 169 40
Content marketing 66 7
Marketing newsletter 77 12
Direct marketing 63 4
Influencer marketing 36 1
Conversion rate optimization solutions 14 3
I don't know / I don't answer / Not applicable 3 3
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 145
The second most used Digital Marketing tool that uses artificial intelligence algorithms is
Google Adwords (Table 2), the platform that allows merchants to display paid ads when
customers search for certain products online. Moreover, Google Adwords allows the
placement of banners and advertisements on the Google Display Network on approximately
60% of all online websites. This also becomes very valuable in the process of collecting
data to map and calibrate consumers' buying intentions for certain products, using artificial
intelligence algorithms (Figure no. 1).
Figure no. 1. The most popular tools for promoting products online
Regardless of the platform for promoting digital products and stores, the objective of
management teams is focused on automating processes and streamlining the data flows to
engage the customer to buy more. Marketing is only one of the activities that management
teams carry out in managing the activity of an online store, but others can be automated
(Table 3).
Table no. 3. Processes that can be automated in online commerce using AI algorithms
It has been
already automated
It requires
automation
I don’t know /
I don’t answer
Product management 103 50 11
Delivery 95 44 14
Refund operations 81 46 12
Accounting report 93 46 10
Marketing campaigns 77 47 10
Relationship with
suppliers 66 36 11
Customer support 72 44 10
Obviously, considering the large number of products, the biggest challenge is to manage
the products, delivery and refund operations. It can be seen in Figure 2 that in certain
processes, for example the relationship with suppliers, management does not prove a high
level of automation and is not interested in addressing this issue.
Online commerce on mobile devices is gaining attention among Romanian consumers, the
big players like emag.ro, altex.ro and aboutyou.ro use aggressive campaigns to encourage
customers to download and use their own applications. For example, customers who use
mobile apps have priority access to Black Friday or public holiday promotional campaigns
AE The Impact of Artificial Intelligence Use on the E-Commerce in Romania
146 Amfiteatru Economic
and receive real-time notifications about the status of their placed order. Management teams
have access to essential data about customers and their behaviour online and offline (e.g.
location) and have also a communication channel open all the time with them for the
submission of promotional offers and product recommendations. Another emerging
technology that can be framed as an alternative to mobile applications is represented by
PWA (progressive web app) solutions, which allow accessing the solution in the browser
but also installing the application, if the customer wishes (Figure no. 3).
Figure no. 2. Analysis of automated processes in e-commerce
Figure no. 3. Online stores managing their own application
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 147
The high cost of developing a mobile app makes this solution extremely unpopular among
online store managers, only 33 confirmed that they already have an application and 9 are in
the process of developing one. Mobile applications have much higher user engagement
rates than online store websites, generating significantly higher conversion rates.
Because this study was conducted in the period May-July 2020, one of the questions
addressed to management representatives of online stores was related to how the COVID-19
pandemic affected online sales. Figure 4 shows how the number of sales during the
COVID-19 pandemic had a noticeable increase for companies whose activity in the online
environment is below 25%, between 50 and 75%, respectively over 75% of yearly turnover.
Figure no. 4. The effect of COVID-19 on online sales
Currently, CRM (Customer Relationship Management) systems in e-commerce platforms
are the main tools used in automating marketing flows. These systems were available on
the market long before e-commerce platforms and the Internet as we know it today. Their
main purpose is to monitor the relationship with customers and monitor the sales process.
From 201 respondents, 57 use dedicated customer relationship management (CRM)
solutions, the remaining 144 using only the standard functions integrated in the e-
commerce platform. Solutions developed specifically for e-commerce platforms are
currently available on the market, but to the open-ended question What CRM system do you
use? addressed on the telephone, no respondent named a specific e-commerce platform. All
managers use platforms that have telesales or B2B sales as their first utility, not being
completely adapted to the e-commerce sales process. Most likely, this is the point where the
confirmation of the following hypothesis starts:
Ha1 – The intention of managers to automate flows in marketing processes is statistically
correlated with owning a CRM application.
The null hypothesis is rejected with p value=0.032 and Phi=0.185, the results are presented
in Table 4.
AE The Impact of Artificial Intelligence Use on the E-Commerce in Romania
148 Amfiteatru Economic
Table no. 4. First hypothesis Test Results Statistical indicator Value Degrees of freedom Asymptotic significance
Pearson Chi-Square 6.883 2 0.032
Phi 0.185 2 0.032
Cramer's V 0.185 2 0.032
Number of cases validated 201
Source: Data processed by the authors through the statistical software SPSS
Ha2 – There is a statistical correlation between the management of online stores that intend
to automate marketing processes using artificial intelligence and the use of questionnaires
to identify customer satisfaction.
From 201 respondents who accepted the interview by phone, 77 believe that they are
currently automating marketing processes, 57 intend to do so in the near future and
67 believe that there is no need to automate. Most of them (40) use machine learning tools
in managing social media campaigns (an option that was introduced 2 years ago), followed
by 33 managers using the same option on Google Adwords (3-4 years expertise) and email
marketing (12). For the use of machine learning algorithms on Google Adwords, a
statistical correlation was identified with the intention of marketing process automation
managers (p=0.034) and users who conducted marketing campaigns on Google Adwords
(p=0.001). However, this correlation does not have a significant managerial implication.
The most unexpected result of this test is the fact that managers who are interested in
automating certain marketing processes are exactly the ones who check customer
satisfaction through marketing questionnaires. The hypothesis was validated with
p value=0.041 and Phi=0.179 (Table no. 5).
Table no. 5. Second hypothesis Test Results
Statistical indicator Value Degrees of
freedom
Asymptotic
significance
Pearson Chi-Square 6.407 2 0.041
Phi 0.179 2 0.041
Cramer's V 0.179 2 0.041
Number of cases validated 201
Source: Data processed by the authors through the statistical software SPSS
Ha3 – There is a statistical correlation between the use of artificial intelligence algorithms in
online store management processes and the position held by the respondent in the company.
From 201 respondents, 50 are employed in Middle Management positions, of which only
6 use systems based on artificial intelligence algorithms, 116 are entrepreneurs, of which
43 systems based on artificial intelligence algorithms, 11 are managers, of which 7 systems
based on artificial intelligence algorithms, and 24 are top managers, of which 11 systems
based on artificial intelligence algorithms.
The cross-tabulation analysis shows that entrepreneurs, directors and top managers are
more interested in using artificial intelligence algorithms than are employees from Middle
Management. This is statistically validated by the p value=0.001 (Table 6).
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 149
Table no. 6. Second hypothesis Test Results Statistical indicator Value Degrees of freedom Asymptotic significance
Pearson Chi-Square 17.201 3 0.001
Phi 0.293 3 0.001
Cramer's V 0.293 3 0.001
Number of cases validated 201
Source: Data processed by the authors through the statistical software SPSS
The level of Cramer’s correlation V=0.293 is much higher than other tests performed in this
research, so a moderate correlation can be emphasized.
7. Discussion on findings
Considering the development and involvement of artificial intelligence in the e-commerce
industry, it can be anticipated that by the end of 2021, approximately 90% of customer
interactions through e-commerce portals will be treated and managed without people, the
role of chatbots becoming crucial (Soni, 2020).
Most studies focused on the benefits of integrating artificial intelligence algorithms into e-
commerce platforms have not considered the managerial perspective on capitalizing on
these advantages. Thus, a study by Wei, Huang and Fu (2007) reveals that personal referral
systems can not only reduce search time for interesting articles, but have the ability to
improve e-commerce portal sales by converting visitors into actual buyers, increasing
cross-selling, and consumer loyalty. Our study contributes with the results regarding the
perception of the investigated managers on the processes that can be automated in online
commerce using artificial intelligence algorithms.
Evaluating key performance indicators is an excellent technique for monitoring an e-
commerce website that connects consumers to regularly updated offers. Artificial
intelligence algorithms allow e-commerce store managers to choose key performance
indicators appropriate to the internal processes of online sales portals (Ahmed, et al., 2017).
In addition to this finding, our study demonstrates the relevance of CRM systems that
integrate artificial intelligence algorithms into the vision of online store managers.
The results of the research conducted by Ballestar, Grau-Carles and Sainz (2019) offers
companies a predictive model (based on machine learning algorithms) to customize the
economic incentives for each referral according to the quality of the leads it brings to a
company involved in e-commerce operations, thus optimizing marketing investments. Our
study also presents the typology of online marketing campaigns that may involve similar
machine learning algorithms.
Various machine learning libraries are available to review the anticipated value of each
parameter in an e-commerce system, being extremely useful to online store managers, who
will enter the platform where artificial intelligence algorithms are integrated to generate
scenarios on online sales. Consistent with this finding reflected in the study by Wang, Cai
and Zhao (2020), the results of our research highlight the high interest of online store
managers in Romania towards the predictive value of e-commerce systems based on
artificial intelligence.
AE The Impact of Artificial Intelligence Use on the E-Commerce in Romania
150 Amfiteatru Economic
In a pandemic context such as COVID-19, high levels of perceived concerns motivate
consumers to rely more on e-commerce portals for transactional activities that have a high
potential to build confidence and reduce risks (Tran, 2020). Our study confirms through its
results the role of e-commerce platforms in facilitating communication between online
retailers and buyers based on strengthening trust and credibility, in the context of the
COVID-19 pandemic.
Recognizing the innovative valences of artificial intelligence in e-commerce, the results of
this study add value to the body of knowledge through the managerial vision of how
artificial intelligence algorithms provide e-commerce portals predictive capabilities,
especially in CRM systems. Thus, the results of our study align with the main contributions
of research coordinated by Soni (2020), proving that one of the most significant and
common uses of artificial intelligence in e-commerce is that it can help estimate sales,
helping experts analyze huge volumes. customer data so that they can obtain useful and
appropriate information for strategic, tactical and operational decisions.
Conclusions
This article develops the theoretical framework based on previous research on e-commerce
and the use of artificial intelligence algorithms, the theoretical contributions being useful to
managers and researchers alike, in designing strategies for digital business transformation,
given the new challenges of digital age, where access to data is extremely easy, and
actionable managerial information establishes the market leader. In e-commerce, the
consumer can compare and analyse products and services without costs related to their
transport in a short time. The same can be done by competitors who can establish their
pricing strategies based on the market offer or can identify high-performance products in
the digital catalogues of similar stores. Artificial intelligence can be a useful tool in
identifying these opportunities and capturing them in marketing campaigns. However,
understanding these algorithms and their efficient use is a managerial challenge, which
reveals that less than a quarter of online store managers use or have used such tools. The
value of this article for the academic community is given by understanding managerial
needs in a new economic branch, digital commerce and how modern technologies that use
artificial intelligence can be used to streamline internal processes.
By accepting the third alternative hypothesis, this study confirms that the interest in using
artificial intelligence algorithms in e-commerce arises from top management level. This
finding has significant implications in competitive areas where management must perform
consistently. At the same time, it is the management team responsibility to have control
over the key processes in the business and to constantly streamline production costs. This
study was largely focused on how artificial intelligence can be used in marketing
campaigns and validates by testing the second alternative hypothesis that management's
interest in these algorithms stems from the need to maintain high customer satisfaction. The
deepest managerial implication for intelligent e-commerce algorithms is the management of
online sales forecasting processes and the logistics of packages and products. An important
issue that online store managers need to consider is the proper training of staff in the
direction of recent technological advances in the field of artificial intelligence, their
ongoing skill development in this regard. Online store managers should focus on ways they
can raise awareness of the role of artificial intelligence systems in e-commerce platforms,
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 151
given that research has found that this is the key to discover the way for predictive
strategies in analysing customer behaviour in the online environment.
The descriptive analysis also identifies other processes in e-commerce that could be
optimized in terms of internal workflows, but in this study, they are not deepened using
statistical studies. Inevitably the use of data collected in the use of machine learning
algorithm training has an impact on the perception of employees and customers who
interact with e-commerce platforms. These privacy policy considerations were not
addressed at this time, but according to the European rules on the protection of personal
data - GDPR, managerial teams are responsible for the management of this data.
In the future research agenda, it is important to address methods of storage optimization and
logistics of goods. E-commerce completely changes the sale-purchase process and the process
innovations that management is currently looking for can certainly be identified. A future
configurational study based on the method of Qualitative-Comparative Analysis (QCA) will
focus on causal recipes of factors that lead to capitalizing on the benefits of applying artificial
intelligence algorithms in an online store. The lack of fully understanding these concepts
limits management in integrating such tools. The easier evaluation of an online store benefits
helps to increase the degree of adoption and can lead to the development of web software
solutions that can be integrated into the e-commerce platform.
Acknowledgement
This work was supported by a grant of the Ministry of Education and Research from
Romania, CCCDI – UEFISCDI, project number PN-III-P1-1.2-PCCDI-2017-0800/
86PCCDI/2018, project name: FUTUREWEB, within PNCDI III.
References
Agarwalla, S. and Sarma, K.K., 2016. Machine learning based sample extraction for
automatic speech recognition using dialectal Assamese speech. Neural Networks, 78,
pp.97-111. https://doi.org/10.1016/j.neunet.2015.12.010.
Ahmed, H., Jilani, T.A., Haider, W., Abbasi, M.A., Nand, S. and Kamran, S., 2017.
Establishing standard rules for choosing best KPIs for an e-commerce business based on
google analytics and machine learning technique. International Journal of Advanced
Computer Science and Applications, 8(5), pp.12-24.
Akter, S. and Wamba, S.F., 2016. Big data analytics in E-commerce: a systematic review
and agenda for future research. Electronic Markets, 26(2), pp.173-194.
Alvarez, G., 2013. Hype Cycle for E-Commerce 2013, [online] Gartner Inc. Available at:
<https://www.gartner.com/en/documents/2571916/hype-cycle-for-e-commerce-2013>
[Accessed 20 September 2020].
Amrhein, V., Greenland, S. and McShane, B., 2019. Scientists rise up against statistical
significance. Nature, 567, pp.305-307.
Ballestar, M.T., Grau-Carles, P. and Sainz, J., 2019. Predicting customer quality in e-
commerce social networks: a machine learning approach. Review of Managerial
Science, 13(3), pp.589-603.
AE The Impact of Artificial Intelligence Use on the E-Commerce in Romania
152 Amfiteatru Economic
Barnes, S.J., 2020. Information management research and practice in the post-COVID-19
world. International Journal of Information Management, 55, pp.102175. https://doi.org/10.1016/j.ijinfomgt.2020.102175.
Bhatti, A., Akram, H., Basit, H.M., Khan, A.U., Raza, S.M. and Naqvi, M.B., 2020. E-
commerce trends during COVID-19 Pandemic. International Journal of Future
Generation Communication and Networking, 13(2), pp.1449-1452.
Bucklin, R.E. and Sismeiro, C., 2009. Click Here for Internet Insight: Advances in
Clickstream Data Analysis in Marketing. Journal of Interactive Marketing, 23(1),
pp.35-48.
Chaffey, D. and Patron, M., 2012. From web analytics to digital marketing optimization:
Increasing the commercial value of digital analytics. Journal of Direct, Data and
Digital Marketing Practice, 14(1), pp. 30-45. https://doi.org/10.1057/dddmp.2012.20.
Chaparro-Peláez, J., Agudo-Peregrina, Á.F. and Pascual-Miguel, F.J., 2016. Conjoint
analysis of drivers and inhibitors of e-commerce adoption. Journal of Business
Research, 69(4), pp.1277-1282.
Dabija, D.C., Bejan, B.M. and Tipi, N., 2018. Generation X versus millennials
communication behaviour on social media when purchasing food versus tourist
services. Economics and Management, 21(1), pp.191-205.
Dospinescu, O., Anastasiei, B. and Dospinescu, N., 2019. Key factors determining the
expected benefit of customers when using bank cards: An analysis on millennials and
generation Z in Romania. Symmetry, 11(12), pp.1449-1469.
Enache, M.C., 2018. E-commerce Trends. Annals of the University Dunarea de Jos of
Galati: Fascicle: I, Economics & Applied Informatics, 24(2), pp. 67-71.
Eurostat, 2018. SDG 12 - Responsible consumption and production, [online] Eurostat.
Available at: <https://ec.europa.eu/international-partnerships/sdg/responsible-
consumption-and-production_en> [Accessed 24 September 2020].
Hirschman, E.C. and Holbrook, M.B., 1982. Hedonic consumption: emerging concepts,
methods and propositions. Journal of Marketing, 46(3), pp. 92-101.
Jaradat, M.I.R.M., Moustafa, A.A. and Al-Mashaqba, A.M., 2018. Exploring perceived
risk, perceived trust, perceived quality and the innovative characteristics in the adoption
of smart government services in Jordan. International Journal of Mobile
Communications, 16(4), pp.399-439.
Jordan, M.I. and Mitchell, T.M., 2015. Machine learning: Trends, perspectives, and
prospects. Science, 349(6245), pp.255-260.
Kakatkar, C. and Spann, M., 2019. Marketing analytics using anonymized and fragmented
tracking data. International Journal of Research in Marketing, 36(1), pp.117-136.
Kang, M., Gao, Y., Wang, T. and Wang, M., 2015. The Role of Switching Costs in O2O
Platforms: Antecedents and Consequences. International Journal of Smart Home, 9(3),
pp.135-150.
Koehn, D., Lessmann, S. and Schaal, M., 2020. Predicting online shopping behaviour from
clickstream data using deep learning. Expert Systems with Applications, 50, pp.113342.
DOI: 10.1016/j.eswa.2020.113342.
Krizhevsky, A., Nair, V. and Hinton, G., 2020. The CIFAR-10 dataset, [online] Available
at: <https://www.cs.toronto.edu/~kriz/cifar.html> [Accessed 20 September 2020].
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 153
Kulcsár, E. and Téglás, S., 2017. In the maze of e-commerce. Online trade defining
variables in Romania. Management & Marketing Journal, 15(1), pp.124-138.
LeCun, Y., Bengio, Y. and Hinton, G., 2015. Deep learning. Nature, 521(7553), pp.436-444.
Leung, K.H., Luk, C.C., Choy, K.L., Lam, H.Y. and Lee, C.K., 2019. A B2B flexible
pricing decision support system for managing the request for quotation process under e-
commerce business environment. International Journal of Production Research, 57(20),
pp.6528-6551.
Lu, H., Li, Y., Chen, M., Kim, H. and Serikawa, S., 2018. Brain intelligence: go beyond
artificial intelligence. Mobile Networks and Applications, 23(2), pp.368-375.
Massaro, A., Vitti, V., Lisco, P., Galiano, A. and Savino, N., 2019. A business intelligence
platform Implemented in a big data system embedding data mining: a case of study.
International Journal of Data Mining & Knowledge Management Process, 9(1), pp.1-20.
McKnight, D.H., Choudhury, V. and Kacmar, C., 2002. Developing and validating trust
measures for e-commerce: An integrative typology. Information Systems Research,
13(3), pp.334-359.
Mehta, P. and Shah, B., 2016. Review on techniques and steps of computer aided skin
cancer diagnosis. Procedia Computer Science, 85, pp.309-316.
Monnot, E., Reniou, F., Parguel, B. and Elgaaied-Gambier, L., 2019. “Thinking outside the
packaging box”: should brands consider store shelf context when eliminating
overpackaging? Journal of Business Ethics, 154(2), pp. 355-370.
Moriset, B., 2020. e-Business and e-Commerce. International Encyclopedia of Human
Geography, pp. 1-10.
O’Mahony, N., Campbell, S., Carvalho, A., Harapanahalli, S., Hernandez, G.V.,
Krpalkova, L. and Walsh, J., 2019. Deep learning vs. traditional computer vision. In:
Science and Information Conference (pp. 128-144). Springer, Cham. DOI:10.1007/978-
3-030-17795-9.
Pantelimon, F.V., Georgescu, T.M. and Posedaru, B.Ş., 2020. The Impact of Mobile e-
Commerce on GDP: A Comparative Analysis between Romania and Germany and how
Covid-19 Influences the e-Commerce Activity Worldwide. Informatica Economica,
24(2), pp. 27-41.
Pavel, S., 2019. Raportul Oficial al Pieței de E-Commerce din România GPeC 2018,
[online] Available at: <https://www.gpec.ro/blog/raportul-pietei-ecommerce-gpec-2018-
romanii-au-facut-cumparaturi-online-de-peste-3-5-miliarde-euro-in-2018> [Accessed
27 November 2020].
Pickton, D., Broderick, A., 2005. Integrated Marketing Communications, 2nd ed. Harlow:
Prentice Hall/Financial Times.
Radu, A., 2020. E-Commerce, 2019. Raport Gpec Romania, [online] Available at:
<https://www.gpec.ro/blog/raport-gpec-e-commerce-romania-2019> [Accessed 20
September 2020].
Ritala, P., Golnam, A. and Wegmann, A., 2014. Coopetition-based business models: The
case of Amazon.com. Industrial Marketing Management, 43(2), pp. 236-249.
Rogers, D.L., 2016. The digital transformation playbook: Rethink your business for the
digital age. New York: Columbia Business School Pub.
AE The Impact of Artificial Intelligence Use on the E-Commerce in Romania
154 Amfiteatru Economic
Shin, D.H. and Choi, M.J., 2015. Ecological views of big data: Perspectives and issues.
Telematics and Informatics, 32(2), pp. 311-320.
Soni, V.D., 2020. Emerging Roles of Artificial Intelligence in ecommerce. International
Journal of Trend in Scientific Research and Development, 4(5), pp.223-225.
Srinivasan, S. and Barker, R., 2012. Global analysis of security and trust perceptions in web
design for e-commerce. International Journal of Information Security and Privacy
(IJISP), 6(1), pp.1-13.
Tadelis, S., 2015. The economics of reputation and feedback systems in e-commerce
marketplaces. IEEE Internet Computing, 20(1), pp.12-19.
Taylor, K., 2019. The retail apocalypse is far from over as analysts predict 75,000 more
store closures, [online] Available at: <https://www.businessinsider.com/retail-
apocalypse-thousands-store-closures-predicted-2019-4> [Accessed 20 September 2020].
Tran, L.T.T., 2020. Managing the effectiveness of e-commerce platforms in a pandemic.
Journal of Retailing and Consumer Services, 58, pp.102287.
Vanneschi, L., Horn, D.M., Castelli, M. and Popovič, A., 2018. An artificial intelligence
system for predicting customer default in e-commerce. Expert Systems with
Applications, 104, pp.1-21. https://doi.org/10.1016/j.eswa.2018.03.025.
Wang, Q., Cai, R. and Zhao, M., 2020. E-commerce Brand Marketing based on FPGA and
Machine Learning. Microprocessors and Microsystems, pp.103446.
Wang, S.W., Ngamsiriudom, W. and Hsieh, C.H., 2015. Trust disposition, trust
antecedents, trust, and behavioral intention. The Service Industries Journal, 35(10),
pp.555-572.
Wei, K., Huang, J. and Fu, S., 2007. A survey of e-commerce recommender systems. In:
2007 International Conference on Service Systems and Service Management. China,
Editors: Tien and Berg, Chengdu.
Zhang, X., Zhou, G., Cao, J. and Wu, A., 2020. Evolving strategies of e-commerce and
express delivery enterprises with public supervision. Research in Transportation
Economics, 80(C), pp.100810. DOI: 10.1016/j.retrec.2019.100810.
Zheng, X., Men, J., Yang, F. and Gong, X., 2019. Understanding impulse buying in mobile
commerce: An investigation into hedonic and utilitarian browsing. International
Journal of Information Management, 48, pp. 151-160. DOI: 10.1016/
j.ijinfomgt.2019.02.010.
Zoghbi, S., Vulić, I. and Moens, M.F., 2016. Latent Dirichlet allocation for linking user-
generated content and e-commerce data. Information Sciences, 367-368, pp.573-599. DOI:10.1016/j.ins.2016.05.047.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 155
CONSUMER ACCEPTANCE OF THE USE OF ARTIFICIAL INTELLIGENCE
IN ONLINE SHOPPING: EVIDENCE FROM HUNGARY
Szabolcs Nagy1* and Noémi Hajdú2 1)2) University of Miskolc, Miskolc, Hungary
Please cite this article as:
Nagy, S. and Hadjú, N., 2021. Consumer Acceptance
of the Use of Artificial Intelligence in Online
Shopping: Evidence From Hungary. Amfiteatru
Economic, 23(56), pp.155-173.
DOI: 10.24818/EA/2021/56/155
Article History
Received: 30 September 2020
Revised: 7 November 2020
Accepted: 26 December 2020
Abstract
The rapid development of technology has drastically changed the way consumers do their
shopping. The volume of global online commerce has significantly been increasing partly
due to the recent COVID-19 crisis that has accelerated the expansion of e-commerce.
A growing number of webshops integrate Artificial Intelligence (AI), state-of-the-art
technology into their stores to improve customer experience, satisfaction and loyalty.
However, little research has been done to verify the process of how consumers adopt and
use AI-powered webshops. Using the technology acceptance model (TAM) as a theoretical
background, this study addresses the question of trust and consumer acceptance of
Artificial Intelligence in online retail. An online survey in Hungary was conducted to build
a database of 439 respondents for this study. To analyse data, structural equation modelling
(SEM) was used. After the respecification of the initial theoretical model, a nested model,
which was also based on TAM, was developed and tested. The widely used TAM was
found to be a suitable theoretical model for investigating consumer acceptance of the use of
Artificial Intelligence in online shopping. Trust was found to be one of the key factors
influencing consumer attitudes towards Artificial Intelligence. Perceived usefulness as the
other key factor in attitudes and behavioural intention was found to be more important than
the perceived ease of use. These findings offer valuable implications for webshop owners to
increase customer acceptance.
Keywords: consumer acceptance, artificial intelligence, online shopping, AI-powered
webshops, technology acceptance model, trust, perceived usefulness, perceived ease of use,
attitudes, behavioural intention, Hungary
JEL Classification: L81, M31, O30
* Corresponding author, Szabolcs Nagy – e-mail: [email protected]
AE Consumer Acceptance of the Use of Artificial Intelligence in Online Shopping: Evidence From Hungary
156 Amfiteatru Economic
Introduction
The rapid development of digital technology has changed online shopping (Daley, 2018). In
recent years, the use of Artificial Intelligence (AI) in online commerce has been increased since
AI is an excellent tool to meet rapidly changing consumer demand and to increase sales
efficiency. The global spending by retailers on AI services is expected to quadruple and reach
$12 billion by 2023, and over 325000 retailers will adopt AI technology (Maynard, 2019).
Smidt and Power (2020) claimed that online product research has significantly increased
over the past years. USA's largest online retailer, Amazon, is the exemplary case of how
to effectively integrate AI into online retail. Besides the rich assortment, fast delivery
and competitive prices, a more localised shopping journey can be created. Thus Amazon
can use location-specific pricing and send destination-specific messages to its
customers, who will pay in their local currency (Barmada, 2020).
Novel marketing techniques supported by new technologies, including the use of AI systems
spark the proliferation of new marketing methods to effectively reach target consumers and to
offer enhanced consumer experiences (Pusztahelyi, 2020). Pursuant to Asling (2017), the use
of AI in online shopping makes customer-centric search and a new level of personalisation
possible resulting in a more efficient sales process. Information technology (IT) has changed
the nature of company-customer relationships (Rust and Huang, 2014). However, any
technology-driven transformation is based on trust (Pricewaterhouse Coopers, 2018).
Online retailers need more in-depth insight into how consumers perceive and accept the use of
AI in webshops and how much they trust them. They also need to know how to use AI most
effectively to increase online spending and online purchase frequency since the importance of
time and cost efficiency in shopping has recently become more and more critical. In this
regard, online shopping means a convenient way for customers to buy the desired products.
So far, only a few researchers have addressed the question of trust and consumer
acceptance of AI in online retail. Based on the technology acceptance model (TAM), this
study aims to fill this research gap and proposes an integrated theoretical framework of
consumers' acceptance of AI-powered webshops. Further objectives of this paper are to
investigate the relationships between the elements of TAM; to analyse the effects of trust,
perceived usefulness and perceived ease of use on attitudes and behavioural intention.
After reviewing the use of AI in online shopping, this paper discusses the role of trust in
online shopping and presents the technology acceptance model. The next section deals with
the research methodology, including the research questions, hypotheses and the sample. In
the results and discussion section, the validity and reliability of the model, as well as the
model fit are presented. Hypothesis testing, detailed analysis of the relationships between
the elements of the nested model, and comparison of the results with the previous research
findings are also discussed here before the conclusions sections.
1. Literature review
According to IBM's U.S. Retail Index, the COVID-19 has speeded up the change from
traditional shopping to online purchasing by circa five years (Haller, Lee and Cheung,
2020). Due to the pandemic situation, there is an increased demand for AI in the retail
industry (Meticulous Market Research, 2020).
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 157
1.1. The use of AI in online shopping
AI systems are a set of software and hardware that can be used to continuously assess and
analyse data to characterise environmental factors and to determine decisions and actions
(European Commission, 2018). Prior research mainly focused on the advantages of the use
of AI in online settings and failed to address how consumers accept AI in online retail.
According to utility theory, this new technology helps consumers to find and choose the
best product alternatives, while decreases the search cost and search time (Pricewaterhouse
Coopers, 2018), thus increasing utility (Stigler, 1961; Bakos, 1977; Stigler and Becker,
1977; André, et al. 2017; Lynch and Ariely, 2000). AI filters the information for each target
customer and provides what exactly is needed (Paschen, Wilson and Ferreira, 2020). AI
supports automating business processes, gains insight through data analysis, and engages
with customers and employees (Davenport and Ronanki, 2018).
Artificial intelligence is widely used to increase the efficiency of marketing (Kwong, Jiang, and
Luo, 2016) and retail (Weber and Schütte, 2019) and to automate marketing (Dumitriu and
Popescu, 2020). AI-powered online stores provide their customers with automated assistance
during the consumer journey (Yoo, Lee and Park, 2010; Pantano and Pizzi, 2020). It is a great
advantage, especially for the elder people, who are averse to technical innovations.
Consumers' online information search and product selection habits can be better understood by
AI to offer a more personalised shopping route (Rust and Huang, 2014). It is a great
opportunity for online shops to analyse the profile of existing and potential customers and
thereby suggest tailor-made marketing offerings for them (Onete, Constantinescu and Filip,
2008). AI also makes the contact with both the customers and the employees continuous and
interactive. Frequently asked questions (FAQs) regarding the products, product-use and
ordering process can be automated by a chatbot. New sales models use automated algorithms
to recommend unique, personalised marketing offerings, thus increasing customer satisfaction
and engagement. To sum up the advantages, AI systems operate automatically and analyse big
data in real-time to interpret and shape consumer behavioural patterns to offer products and
services in a personalised way, thus enhancing the shopping experience.
However, AI systems also have some disadvantages. They work most effectively with big data;
therefore, the implementation of AI systems requires huge investments (Roetzer, 2017).
1.2. The role of trust in online shopping
Trust is of great importance in online commerce. According to Kim, Ferrin and Rao (2008),
consumer confidence has a positive effect on a consumer's intention to buy. The higher the
consumer trust in an online shop is, the more likely the consumer will be to go through the
buying process. Trust is especially crucial when the customer perceives a financial risk.
Thatcher et al. (2013) identified two types of trust: general and specific trust. General trust
concerns the e-commerce environment, consumer beliefs about and attitudes towards it.
Specific trust is related to the shopping experience in a specific virtual store. Confidence can be
enhanced through interactive communication between the retailer and the buyer by using
appropriate product descriptions and images to reduce the perceived risk. As stated in Cătoiu et
al. (2014) there is a strong negative correlation between perceived risks and trust. According to
Reichheld and Schefter (2000, p. 107), “price does not rule the Web; trust does”.
AE Consumer Acceptance of the Use of Artificial Intelligence in Online Shopping: Evidence From Hungary
158 Amfiteatru Economic
Aranyossy and Magisztrák (2016) found that a higher level of e-commerce trust was
associated with more frequent online shopping. However, when shopping online, customers
do not necessarily notice that a website uses AI tools (Daley, 2018).
All things considered, AI marks a new era in online sales. However, continuous
technological development such as the use of AI-powered websites divides society, as there
are those who accept novelty while others reject it.
1.3. Technology Acceptance Model (TAM)
Consumers' adaptation to new technologies can be explained by several models. Dhagarra,
Goswami and Kumar (2020) summarised them as follows: (1) Theory of Reasoned Action
(TRA) by Fishbein and Ajzen (1975); (2) Theory of Planned Behaviour (TPB) by Ajzen
(1985); (3) Technology Acceptance Model (TAM) by Davis (1986); (4) Innovation
Diffusion Theory (IDT) by Rajagopal (2002); (5) Technology Readiness Index (TRI) by
Parasuraman, (2000); and (6) Unified Theory of Acceptance and Use of Technology
(UTAUT) by Venkatesh, et al. (2003).
Technology acceptance model (TAM), an extension of (TRA), is one of the most widely-
used theoretical models (Venkatesh, 2000) to explain why an IT user accepts or rejects
information technology and to predict IT user behaviour (Legris, Ingham, and Collerette,
2003). The original TAM contains six elements: external variables, perceived usefulness,
perceived ease of use, attitude, behavioural intention to use and actual use. According to
TAM, external variables have a direct influence on perceived usefulness (PU) and
perceived ease of use (PEU), i.e. the two cognitive belief components. Perceived ease of
use directly influences PU and attitude, whereas perceived usefulness has a direct impact on
attitude and behavioural intention to use, which affects actual use (Figure no. 1).
Figure no. 1. The original technology acceptance model (TAM)
Source: Davis, 1986.
Ha and Stoel (2008) examined the factors affecting customer acceptance of online shopping
and found that perceived ease of use, perceived trust and perceived shopping enjoyment
had the greatest impact on customer acceptance. Ease of use, trust and shopping enjoyment
had a significant impact on perceived usefulness; trust, shopping enjoyment, and usefulness
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 159
had a significant effect on attitude towards online shopping. They also found that attitude
and perceived usefulness had an influential role in consumer intention to purchase online.
According to Vijayasarathy (2004), there is a positive association between consumer attitude
towards online shopping and the beliefs concerning usefulness, compatibility, security and ease
of use. Also, the intention to purchase online is strongly influenced by consumer beliefs about
online shopping, self-efficacy and attitude. Surprisingly, no positive relationship between
purchasing intention and consumer beliefs about the usefulness of online shopping was
reported (Vijayasarathy, 2004). Gefen, Karahanna and Straub (2003) found that perceived
usefulness and perceived ease of use influence consumer repurchase intention.
It must be noted that Schepman and Rodway (2020) expressed some criticisms about the
applicability of TAM to measure attitudes towards AI. According to them, it is the online
retailers that can decide to integrate AI into webshops, and consumers have no choice but to
use it when shopping online in such stores. Therefore, traditional technology acceptance
models might not be ideal to measure attitudes towards AI. However, we are convinced that
consumers still have the free will to decide whether to use new technology, i.e. to shop
online in an AI-powered webshop, or not.
2. Methodology and research questions
2.1. Methodology
The constructs and the measurement instruments presented in Table no. 1 were developed
based on the literature review, and according to the Technology Acceptance Model.
Variables with asterisk and in italics were adapted from Park (2009), the others were
adapted from Hu and O'Brien (2016). However, each variable was modified by the authors
to make it possible to measure the perceived role of AI in online shopping.
For data collection, a questionnaire made up of 26 questions (variables) was used (Table
no. 1). Additionally, six demographics variables - gender, education, age, occupation, place
of residence and internet subscription - were also included in the survey. All measurement
instruments were listed in Table no. 1 but the demographics variables were measured on a
seven-point Likert-scale ranging from strongly disagree (1) strongly agree (7).
In the very first section of the questionnaire, respondents were provided with a detailed
explanation of AI-powered webshops and shopping apps, which are online stores where
shopping is supported by artificial intelligence. AI-powered webshops present personalised
product/service offerings based on previous search patterns and purchases that we made
before, and automatically display products that AI chooses for us. Also, AI offers similar
products to those that were originally viewed but were not available in the right size
(product recommendation based on visual similarity). Another typical sign of an AI-
powered webshop is that when the customer is leaving the web store, AI warns about the
products left in the cart, to complete the purchase. AI-powered webshops often use
chatbots, i.e. a virtual assistant is available if the customer has any questions, and visual
(image-based) search is also possible: after uploading a product picture, AI recommends
the most similar ones to that. Virtual changing rooms, voice recognition and automatic
search completion are also available in AI-powered webshops such as Amazon, e-Bay,
Alibaba, AliExpress, GearBest, eMAG.hu, PCland.hu, Ecipo, Bonprix, Answear, Reserved,
Fashiondays, Fashionup, Spartoo, Orsay, to mention just a few.
AE Consumer Acceptance of the Use of Artificial Intelligence in Online Shopping: Evidence From Hungary
160 Amfiteatru Economic
Table no. 1. Constructs and measurement instruments Construct Definition Measurement Instruments
Perceived
Usefulness
(PU)
The degree to which a
consumer believes that AI
used in online shopping
would make his or her
purchases more effective.
PU1. The use of AI in retail (shopping ads and
webshops) allows me to find the best deals.
PU2. The use of AI in retail enhances my
effectiveness in purchasing.
PU3. The use of AI in retail is useful to me.
PU4 The use of AI in retail saves time for me. *
Perceived
Ease of Use
(PEU)
The degree to which a
consumer believes that
using AI in webshops will
be free of effort.
PEU1. AI-powered shopping apps and webshops
are easy to use.
PEU2. Shopping does not require a lot of my
mental efforts if supported by AI (alternatives are
offered by AI).
PEU3. Shopping is not so complicated if AI offers
products to me.
PEU4 Learning how to use AI-powered shopping
apps and webshops is easy for me. *
PEU5 It is easy to become skilful at using AI-
powered shopping apps and webshops*
Experience
(EXP)
The consumers'
knowledge about and the
experience with
purchasing in an AI-
powered webshop.
EXP1. I'm experienced in online shopping.
EXP2. I have already used AI-powered applications
(chatbots, etc.)
Trust
(TRUST)
The subjective probability
with which people believe
that AI works for their
best interest.
T1. I am convinced that AI in retail is used to
provide customers with the best offerings.
T2. I trust in apps and webshops that use AI.
Subjective
Norm
(SN)
The degree to which a
consumer perceives that
most people who are
important to him or her
think he or she should or
should not make
purchases in AI-powered
webshops.
SN1. People who influence my behaviour would
prefer me to use AI-powered shopping apps and
webshops.
SN2. I like using AI-powered webshops and
shopping apps based on the similarity of my values
and the social values underlying its use. *
Task
Relevance
(TR)
The degree to which a
consumer believes that
AI-powered webshops are
applicable to his or her
shopping task.
TR1 I think AI can be used effectively in webshops
and shopping apps.
Compen-
sation
(COMP)
The degree to which a
consumer believes that he
or she has the ability to
make purchases in AI-
powered webshops.
I would prefer AI-powered shopping apps and
webshops…
C1. if there was no one around to visit physical
shops/shopping malls with.
C2. if I had less time.
C3. if I had a built-in help facility for assistance
when needed.
Perceived
Quality
PQ
The degree of how good a
consumer perceives the
quality of a product in AI-
powered webshops.
PQ1 AI finds/offers better products for me than I
could.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 161
Construct Definition Measurement Instruments
Perceived
Enjoyment
PE
The extent to which
shopping in AI-powered
webshops is perceived
to be enjoyable.
PE1 Shopping is more fun, enjoyable when AI
helps me to find the best-suited products.
Attitude
ATT
The consumer's attitude
towards shopping in AI-
powered webshops.
ATT1 Shopping in a webshop/shopping app that is
powered by AI is a good idea
ATT2 Shopping in a webshop/shopping app that is
powered by AI is a wise idea
ATT3 I am positive towards webshop/shopping app
that is powered by AI
Behavioural
Intention
BI
A consumer's behavioural
intention to do the
shopping in AI-powered
webshops.
BI1 I intend to visit webshops and to use shopping
apps that are powered by AI more frequently.
BI2 I'm willing to spend more on products offered
by webshops and apps powered by AI
Sources: Adapted from Hu and O'Brien, 2016; *Park, 2009.
An online survey in Google Form was conducted to collect data in July and August 2020 in
Hungary. Because of the Theory Acceptance Model, previous online shopping experience
with AI-powered webshops was the one and only eligibility criterion for respondents to
participate in this study. Convenience sampling method was used to reach the maximum
number of respondents. Data was migrated from Google Form to MS Excel, SPSS 24 and
AMOS, and was checked for coding accuracy. The database was complete and contained
no missing data. Descriptive statistical analyses were done in SPSS. AMOS was employed
to test the hypotheses and the theoretical model by structural equation modelling (SEM).
2.2. Research questions and hypotheses
Based on the literature review, this study aims to address the following research questions
respectively:
R1: Can the technology acceptance model (TAM) be used for investigating consumer
acceptance of the use of artificial intelligence in online shopping?
R2: If so, what are the key factors influencing behavioural intention to visit AI-
powered webshops and apps?
Based on the Technology Acceptance Model, an initial theoretical model was developed
(Figure no. 2). The arrows that link constructs (latent variables such as COMP, EXP,
TRUST, SN, PEU, PU, ATT, BI) represent hypothesised causal relationships (hypotheses)
in the direction of arrows. One of the objectives of this study is to test those hypotheses.
Error terms for all observed indicators are indicated by e1 to e35, respectively.
AE Consumer Acceptance of the Use of Artificial Intelligence in Online Shopping: Evidence From Hungary
162 Amfiteatru Economic
Figure no. 2. The initial theoretical model
2.3. The sample
A sample size of 200 is an appropriate minimum for SEM in AMOS (Marsh, Balla, and
MacDonald, 1988), and a minimum of 10-20 subjects per parameter estimates in the model
are optimal (Schumacker and Lomax, 2010). Therefore, the ideal sample size is between
380 and 760, considering the number of parameter estimates (38) in the initial model. The
actual sample size of 439 respondents fits into this category.
Of the sample of 439 respondents, 62.2% were female, 37.8% male. Their average age was
32.2 years. 60,8% of the respondents had tertiary education, 38% had secondary education,
and 1.2% had primary education. Most respondents resided in county seats (47.6%); the
rest lived in other towns/cities (24.1%), villages (17.8%) and the capital (10.5%). Most
respondents were subscribed to both mobile and wired internet services (84.3%), while
7.7% had only mobile internet, and 7.1% had only wired internet services. Only 0.9% of
respondents had not got any subscription to internet services (wired or mobile). There is no
data available on the distribution of the e-shoppers in Hungary, therefore, it is impossible to
tell if this sample reflects the characteristics of the e-shoppers’ population in Hungary.
3. Results and discussion
The initial model (Figure no. 2), which proved to be too complex and did not fit the current
data (CMIN/DF=7.72; p=.00; GFI=,693; CFI=.723; RMSEA=.124; HOELTER 0.5= 65),
was absolutely rejected. Therefore, it was not appropriate to interpret any individual
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 163
parameter estimates, and further model modifications were required to obtain a better-
fitting model. Respecification of the initial model led to a nested model that fitted well and
is discussed further. During the respecification, the alternative model approach was used
(Malkanthie, 2015). To test the model, the same data set was used. Several modified
models were developed, and out of the theoretically justifiable models, the model with the
best data fit was selected (Figure no. 3) as suggested by Mueller and Hancock (2008).
The respecification process was started with testing the measurement model by a series of
Principal Component Analysis (PCA). Variables with factor loadings under 0.7 were
deleted. A rule of thumb in confirmatory factor analysis suggests that variables with factor
loadings under |0.7| must be dropped (Malkanthie, 2015). As a result, only one external
variable, which is related to trust (T2), remained in the model (Table no. 2). Perceived
Usefulness (PU) was measured by three variables (PU1, PU2 and PU3), whereas Perceived
Ease of Use was made up of two variables (PEU2 and PEU3), and Behavioural Intention
became unidimensional (B1). The attitude was composed of three variables (ATT1, ATT2
and ATT3). The nested model, which is theoretically consistent with the research goals,
contains eight hypotheses:
H1: Attitude has a positive effect on behavioural intention.
H2: Perceived usefulness positively affects behavioural intention.
H3: Perceived usefulness has a positive effect on attitude.
H4: Perceived ease of use positively influences attitude.
H5: Perceived ease of use positively influences perceived usefulness.
H6: Perceived ease of use has a positive impact on trust.
H7: Trust has a positive effect on perceived usefulness.
H8: Trust positively influences attitude.
Figure no. 3. The nested model
AE Consumer Acceptance of the Use of Artificial Intelligence in Online Shopping: Evidence From Hungary
164 Amfiteatru Economic
3.1. Validity
To investigate the extent to which a set of items reflect the theoretical latent-construct they
are designed to measure, both convergent and discriminant validity were checked.
Convergent validity suggests that the variables of a factor that are theoretically related are
expected to correlate highly. According to the Fornell-Larcker criterion for convergent
validity, the Average Variance Extracted (AVE) should be greater than 0.5. According to
the Hair, et al. (1998) criteria, AVE should be greater than 0.5, standardised factor loading
of all items should be above 0.5, and composite reliability should be above 0.7.
In the nested measurement model, each factor loading was above .84 (Table no. 2).
Table no. 2. Summary of means, standard deviations, normality,
validity and reliability measures Cons-
truct Measurement Instrument Mean STD Z Skew
Z
Kurt
Loa-
ding α AVE CR
Per
ceiv
ed
Use
fuln
ess
PU1. The use of AI in
retail (shopping ads and
webshops) allows me to
find the best deals.
4.68 1.53 -4.06 -1.40 0.85
0.91 0.76 0.91 PU2. The use of AI in
retail enhances my
effectiveness in
purchasing.
4.67 1.63 -4.56 -1.87 0.89
PU3. The use of AI in
retail is useful to me. 4.73 1.69 -4.16 -2.70 0.89
Per
ceiv
ed E
ase
of
Use
PEU2. Shopping does not
require a lot of my mental
efforts if supported by AI
(alternatives are offered by
AI).
5.15 1.62 -6.38 -0.59 0.90
0.88 0.81 0.9
PEU3. Shopping is not so
complicated if AI offers
products to me.
5.06 1.64 -6.44 -0.68 0.90
Tru
st
T2. I trust in apps and
webshops that use AI. 4.11 1.62 -2.00 -2.90 1.00 1 n.a. n.a.
Att
itude
ATT1 Shopping in a
webshop/shopping app that
is powered by AI is a good
idea
5.02 1.63 -4.99 -1.58 0.90
0.9 0.79 0.92
ATT2 Shopping in a
webshop/shopping app that
is powered by AI is a wise
idea
4.23 1.62 -1.60 -2.39 0.86
ATT3 I am positive
towards webshop/shopping
app that is powered by AI
4.72 1.70 -4.11 -1.87 0.90
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 165
Cons-
truct Measurement Instrument Mean STD Z Skew
Z
Kurt
Loa-
ding α AVE CR
Beh
avio
ura
l
Inte
nti
on BI1 I intend to visit
webshops and use
shopping apps that are
powered by AI more
frequently.
3.35 1.78 2.13 -3.93 1.0 1 n.a. n.a.
Notes: STD=Standard Deviation, Z Skew=Z score for skewness, Z Kurt=Z score for
Kurtosis, α=Cronbach's alpha, AVE=Average Variance Extracted, CR=Composite
Reliability, N=439.
Moreover, all AVE scores were also well above the threshold level (AVE (ATT)=0.79;
AVE (PU)=.76 and AVE (PEU)=0.81), and all CR scores exceeded 0.7 (CR (PU)=.91; CR
(PEU)=0.90 and CR (ATT)=0.92). Therefore, the model meets both the Fornell-Larcker
(1981) criterion and the Hair et al. (1998) criteria for convergent validity, so the internal
consistency of the model is acceptable.
To assess discriminant validity, i.e. the extent to which a construct is truly distinct to other
constructs, AVEs were compared with squared inter-construct correlations (SIC). AVE
scores higher than SIC scores indicate that discriminant validity is acceptable (ATT
AVE=0.79, SIC1=0.61 and SIC2=0.32; PU AVE=0.76, SIC1=0.40 and SIC2=0.61; PEU
AVE=0.81, SIC1=0.40 and SIC2=0.32). Discriminant validity was also confirmed by
investigating correlations among the constructs. Since there were no correlations above .85,
which is a threshold limit of poor discriminant validity in structural equation modelling
(David, 1998), results also confirmed adequate discriminant validity (PEU*T2=0.52;
PEU*PU=0.64; PEU*ATT=0.57; PEU*BI1=0.43; T2*PU=0.73; T2*ATT=0.74;
T2*BI1=0.53; PU*ATT=0.78; PU*BI1=0.64; ATT*BI1=0.66).
3.2. Reliability
To test the accuracy and consistency of the nested model, three reliability tests were used:
Cronbach's alpha (α), the Average Variance Extracted index (AVE) and Composite
Reliability (CR). The threshold value for an acceptable Cronbach's alpha is .70 (Cronbach,
1951). The measurement model is acceptable if all estimates are significant and above 0.5
or 0.7 ideally; AVEs for all constructs are above 0.5 (Forner and Larcker, 1981); and
finally, CRs for all constructs are above 0.7 (Malkanthie, 2015). Table no. 2 shows that the
calculated Cronbach's alphas of all constructs were at least .87 or higher, and the AVE
scores were also higher than 0.76, as well as the CRs were above 0.9; therefore, the
reliability of the measurement model is optimal.
3.3. Model fit
Absolute- and relative model fits were tested. Each absolute measure was significant and
indicated a good fit. Although Chi-square statistics are sensitive to large sample size and
assume a multivariate normal distribution (Kelloway, 1998), even those measures were
acceptable. However, other model fit indexes are better to consider as criteria. Therefore,
the goodness-of-fit index (GFI), the adjusted goodness-of-fit index (AGFI), the root mean
squared error of approximation (RMSEA) and the standardised root mean squared residual
AE Consumer Acceptance of the Use of Artificial Intelligence in Online Shopping: Evidence From Hungary
166 Amfiteatru Economic
(SRMR) were also examined. All of them indicated a good absolute model fit. (Absolute
measures: Chi square=34.154 (DF=29); Probability level=0.23; CMIN/DF=1.18;
GFI=0.98; AGFI=0.96; RMSEA=0.02; SRMR=0.04). As far as the relative model fit is
concerned, TLI or NNFI, GFI, AGFI, NFI, IFI, CFI and Critical N (CN or HOELTER)
were calculated. All but CN range from zero to one. Values exceeding .9 show an
acceptable fit, above .95 a good fit (Bentler and Bonnet, 1980). CN (HOELTER), which
favours large samples over small ones (Bollen, 1990), is an improved method for
investigating model fit (Hoelter, 1983). CN should be above 200 to indicate a good model
fit. (Relative measures: TLI/NNFI=0.98; GFI=0.98; AGFI=0.96; NFI=0.93; IFI=0.99;
CFI=0.99 and HOELTER (CN)=546). The results of the absolute and relative model fit test
confirmed that the structural model is acceptable and suitable for the analysis and
interpretation of the parameter estimates. Therefore, it can be concluded that the technology
acceptance model is suitable for investigating consumer acceptance of the use of artificial
intelligence in online shopping, which is the answer to the first research question (R1).
3.4. Hypothesis testing and estimates
Because of the non-normality of the variables in the nested model, the asymptotically
distribution-free (ADF) method was used to estimate parameters in AMOS. ADF calculates
the asymptotically unbiased estimates of the chi-square goodness-of-fit test, the parameter
estimates, and the standard errors. The limitation of ADF is that it needs a large sample size
(Bian, 2012), which criterion was met in this study (N=439). Skewness and Kurtosis z-
values of the variables were out of the range of the normal distribution that is -2 and +2
(George and Mallery, 2010). Moreover, the p values of the variables were significant
(p=.000) in the Shapiro-Wilk and Kolmogorov-Smirnov tests, which also confirmed non-
normality.
To address the second research question (R2) and to determine the key factors influencing
behavioural intention to use AI-powered webshops and apps, hypotheses were tested in the
structural model (Table no. 3).
Table no. 3. Direct, indirect, total effects and hypothesis testing
Hypothesis Relationship P St. direct eff. St. indirect eff. St. total eff. Result
H1 BI1 ← ATT *** 0.41 0.00 0.41 accepted
H2 BI1 ← PU *** 0.32 0.19 0.51 accepted
H3 ATT ← PU *** 0.48 0.00 0.48 accepted
H4 ATT←PEU 0.1 0.09 0.48 0.57 rejected
H5 PU ← PEU *** 0.35 0.28 0.64 accepted
H6 T2 ← PEU *** 0.52 0.00 0.52 accepted
H7 PU ← T2 *** 0.55 0.00 0.55 accepted
H8 ATT ← T2 *** 0.35 0.26 0.61 accepted
The arrows linking constructs represent hypotheses in the direction of arrows in the nested
model (Figure no. 3 and Figure no. 4). Asterisks signal statistically significant relations
between constructs. Gamma estimates were calculated from exogenous construct to
endogenous construct, and beta estimates between two endogenous constructs. Figure no. 4
shows the standardised estimates, loadings and residuals regarding the relationships
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 167
between constructs and observed indicators. A hypothesis was accepted if the presence of a
statistically significant relationship in the predicted direction was confirmed.
As Table no. 3 shows, all hypotheses were accepted except for H4. So, the present findings,
except for the relationship between perceived ease of use and attitude, are consistent with
the Technology Acceptance Model proposed by Davis (1986). Surprisingly, perceived ease
of use (PEU) was found to have no direct, significant effect on attitude (ATT), which is not
in agreement with the original TAM (H4 rejected). This discrepancy could be attributed to
the fact that shopping is not too complicated in AI-powered webshops, and it does not
require too much mental effort. However, this slightly unexpected result coincides with the
findings of a previous research by Ha and Stoel (2008), who examined the effect of PEU on
attitude towards online shopping.
In this study, with H5 and H6 accepted, perceived ease of use (PEU) was found to have a
significant, direct, positive impact on both the perceived usefulness (PU) and trust (T2). It
suggests that the easier it is for a consumer to use an AI-powered webshop, the higher level
of customer trust and perceived usefulness can be expected. Consumers trust in AI-powered
shopping apps and stores that are easy to use, and consider those that are too complicated
less useful. Similar results were obtained by Ha and Stoel (2008), who focused on
consumers' acceptance of e-shopping. Gefen, Karahanna and Straub (2003) also found that
perceived ease of use positively affected the perceived usefulness of a B2C website and the
trust in an e-vendor.
Figure no. 4. Parameter estimates of the nested model
Trust in AI-powered webshops has a central role in forming attitudes and perceived
usefulness. Similar to what Gefen, Karahanna and Straub (2003), and Ha and Stoel (2008)
found, trust directly influenced perceived usefulness (H7 accepted). Moreover, trust also
impacted attitude (H8 accepted), in line with the research findings of Ha and Stoel (2008).
AE Consumer Acceptance of the Use of Artificial Intelligence in Online Shopping: Evidence From Hungary
168 Amfiteatru Economic
The strongest direct effect was found between trust and perceived usefulness (H7
accepted). It suggests that the more we trust in Artificial Intelligence during the online
shopping journey, the more likely it is that we consider AI-powered apps and webshops
useful. Besides, a higher level of trust forms a more positive attitude towards shopping in
such webshops. Perceived usefulness has a central role in this model as it (PU) significantly
impacted attitude (H3 accepted) and behavioural intention (H2 accepted). The more useful
we find the use of artificial intelligence in online shopping believing that it allows us to
grab the best deals, the more likely we are to consider it a wise decision to do the shopping
in AI-powered webshops and apps more frequently. Not surprisingly, attitude towards AI-
powered webshops and apps was found to have a strong, significant, positive direct impact
on behavioural intention (H1 accepted). It suggests that forming consumers’ attitude plays a
vital role in increasing the traffic of AI-powered webshops and apps (Figure no. 4).
Although there was no significant direct relationship between perceived ease of use and
attitude, the indirect effect of PEU on attitude (PEU->ATT=0.48) was quite strong, similar
to its indirect impact on behavioural intention (PEU->BI1=0.43). Also, trust was found to
indirectly influence behavioural intention (T2->BI1=0.42). It suggests that if shopping
requires much mental effort and seems to be complicated in AI-powered webshops and
apps, consumers tend to form stronger negative attitudes towards them and also tend to
trust them less, which will result in weaker consumer intention to visit such webshops.
In the nested model perceived usefulness had the highest total effect on behavioural
intention. Therefore, AI-powered webshops and apps are advised to increase the level of
perceived usefulness to succeed by enabling customers to maximise purchase effectiveness
to grab the best deals, i.e. the ideal product with the highest utility.
Conclusions
This research extends our knowledge of consumer acceptance of the use of artificial
intelligence in online shopping in many aspects. The widely used technology acceptance
model (TAM) was proved to be suitable for investigating consumer acceptance of the use
of artificial intelligence in online shopping.
As expected, it was confirmed in the nested model that the key factors influencing
consumer’ behavioural intention to use AI-powered webshops and apps are trust, perceived
usefulness, perceived ease of use and attitudes. In contrast to the original TAM (Davis,
1986), the direct relationship between perceived ease of use and attitudes was insignificant.
Nevertheless, it does not mean that user-friendliness of a webshop is not crucial as
perceived ease of use indirectly affects attitude and the behavioural intention. Instead, user-
friendliness and flawless operation of an artificial intelligence-powered website are the
prerequisites for market success.
Building trust has a central role in consumer acceptance of the use of artificial intelligence
in online shopping. If consumers do not trust in an AI-powered webshop/app, they tend to
consider it less useful and form a negative attitude towards it, which will result in less
online traffic. Also, AI must provide online consumers with tailor-made offerings to grab
the best deals, i.e. products with the highest value; and it is expected to shorten the product
search time to enhance shopping effectiveness. Not surprisingly, the favourable attitude
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 169
towards AI-powered webshops leads to more frequent online traffic in such electronic
stores.
Considering the strong positive impact of the recent COVID-19 crisis on e-commerce, the
use of artificial intelligence in online shopping is expected to expand further. According to
Bloomberg (2020) the pandemic lockdowns have a dual effect on consumer behaviour on
the development of AI. Nowadays, it is more important than ever to create a personalised
customer journey, to meet customers' demand and to provide a greater online shopping
experience. In these efforts, artificial intelligence can be a very effective tool, which was
confirmed by the research findings of this paper.
This study has several practical applications. It is useful for webshop owners and online
marketing managers to understand how consumers adapt to the new technology, i.e. the use
of artificial intelligence in online shopping. It is also beneficial to academics and
researchers who are interested in the adaptation of the Technology Acceptance Model in
online shopping. Those who are interested in the role of trust in consumer choices in the
online environment will also benefit from this study.
As far as the future research directions are concerned, it would be advisable to repeat this
study in a multi-cultural context. It might also be useful to test the model of the Technology
Readiness Index proposed by Parasuraman (2000) and to compare the results presented
here with the new findings.
Acknowledgements
“The described article/presentation/study was carried out as part of the EFOP-3.6.1-16-
2016-00011 “Younger and Renewing University – Innovative Knowledge City –
institutional development of the University of Miskolc aiming at intelligent specialisation”
project implemented in the framework of the Szechenyi 2020 program. The realization of
this project is supported by the European Union, co-financed by the European Social
Fund.”
References
Ajzen, I., 1985. The Theory of Planned Behavior. Organisational Behavior and Human
Decision Processes, 50, pp.179-211.
André, Q., Carmon, Z., Wertenbroch, K., Crum, A., Frank, D., Goldstein, W., Huber, J.,
Boven, L., Weber, B. and Yang, H., 2017. Consumer Choice and Autonomy in the Age
of Artificial Intelligence and Big Data. Customer needs and solutions, 5(1-2), pp.28-37.
Aranyossy, M. and Magisztrák, B., 2016. A vásárlói bizalom hatása az e-kereskedelmi
vásárlási hajlandóságra. Marketing & Menedzsment, 3-4, pp.73-87.
Asling, D., 2017. 19 Powerful Ways To Use Artificial Intelligence In eCommerce. [online]
Available at: <https://blog.linnworks.com/artificial-intelligence-in-ecommerce>
[Accessed 27 August 2020]
Bakos, J.Y., 1997. Reducing buyer search costs: implications for electronic marketplaces.
Management Science, 43(12), pp.1676-1692.
AE Consumer Acceptance of the Use of Artificial Intelligence in Online Shopping: Evidence From Hungary
170 Amfiteatru Economic
Barmada, N., 2020. Grow your business in the Nordics when Amazon Sweden launches this
year. [online] Available at: <https://blog.linnworks.com/grow-your-business-across-
borders-when-amazon-sweden-launches-this-year> [Accessed 27 August 2020].
Bentler, P.M. and Bonnet, D.G., 1980. Significance tests and goodness-of-fit in the analysis
of covariance structure. Psychological Bulletin, 88(3), pp.588-606.
Bian, H., 2012. Structual Equation Modeling Using Amos. New York: Routledge.
Bloomberg, 2020. Coronavirus will finally give artificial intelligence its moment. [online]
Available at: <https://economictimes.indiatimes.com/small-biz/startups/features/
coronavirus-will-finally-give-artificial-intelligence-its-
moment/articleshow/76477021.cms?from=mdr> [Accessed 27 November 2020].
Bollen, K.A., 1990. Overall fit in covariance structure models: two types of sample size
effects. Psychological Bulletin, 107(2), pp.256-259.
Cătoiu, I., Orzan, M., Macovei, O.I. and Iconaru, C., 2014. Modelling Users’ trust in online
social networks. Amfiteatru Economic, 16(35), pp.289-302.
Cronbach, L.J., 1951. Coefficient alpha and the internal structure of tests. Psychometrika
16, pp.297-334. https://doi.org/10.1007/BF02310555.
Daley, S. 2018. 19 examples of artificial intelligence shaking up business as usual [online]
Available at: <https://builtin.com/artificial-intelligence/examples-ai-in-industry>
[Accessed 17 August 2020]
Davenport, T.H. and Ronanki, R., 2018. Artificial Intelligence for the Real World. Harvard
Business Review, January-February 2018., pp.108-116.
Davis, F.D., 1986. A Technology Acceptance Model for empirical testing new end-user
information systems: Theory and results, Doctoral Dissertation, MIT. [online]
<https://dspace.mit.edu/handle/1721.1/15192> [Accessed 17 August 2020].
Dhagarra, D., Goswami, M. and Kumar, G., 2020. Impact of Trust and Privacy Concerns
on Technology Acceptance in Healthcare: An Indian Perspective. International Journal
of Medical Informatics, 141, pp.104164. doi:10.1016/j.ijmedinf.2020.104164.
Dumitriu, D. and Popescu, M.A.M., 2020. Artificial Intelligence Solutions for Digital
Marketing. Procedia Manufacturing, 46, pp.630-636.
Gefen, D., Karahanna, E. and Straub, D.W., 2003. Trust and TAM in Online Shopping: An
Integrated Model. MIS Quarterly, 1, pp.51-90.
European Commission, 2018. A definition of AI: Main Capabilities and Disciplines.
Definition developed for the purpose of the AI HLEG's deliverables. [online] Available
at: <https://ec.europa.eu/digital-single-market/en/news/definition-artificial-intelligence-
main-capabilities-and-scientific-disciplines> [Accessed 07 August 2020].
Fishbein, M. and Ajzen, I., 1975. Belief, attitude, intention and behavior: An introduction
to theory and research. Massachusetts: Addison-Wesley.
Fornell, C. and Larcker, D., 1981. Evaluating Structural Equation Models with
Unobservable Variables and Measurement Error. Journal of Marketing Research, 18(1),
pp.39-50. doi:10.2307/3151312.
George, D. and Mallery, M., 2010. SPSS for Windows Step by Step: A Simple Guide and
Reference, 17.0 update (10a ed.) Boston: Pearson.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 171
Ha, S. and Stoel, L. 2008. Consumer e-shopping acceptance: Antecedents in a technology
acceptance model. Journal of Business Research, 62, pp.565-571.
Hair, J.F., Anderson, R.E., Tatham, R.L. and Black, W.C. 1998. Multivariate data analysis:
(5th ed.). Upper Saddle River, NJ: Prentice Hill.
Haller, K., Lee, J. and Cheung, J., 2020. Meet the 2020 consumers driving change. Why
brands must deliver on omnipresence, agility, and sustainability. IBM Institute for
Business Value. [online] Available at: <https://www.ibm.com/thought-leadership/
institute-business-value/report/consumer-2020#> [Accessed 27 November 2020].
Hoelter, J.W., 1983. The analysis of covariance structures: Goodness-of-fit indices.
Sociological Methods and Research, 11, pp.325-344.
Hu, K. and O'Brien, S., 2016. Applying TAM (Technology Acceptance Model) to testing MT
acceptance. [online] Available at: <https://ec.europa.eu/info/sites/info/files/
tef2016_kehu_en.pdf > [Accessed 07 July 2020].
Kim, D., Ferrin, D.L. and Rao, H.R. 2008. A trust-based consumer decision-making model
in electronic commerce: The role of trust, perceived risk, and their antecedents.
Decision Support Systems, 44(2), pp.544-564.
Kelloway, E.K., 1998. Using LISREL for structural equation modeling: A researcher's
guide. Thousand Oaks, CA: Sage.
Kwong, C.K., Jiang, H. and Luo, X., 2016. AI-based methodology of integrating affective
design, engineering, and marketing for defining design specifications of new products.
Engineering Applications of Artificial Intelligence, 47, pp.49-60.
Legris, P., Ingham, J. and Collerette, P., 2003. Why do people use information technology?
A critical review of the technology acceptance model. Information & Management
40(3), pp.191-204.
Lynch, J.G. and Ariely, D., 2000. Wine Online: Search Costs Affect Competition on Price,
Quality, and Distribution. Marketing Science, 19(1), pp.83-103.
Malkanthie, A., 2015. Structural Equation Modeling with AMOS. Lap Lambert Academic
Publishing, Germany.
Marsh, H.W, Balla, J.R. and MacDonald, R.P., 1988. Goodness-of-fit indexes in
confirmatory factor analysis: The effect of sample size. Psychological Bulletin, 88,
pp.245-258.
Maynard, N., 2019. Juniper research. How AI can revive retail? [online] Available at:
<https://www.juniperresearch.com/document-library/white-papers/how-ai-can-revive-
retail> [Accessed 07 August 2020]
Meticulous Market Research, 2020. Artificial Intelligence (AI) in Retail Market Worth
$19.9 billion by 2027- Exclusive Report Covering Pre and Post COVID-19 Market
Analysis. [online] Available at: <https://www.prnewswire.com/news-releases/artificial-
intelligence-ai-in-retail-market-worth-19-9-billion-by-2027--exclusive-report-covering-
pre-and-post-covid-19-market-analysis-by-meticulous-research-301098029.html>
[Accessed 27 November 2020]
Mueller, R. and Hancock, G., 2008. Best practices in structural equation modeling. In:
J. Osborne, ed. 2008. Best practices in quantitative methods. Thousand Oaks, CA: Sage
Publications, Inc. pp. 488-508.
AE Consumer Acceptance of the Use of Artificial Intelligence in Online Shopping: Evidence From Hungary
172 Amfiteatru Economic
Onete, B., Constantinescu, M. and Filip, A., 2008. Internet buying behavior. Case study:
research of AES students' behavior regarding online shopping. Amfiteatru Economic,
November, pp.18-24.
Pantano, E. and Pizzi, G., 2020. Forecasting artificial intelligence on online customer
assistance: evidence from chatbot patents analysis. Journal of Retailing and Consumer
Services, 55, pp.102096.
Parasuraman, A., 2000. Technology Readiness Index (TRI): A Multiple item scale to
measure readiness to embrace new technologies. Journal of Service Research, 2(4),
pp.307-320.
Park, S.Y., 2009. An Analysis of the Technology Acceptance Model in Understanding
University Students' Behavioral Intention to Use e-Learning. Educational Technology &
Society, 12(3), pp.150-162.
Paschen, J., Wilson, M. and Ferreira, J., 2020. Collaborative intelligence: How human and
artificial intelligence create value along the B2B sales funnel. Business Horizons, 63(3),
pp.403-414.
Pricewaterhouse Coopers, 2018. Künstliche Intelligenz als Innovationsbeschleuniger in
Unternehmen – Zuversicht und Vertrauen in Künstliche Intelligenz. [online] Available
at: <https://www.pwc.de/de/digitale-transformation/ki-als-innovationsbeschleuniger-in-
unternehmen-whitepaper.pdf > [Accessed 07 August 2020].
Pusztahelyi, R., 2020. Emotional AI and its challenges in the viewpoint of online
marketing. Curentul Juridic, 23(2), pp.13-31.
Rajagopal, P., 2002. An innovation-diffusion view of implementation of enterprise resource
planning (ERP) systems and development of a research model. Information &
Management, 40(2), pp.87-114.
Reichheld, F.F. and Schefter, P., 2000. E-loyalty your secret weapon on the web. Harvard
Business Review, 78(4), pp.105-113.
Roetzer, P., 2017. 6 Limitations of Marketing Artificial Intelligence, According to Experts.
[online] Available at: <https://www.marketingaiinstitute.com/blog/limitations-of-
marketing-artificial-intelligence> [Accessed 25 September 2020].
Rust, R.T. and Huang, M.H., 2014. The Service Revolution and the Transformation of
Marketing Science. Marketing Science, 33(2), pp.206-221.
Schepman, A. and Rodway, P., 2020. Initial validation of the general attitudes towards
Artificial Intelligence Scale. Computers in Human Behavior Reports, 1, pp.
100014. DOI: 10.1016/j.chbr.2020.100014.
Schumacker, R.E. and Lomax, R.G., 2010. A Beginner's Guide to Structural Equation
Modeling. New York: Routledge. https://doi.org/10.4324/9780203851319.
Shankar, V., 2018. How Artificial Intelligence (AI) Is Reshaping Retailing. Journal of
Retailing, 94(4), pp.343-348.
Smidt, F. and Power, B. 2020. 8 ways consumers across Europe adapted their shopping
behaviour this year. [online] Available at: <https://www.thinkwithgoogle.com/intl/en-
cee/insights-trends/industry-perspectives/consumers-adapted-shopping-behaviour-
covid/> [Accessed 27 August 2020].
Stigler, G.J., 1961. The Economics of Information. Journal of Political Economy, 69(3),
pp.213-225.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 173
Stiegler, G.J. and Becker, G.S., 1977. De Gustibus Non Est Disputandum. The American
Review. 67(2), pp.76-90.
Thatcher, J.B., Carter, M., Li, X. and Rong, G., 2013. A Classification and Investigation of
Trustees in B-to-C e-Commerce: General vs. Specific Trust. Communications of the
Association for Information Systems. 32(4). 10.17705/1CAIS.03204.
Venkatesh, V., 2000. Determinants of perceived ease of use: integrating control, intrinsic
motivation, and emotion into the technology acceptance model. Information Systems
Research, 11(4), pp.342-365.
Venkatesh, V., Morris, M.G., Davis, G.B. and Davis, F.D., 2003. User Acceptance of
Information Technology: Toward a Unified View, MIS Quarterly, 27(3), pp.425-478.
Vijayasarathy, L.R., 2004. Predicting consumer intentions to use online shopping: the case
for an augmented technology acceptance model. Information & Management, 41,
pp.747-762.
Weber, F. and Schütte, R., 2019. A Domain-Oriented Analysis of the Impact of Machine
Learning-The Case of Retailing. Big Data Cognition Computation, 3(1), pp.1-14.
Yoo, W.-S., Lee, Y. and Park, J.K., 2010. The role of interactivity in e-tailing: Creating
value and increasing satisfaction. Journal of Retailing and Consumer Services, 17(2),
pp.89-96.
AE Benefits and Risks of Introducing Artificial Intelligence Into Trade and Commerce: The Case of Manufacturing Companies in West Africa
174 Amfiteatru Economic
BENEFITS AND RISKS OF INTRODUCING ARTIFICIAL INTELLIGENCE
IN COMMERCE: THE CASE OF MANUFACTURING COMPANIES
IN WEST AFRICA
Zelin Zhuo1, Frank Okai Larbi2* and Eric Osei Addo3
1)2) Institute of International and Comparative Education, Research Center
for Hong Kong &Macau Youth Education, South China Normal University,
Guangzhou, China 3) School of International Trade and Economics, University of International
Business and Economics, Beijing, China
Please cite this article as:
Zhuo, Z., Larbi, F.O. and Addo, E.O., 2021. Benefits
and Risks of Introducing Artificial Intelligence Into
Trade and Commerce: The Case of Manufacturing
Companies in West Africa. Amfiteatru Economic,
23(56), pp. 174-194.
DOI: 10.24818/EA/2021/56/174
Article History
Received: 13 August 2020
Revised: 12 November 2020
Accepted: 29 December 2020
Abstract
With innovations in technology, the application of artificial intelligence (A.I) in the area of
commerce is rising to the top with an expected growing number of business transactions not
just for entrepreneurs but for consumers as well. It advances the understanding of how A.I.
can be used to enhance businesses around the world by establishing their presence online to
reach customers beyond borders. This study highlights the benefits and risks of introducing
A.I. into trade in terms of how the commerce industry operates and revolutionize products
shopping. Significantly, the primary aim of this paper is to explore ways A.I. is integrated
into commerce to help understand its impact on existing/potential customers and its
efficiency in sales processes. With a sample size of 2,903 manufacturing companies in four
West-African countries, the empirical results show that value-based adoption of A.I.
approach outperforms the traditional/human search of customers’ products delivery in both
convenience, accuracy and profitability. Furthermore, A.I. approach within commerce
achieved competitive advantage with several modernized customer service machine
learning approach such as automated content creation, voice assistance, image search
among others. Clearly, this shows that the application of A.I system into commerce
introduces significant competitive advantages in terms of trust, quality, openness and
security.
Keywords: Artificial Intelligence, Human Interaction, Commerce, Value-based Adoption
model (VAM), Probit Model, West Africa
JEL Classification: O2, O3, O33
* Corresponding author, Frank Okai Larbi – e-mail: [email protected]
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 175
Introduction
In recent years, artificial intelligence (A.I.) has drawn attention as a key for economic
growth in developed countries. Indeed, Sutton and Trefler (2016) describe both
theoretically and empirically how developing countries such as China initially entered new
markets at a low level of quality but over time developed the capabilities to deliver high-
quality, and internationally competitive goods and services via A.I. technology. This is
mainly due to the attention been focused on developing new A.I. information
communication technology (Lu et al., 2018). The use of A.I. technologies offer many
benefits (Canbek and Mutlu, 2016) and risks (Alzahrani, 2019). It brings the appearance of
normal human language use into a new social relation between machines and humans
(Barrett et al., 2019; Sanzogni, 2017). This innovative media presents a powerful
technology that uses analytics to determine news feeds, information, products and
purchases (Cunnean et al., 2019). Notably, it is a fundamental, pervasive economic and
organizational phenomenon that holds many opportunities in store for management (Foss,
2005). Again, A.I. unlike the natural intelligence displayed by humans is transforming the
face of commerce in the business world and how it creates products and services to
customers (Armbrust et al., 2011). Additionally, China has become the focal point for much
of the A.I. discourse. For instance, China has developed significant commercial A.I
capabilities, evidenced by companies such as Baidu (a search engine like Google), Alibaba
(an e-commerce web portal like Amazon), and Tencent (the developer of WeChat, which
can be seen as combining the functions of Skype, Facebook and Apple Pay). Today, A.I
remains the most spectacular I.T application, a technology that has gone through an
unequalled development over the last decades (Blanchet et al., 2019; Lee et al., 2018;
Wiljer and Hakim, 2019). In business, A.I is relevant to any intellectual task because it is
becoming an imperative for firms that want to maintain a competitive edge. In this way,
humans can use A.I. to help game out possible consequences of each action and streamline
the decision-making process. However, the growing of A.I innovation led businesses to
make decisions to adopt new technology to address customer needs and support product
services aimed at satisfying commerce transactions (Ekufu, 2012). On the other hand,
consumer trust is more important in cyber transactions than it is in traditional transaction
because trust is a prerequisite for successful commerce and as a result customers are
hesitant to make purchases unless they trust the seller (Gefen, 2002; Jarvenpaa et al., 1998;
Kim et al., 2007). Although, there have been some limitations with A.I. adoption in
commerce, there are various avenues of corporate decision making and problem-solving by
A.I usage such as data mining, credit worthiness, stock market predictions among others.
By definition, A.I is best understood as a set of techniques aimed at approximating some
aspects of human or animal cognition using machines (Calo, 2017). According to Huang
and Rust (2018) and Syam and Sharma (2018) A.I is manifested by machines that exhibits
aspects of human intelligence and involves machines mimicking intelligent human
behaviour. This means that it relies on several key technologies, such as machine learning,
natural language processing, rule-based expert systems, neural networks, deep learning,
physical robots, and robotic process automation (Davenport, 2018). Furthermore, A.I
involves the use of a computer to model intelligent behaviour with minimal human
intervention (Benko and Lanyi, 2009; Haenlein and Kaplan, 2019; McCorduck et al., 1977).
In sum, A.I. can be defined as a section of informatics and applied computer science to
pattern human proceedings of problem solving and transfer them to computers in order to
invent efficient and new solutions as well as course of actions. Therefore, A.I. is a computer
AE Benefits and Risks of Introducing Artificial Intelligence Into Trade and Commerce: The Case of Manufacturing Companies in West Africa
176 Amfiteatru Economic
program running on any possible device or data center with the skill to interact with its
environment (Dautenhahn, 2007). However, the adoption of such technology starts instead
with profits because profits are at the core of arguments supporting the introduction of A.I.
in trade (Agrawal et al., 2018). Nevertheless, information technologies have become
ubiquitous in professional activities, disrupting and affecting all core processes and
operations (Devaraj and Kohli, 2003). Considering the importance of A.I. in today’s
commerce, this paper explores the benefits and risks of introducing such a mechanism into
trade. Furthermore, this study reviews the process of A.I. in commerce together with the
processes of conducting businesses for profit or not for profit goods, commodities, property
or services in the field of commerce. Therefore, the objective of this study is to address the
following research questions: firstly, does the introduction of A.I improves commerce
performance at both the organizational and process level and secondly, what is the business
value of A.I based projects within organizations. From above background, the authors draw
seven key hypothesis:
H1: Usefulness of A.I positively affects the customer’s purchase decisions.
H2: Trust of A.I negatively affects the consumer’s risk of a transaction.
H3: Data management in A.I will more likely affects the adoption of its usage.
H4: Experts knowhow in A.I will more likely affects the adoption of its usage.
H5: Cost incur will more likely affects the firm’s adoption of its usage.
H6: A.I privacy protection positively affects the consumer’s intention to purchase on the
internet.
H7: Overall value of A.I positively affects the introduction of such technological tool.
To conclude this overview, it is worth noting that A.I.is here to stay and will be an integral
part of the future of the retailing and commerce sector affirming the power for commerce
businesses to explore countless opportunities to improve customer experiences, better
understand their customers and generate profits from firms operations.
1. Literature review
1.1. Value-based Adoption Model (VAM)
Value-based adoption model (VAM) proposed by Kim et al. (2007) empirically test this
novel approach towards understanding consumers’ adoption of technology. Kim et al.
argued that the TAM has limitations in explaining new ICT acceptance and that those who
accept new ICT are not just technology users but also consumers. Furthermore, they
claimed that the main interests of technology users in an organization are usefulness and
ease of use, but that rational consumers focus more on maximization of value (Lin et al.,
2012). VAM saw benefits and sacrifice as the main factors of value and analyzed intention
to use. Additionally, it is based on a cost-benefit paradigm which reflects the decision-
making process where the decision to use is made by comparing the cost of uncertainty in
choosing a new technology or product. Empirically, this means that VAM aims to explain
the adoption of technology in order to overcome the limits of technology acceptance model
in a new ICT environment (Lin et al., 2012). However, Davis et al. omitted attitude in the
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 177
final TAM due to its weak mediation of beliefs on adoption intention. Empirical studies
have found that attitude does not influence intention directly, and that TAM retains its
robustness even without including attitude (Davis et al., 1989; Venkatesh et al, 2003).
Again, they concluded in their review of IT acceptance research that attitudinal constructs
are significant only when specific cognitions (performance and effort expectancies) are not
included in the model (Venkatesh, 2003).
From figure no. 1 results that perceived value is affected by benefits and sacrifices.
Figure no. 1. The Basic Concept of Value-based Adoption Model (VAM)
Source: Kim et al., 2007
Perceived value affects adoption intention and as a result perceived benefits are derived
from the cognitive evaluation theory (Deci, 1971) which classifies motivations into
extrinsic and intrinsic subsystems. Extrinsic motivation refers to the performance of an
activity to achieve a specific goal while intrinsic motivation refers to the performance of an
activity for no apparent reinforcement other than the process of performing the activity per
se (Davis et al., 1989). Both extrinsic and intrinsic factors have been found to influence
perceived value and behavioral intention and these findings also apply to information
systems (Moore and Benbasat, 1991). Clearly, this means that perceived value is defined as
the subjective evaluation of consumers of the trade-off between benefits and costs of
products or services (Zeithaml, 1988). The definition of perceived value by Zeithaml (1988)
is widely used and indicates an overall evaluation by consumers of the usefulness of
products. Again, value is derived from comparison of the acquired benefits with costs paid,
and costs paid must consider sacrifice of effort and time as well as monetary aspects
(Bolloju et al., 2002). Additionally, sacrifices are both monetary and nonmonetary.
Monetary spending includes the actual price of the product, and it is generally measured
based on customers’ perceptions of the actual price paid. Non-monetary costs usually
include time, effort and other unsatisfactory spending for the purchase and consumption of
the product (Thaler, 1985; Zeithaml, 1988).
1.2. Application of proposed framework and hypothesis
Taking into account above arguments, the study develop hypothesis around Value-based
Adoption Model (VAM) into trade / commerce. According to this theory of utility, users try
to achieve maximum utility or satisfaction given their resource limitations. It is crucial to
mention that artificial intelligence (A.I) technologies are developing apace with many
potential benefits for economies, societies, communities and individuals. In considering the
potential impact of A.I on commerce provide for a suite of technologies that perform tasks
usually associated with human intelligence.
AE Benefits and Risks of Introducing Artificial Intelligence Into Trade and Commerce: The Case of Manufacturing Companies in West Africa
178 Amfiteatru Economic
Taking into account figure no. 2, usefulness ensures the quality of having technological
system and especially practical worth or applicability while trust in A.I usage bring
confidence and/or reliance on the integrity and strength of products transaction.
Figure no. 2. Proposed Framework and Hypothesis of Value-based Adoption Model
(VAM)
Also, Data management allows permit smooth analysis and exploitation of the data derived
from production to be carried out in real time. With regards to barriers to sacrifice, the lack
of experts of A.I pose as the scarcity of professionals with skills and experience in this type
of implementations. Moreover, it’s crucial in these cases to have individuals (professionals)
who have experience on projects of the same magnitude. Likewise the fee of introducing
A.I into commerce which requires huge costs as it is a complex machine. Apart from the
installation cost, its repair and maintenance also require huge costs. Lastly, A.I technology
privacy as to protecting personal data and ensuring users of A.I that their information is
confidential and management of data protection is a challenge to most organizations. It is
on this premise that the study seeks to address above-mentioned hypothesis taking into
consideration proposed variables.
1.3. Success factors impacting artificial intelligence into trade/commerce
The introduction of A.I. has immense contribution to the development of the
business/commerce industry in the area of product customization, market trend analysis,
target marketing, customer relationship management, web personalization among others.
Below are some key benefits of introducing artificial intelligence into trade/commerce.
Firstly, artificial intelligence enhance creative tasks by freeing users from routine and
repetitive tasks and allows them to spend more time on creative functions. In doing so,
allows robots to develop repetitive, routine and process optimization tasks automatically
and without human intervention. Secondly, A.I reduces failures caused by human
limitations. In some production lines, A.I is used to detect, by means of infrared sensors,
small cracks or defects in parts that are undetectable by the human eye. Thirdly, A.I control
and optimize of productive processes and production lines more efficiently via error-free
processes and obtain greater control over production lines in the company. This not only
increases productivity at the machine level, it also makes workers more productive and
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 179
increases the quality of the work they do because having more information allows workers
to have a more focused view of their work and make better decisions. Furthermore, there is
improvement in decision making at both production and business levels. Meaning, by
having more information in a structured way, it allows each of the people in charge to make
decisions in a faster and more efficient way. Lastly, there is efficiency in data acquisition
and analysis whereby computers always worked with data extraordinarily well and A.I is
extremely good at working with high volumes of data that humans simply cannot handle.
1.4. Barriers of introducing artificial intelligence into trade/commerce
With regard to the limitations of introducing A.I into commerce, below are some of the most
common factors that can occur in the business environment. The first risk is the cost and
installation time of A.I projects. The cost of installation both at the time and the economic
level, is a very important factor in choosing to execute this type of project. Companies that
lack internal skills or are not familiar with AI systems, must value the outsourcing of both
implementation and maintenance in order to obtain successful results in their project. Another
obstacle that often occurs at the business level for introducing A.I is the lack of qualified
professionals to manage / operate the technology. Thirdly, artificial intelligence cannot be
improved with experience because they perform the same function again if no different
command is given to them. With time, it can lead to wear and tear. Again, it stores a lot of
data but the way it can be accessed and used is very different from human intelligence. This
means that A.I technology cannot cope up with the dynamic environment and so they are
unable to alter their responses to changing environments. Moreover, artificial intelligence lack
privacy consideration as it may pose a great challenge for humanity if it reaches a very
advanced stage. At what point may a machine be deemed sentient, conscious, and therefore
entitled to similar to what we call human rights. We may never reach this stage of A.I seen it
may not be as hard as some imagine but early awareness of A.I’s privacy considerations is
necessary for safeguarding users personal information.
1.5. Effects of COVID-19 pandemic on implementing artificial intelligence
A.I offers industries an avenue to sustain economic activities and business performance during
times of crisis. Fundamentally, the lapses of increasing security threat needs to be address
especially in developing countries. Currently, the impact of COVID-19 situation has resulted in
the growing use of artificial intelligence in trade / commerce. Significantly, the effect of A.I
stands to promote reliable and practical implementation of digital marketing across businesses.
Moreover, the applications of Artificial Intelligence has the potential to solve challenges in the
distribution of goods during unforeseeable events like COVID-19 pandemic among others.
Again, its implementation would provide some practical insights on how the pandemic
threatens the closure of small businesses and also, preventing consumers’ access to essential
household goods. Undoubtedly, it should be echoed that to implement A.I, the development of
strong institutions with strict regulation / governance on cyber transaction is pivotal. Moreover,
the effect of COVID-19 on implementing A.I, for instance, in the manufacturing sector can
help citizens limit the spread of infection as it prevents people from moving around since a lot
of orders will be carried out online. In sum, governments need to ensure that the supporting
technology for A.I is adequate in its performance and the establishment of centralized
platforms for data management.
AE Benefits and Risks of Introducing Artificial Intelligence Into Trade and Commerce: The Case of Manufacturing Companies in West Africa
180 Amfiteatru Economic
2. Methodology
In order to perform our empirical analyses, the study uses questionnaire survey conducted
in four West-African countries, namely: Ghana, Nigeria, Togo and Burkina Faso. The
selected countries were chosen because of the large number of manufacturing and
production companies located in these countries. Furthermore, the responded data for this
study were collected via an online platform. A total of 3,000 questionnaires were
administered online and afterwards, the study analyzed completed data of 2,903 from
manufacturing companies (after accounting for missing data). Nevertheless, the study
examines perceived value as the comparison between benefit and cost (sacrifice) which
compares (1) usefulness, trust and data management (2) experts, cost and privacy and
(3) Overall value of introducing artificial intelligence into trade/commerce. Significantly,
the research model and the proposed hypothesis were evaluated by the probit model in
examining the benefits and risks of introducing A.I. into commerce using manufacturing
companies in four West African countries. Also, the probit model regression was performed
using the 40 items drafted in the questionnaire survey via manufacturing companies’
performance status as dependent variable. Empirically, let Mijt denote the performance
status of technology i = 1,……, nj in manufacturing j = 1,……, J at time point t ϵ (0,1).
Performance status is assumed to be continuous and ranges from Strongly Disagree (1),
Disagree (2), Not Sure (3), Agree (4), or Strongly Agree (5) in the questionnaire survey
using SPSS version 26. In addition, the likert scale data is defined in SPSS as “1/2/3/4/5”.
The coding of underlying status Mijt to observed, discrete technology performance category
Mijt is given by the standard measurement model:
3 if Mijt ≤ k1
Mijt 2 if k1 < Mijt ≤ k2
1 if Mijt > k2 (1)
where the parameters, k, are unobserved and must be estimated from the data. The
categories are ordered from worst to best. This facilitates the qualitative interpretation of
regression coefficients, where a positive sign indicates acceptance and improvement of
technological usage (A.I) and, thus, the probability of reporting no problems. In addition,
introducing A.I. technology at any time point t is described by the equation:
Mijt = αij + ϕj +xijβ + Tvj +(T · xij)ψ + εijt (2)
with
vj = µ + yj (3)
The bearing xij is a set of benefits-risk adjustment variables that are, in this case, time
invariant, where beta (β) is the estimate of the influence of each variable. Treatment is
modelled as a dummy variable T, which takes a value of 1 if t = 1 (post-introduction) and
0 otherwise. The direct effect of treatment performance on post-introduction technology
(A.I) usage is given by the coefficient υj. Afterwards, the study computes the probability of
reporting a specific post-introduction performance status category (m=1, 2, 3), based on the
estimated load exerted by the manufacturing and production companies in providing better
services as determined indirectly from above equations. This is given as:
Prob (Mj1 = m|x, yj âij = φj = 0) = ϕ(km-Sj1)– ϕ(km-1-Sj1) (4)
where
Sj1=µ+x'β+x'φ+yj (5)
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 181
Moreover, using SPSS version 26, the study performed a reliability test. By reliability
measurement, all proposed variables show good internal consistency with resulting
Cronbach’s alphas (α) ranging between 0.802 and 0.932.
According to Cronbach (1951), if all the scale items are entirely independent from one
another (i.e., are not correlated or share no covariance), then α = 0 and if all the items have
high covariance, then α will approach 1 as the number of items in the scale approaches
infinity. In other words, the higher the alpha (α) coefficient, the more the items have shared
covariance and measure the underlying concept. From table no. 1, the coefficient test of
internal consistency is acceptable.
Table no. 1: Summary Coefficient Output
No. Variable Cronbach Alpha
1 Usefulness 0.932
2 Trust 0.802
3 Data 0.871
4 Experts 0.836
5 Cost 0.825
6 Privacy 0.902
7 Value 0.916
8 Adoption 0.873
3. Analysis and discussion of results
The characteristics of respondents described and grouped by sex, age, level of education,
length of work and categories of manufacturing companies in four selected West African
countries. The profile of respondents are indicated in table no. 2 below. It is explained that
male manufacturers occupies 55.39%, which means that men are in control of such an
industry and key decision makers of such digital technological usage into commerce than
women with a score of 44.61%. Again, the study recorded 47.61% of industry players who
are over 41-50 years of age. As Ahadiat (2008) found the same result that such age bracket
are more positive in the utilization of such technological media. Furthermore, 54.15% of the
respondents attained a bachelor’s degree and this juxtapose the technological skills they
possesses in using artificial intelligence tool in their line of operations. Likewise, the
working period which influence their capabilities in digital platform. From the table, 36.31
% of the respondents have been in the industry for over 21 years and this immensely had
contributed to the success story of introducing artificial intelligence into trade/commerce.
Lastly, top three sub-sector manufacturers that contributed hugely to the study were the
aluminium sector (20.01%), followed by textiles (15.77%) and then automotive (13.88%).
Clearly, this shows the adoption rate of manufacturers in promoting the use of artificial
intelligence in their business transactions.
AE Benefits and Risks of Introducing Artificial Intelligence Into Trade and Commerce: The Case of Manufacturing Companies in West Africa
182 Amfiteatru Economic
Table no. 2. Demographic of Respondents Country 4
Variables Total Sample Country 1 Country 2 Country 3 Country 4
(N) Ghana Nigeria Togo Burkina Faso
2,903 738 984 529 652
2,903 738 984 529 652
(100%) (25.42%) (33.90%) (18.22%) (22.46%)
Gender:
Male 1,608 451 589 233 335
(55.39%) (15.54%) (20.29%) (8.03%) (11.54%)
Female 1,295 287 395 296 317
(44.61%) (9.89%) (13.61%) (10.20%) (10.92%)
Total l 2,903 738 984 529 652
Age:
Less than30 318 87 93 65 73
(10.95%) (2.97%) (3.20%) (2.24%) (2.51%)
31-40 542 134 197 80 131
(18.67%) (4.62%) (6.79%) (2.76%) (4.51%)
41-50 1,382 357 479 248 298
(47.61%) (12.30%) (16.50%) (8.54%) (10.27%)
51 and above 661 160 215 136 150
(22.77%) (5.51%) (7.41%) (4.68%) (5.17%)
Total 2,903 738 984 529 652
Marital Status:
Single 102 23 19 22 38
(3.51%) (0.79%) (0.65%) (0.76%) (1.31%)
Married 1,938 518 638 356 426
(66.76%) (17.84%) (21.98%) (12.26%) (14.67%)
Divorced 863 197 327 151 188
(29.73%) (6.79%) (11.26%) (5.20%) (6.48%)
Total 2,903 738 984 529 652
Educational
Qualification:
Technical/Professional 286 79 94 67 46
(9.85%) (2.72%) (3.24%) (2.31%) (1.58%)
Bachelor’s Degree 1,572 418 517 257 380
(54.15%) (14.40%) (17.81%) (8.85%) (13.09%)
Master 704 132 304 108 160
(24.25%) (4.55%) (10.47%) (3.72%) (5.51%)
PhD 341 109 69 97 66
(11.75%) (3.75%) (3.38%) (3.34%) (2.27%)
Total 2,903 738 984 529 652
Working Experience:
0-5 years 294 68 91 87 48
(10.13%) (2.34%) (3.13%) (3.00%) (1.65%)
6-10 years 415 109 123 95 88
(14.30%) (3.75%) (4.24%) (3.27%) (3.03%)
11-15 years 477 126 142 108 101
(16.43%) (4.34%) (4.89%) (3.72%) (3.48%)
16-20 years 663 103 276 106 178
(22.84%) (3.55%) (9.51%) (3.65%) (6.13%)
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 183
Variables Total Sample Country 1 Country 2 Country 3 Country 4
(N) Ghana Nigeria Togo Burkina Faso
21 and above 1,054 332 352 133 237
(36.31%) (11.44%) (12.13%) (4.58%) (8.16%)
Total 2,903 738 984 529 652
Specialization:
Textiles 458 104 153 83 118
(15.77%) (3.58%) (5.27%) (2.86%) (4.06%)
Tobacco 102 32 39 17 14
(3.51%) (1.10%) (1.34%) (0.59%) (0.48%)
Automotive 403 115 122 71 69
(13.88%) (3.96%) (4.20%) (2.45%) (2.38%)
Decorative items 283 69 93 58 63
(9.75%) (2.38%) (3.20%) (2.00%) (2.17%)
Essential oils 102 21 39 24 18
(3.51%) (0.72%) (1.34%) (0.83%) (0.62%)
Soap 143 43 58 19 23
(4.93%) (1.48%) (2.00%) (0.65%) (0.79%)
Electronics 119 32 49 17 21
(4.10%) (1.10%) (1.69%) (0.59%) (0.72%)
Aluminium 581 131 146 82 208
(20.01%) (4.51%) (5.03%) (2.82%) (7.17%)
Printing 210 44 62 59 45
(7.23%) (1.52%) (2.14%) (2.03%) (1.55%)
Cement 195 53 79 37 26
(6.72%) (1.83%) (2.72%) (1.27%) (0.89%)
Beer brewing 107 29 53 12 13
(3.69%) (1.00%) (1.83%) (0.41%) (0.45%)
Tomato Paste 116 45 59 32 20
(4.00%) (1.55%) (2.03%) (1.10%) (0.69%)
Cosmetics 84 20 32 18 14
(2.89%) (0.69%) (1.10%) (0.62%) (0.48%)
On the other hand, below statistics shows the summary description of the dataset taking into account the mean and standard deviation. On average, A.I has enhanced the way industry players effectively complete the task and product delivery at a weighted mean of 80.97 percent. This means that A.I has help improve efficiencies and augment our human capabilities with new products and processes in the manufacturing industry. Additionally, the use of A.I has facilitated good decision making process and also, build confidence in business transaction as a weighted average growth rate of 84 percent (approximately). Likewise, data management in artificial intelligence technology which saw a rate of 82 percent with its ability to stores lots of data in a structured electronic format. Moreover, the acquisition, installation and maintenance cost (fees) of the technological device together with its expertise support contributed equally to its usage with a weighted average of 78 percent and 80 percent respectively. On the basis of such device privacy protection scored an average figure of 73 percent approximately. Overall, the beneficial value of such technological device recorded a weighted average mark of 88 percent while respondents’ reasons for adopting such artificial intelligence technology in their business operations had a 90 percent acceptance rate. In sum, this portray that A.I is an integral part of our business system because it allows companies to design, produce and deliver products and service better than ever before. (Table no. 3)
AE Benefits and Risks of Introducing Artificial Intelligence Into Trade and Commerce: The Case of Manufacturing Companies in West Africa
184 Amfiteatru Economic
Table no. 3. Descriptive Statistics
Total Sample (N = 2,903)
Variable/Description
Mean SD Min Max
Usefulness
1. A.I enables quicker completion of tasks
87.45 0.07 26.19 102.38
2. A.I enhance my task effectively
79.62 0.18 43.02 93.82
3. A.I makes it easier to do my task
81.09 0.24 31.93 119.39
4. A.I improves my task performance
76.13 0.06 54.28 96.94
5. A.I save time and effort in performing task
80.54 0.12 21.94 94.63
Trust
6. A.I has influence my business performance
84.27 0.27 42.85 103.48
7. A.I has impacted on my personal life
73.16 0.43 51.94 97.03
8. A.I has change operations in the industry
82.43 0.05 40.48 110.32
9. A.I facilitate good decision making
90.18 0.24 53.95 100.53
10. A.I build confidence in business transactions
89.43 0.10 41.06 90.39
Data
11. A.I efficient in managing companies data
78.05 0.29 38.12 94.65
12. A.I ability to convert raw data into structured electronic form
89.33 0.08 54.93 98.32
13. A.I back-up storage device on working data
81.92 0.11 29.01 103.28
14. A.I collaborating software with other Apps.
73.54 0.24 32.09 96.49
15. A.I ability to deal with different versions of working data files
86.32 0.31 21.85 94.85
Experts
16. Need a technical person / professional on board
90.12 0.25 39.24 128.49
17. Requires frequent practical training on its usage
72.94 0.13 31.45 97.93
18. Handle relations with technicians and specialists
86.07 0.28 21.58 102.39
19. Check compliance and technical report on usage
70.54 0.37 37.29 91.37
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 185
20. Ensure periodic update of the digital tool
81.32 0.15 45.02 95.83
Cost (Fee)
21. A.I involves a lot of money to acquire it
84.03 0.54 54.39 95.43
22. A.I fee is reasonable / adequate
69.43 0.31 26.01 83.59
23. A.I functions justify its costs
73.91 0.18 49.29 91.38
24. A.I fee on maintenance, web servers and services
85.32 0.09 32.11 101.29
25. A.I cost of web design and/or software development
75.49 0.15 21.38 98.42
Privacy
26. Guidance towards collection, storing and processing
86.03 0.13 28.74 93.63
27. Reports unauthorized entry of hackers
71.24 0.09 32.03 89.02
28. Software access reviewed and consumer access rights updated
69.43 0.17 21.47 92.48
29. Ability to grant access only authorized staff
73.04 0.05 32.75 91.04
30. Routine review of stored data on business transactions
65.16 0.13 53.94 85.34
Value
31. There is value for money in terms of cost
83.95 0.38 30.48 92.48
32. There is value for money in terms of task performed
90.48 0.21 54.96 128.05
33. It is worthwhile because its saves time
93.23 0.13 23.04 97.39
34. Deliver good results in business transactions
81.09 0.19 41.38 90.06
35. Consumer and community value of its usage
89.37 0.15 32.95 94.17
Adoption
36. Intend to use it forever
95.83 0.21 24.85 107.49
37. Plan to recommend to others
84.39 0.03 32.94 97.06
38. Intend to subscribe for latest tech. version
92.05 0.27 41.02 94.39
39. Laws and policies reasonable on the adoption process
87.75 0.18 31.94 101.83
40. Beneficial functions outperform the risk attached
90.26 0.12 29.02 98.72
AE Benefits and Risks of Introducing Artificial Intelligence Into Trade and Commerce: The Case of Manufacturing Companies in West Africa
186 Amfiteatru Economic
Again, below table no. 4 shows the results of the multiple regression analyses. On the basis
of analyzing the benefit of introducing A.I into commerce, for instance; A.I improves task
performance across Ghana (β=0.84, p-value=0.01), Nigeria (β=0.82, p-value=0.01), Togo
(β=0.81, p-value=0.01) and Burkina Faso (β=0.88, p-value=0.01) as significantly related to
the usage of such technological tool. Hence, the acceptance of the proposed framework and
H1 (see above Figure no. 2). Secondly, the other variables, namely trust and data
management were both found to be significantly related to perceived value with an R-
squared of Ghana (0.804), Nigeria (0.853), Togo (0.724) and Burkina Faso (0.705).
Moreover, technical (experts) assistance on the use of A.I scored a significant mark of
Ghana (β=0.76, p-value=0.01), Nigeria (β=0.74, p-value=0.01), Togo (β=0.80, p-
value=0.05) and Burkina Faso (β=0.79, p-value=0.01). Likewise cost (fee) and privacy
protection that recorded a similar digits of introducing artificial intelligence into commerce.
Empirically, this supports H2, H3, H4 and H5 framework constructed (see above Figure no.
2). Finally, the value of A.I positively affects the adoption of such technological tool via its
ability to deliver good results in business transactions at respective rate of Ghana (β=0.85,
p-value=0.01), Nigeria (β=0.80, p-value=0.01), Togo (β=0.73, p-value=0.05) and Burkina
Faso (β=0.81, p-value=0.01). Also, the total sample regression results in table no. 5 affirms
the significant magnitude of introducing artificial intelligence into trade / commerce.
Clearly, the result estimates shows that artificial intelligence is helping companies of all
sizes and in all industries improve productivity and the bottom line at every stage of the
business lifecycle from sourcing material to sales and accounting to customer service. With
A.I, the study justifies that technology has become even more entangled into our daily
existence, workplaces and society in which we operates.
Table no. 4. Summary Regression Results
Ghana
Nigeria Togo Burkina Faso
Variable/Description β SE
β SE β SE β SE
Usefulness
1. A.I enables quicker completion of tasks 0.89 0.09
*** 0.91 0.14 *** 0.80 0.08 ** 0.82 0.12 ***
2. A.I enhance my task effectively 0.78 0.17 ***
0.79 0.23 ** 0.76 0.14 ** 0.76 0.09 ***
3. A.I makes it easier to do my task 0.90 0.05 ***
0.85 0.10 *** 0.81 0.06 ** 0.87 0.08 ***
4. A.I improves my task performance 0.84 0.19 ***
0.82 0.09 *** 0.81 0.12 *** 0.88 0.13 ***
5. A.I save time and effort in performing task 0.79 0.22 **
0.80 0.13 ** 0.83 0.06 *** 0.81 0.07 ***
Trust
6. A.I has influence my business performance 0.89 0.08 **
0.74 0.29 *** 0.69 0.27 ** 0.79 0.12 ***
7. A.I has impacted on my personal life 0.83 0.12 ***
0.79 0.11 *** 0.75 0.09 ** 0.67 0.08 **
8. A.I has change operations in the industry 0.81 0.20 **
0.90 0.08 *** 0.79 0.12 ** 0.80 0.24 ***
9. A.I facilitate good decision making 0.77 0.10 **
0.82 0.23 ** 0.81 0.15 ** 0.82 0.05 **
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 187
10. A.I build confident in business transactions 0.84 0.12 ***
0.84 0.20 ** 0.65 0.08 ** 0.76 0.13 ***
Data
11. A.I efficient in managing companies data 0.80 0.10 **
0.86 0.13 ** 0.59 0.19 ** 0.71 0.16 ***
12. Convert raw data into structured electronic form 0.79 0.27 **
0.76 0.21 ** 0.51 0.20 ** 0.84 0.20 **
13. A.I back-up storage device on working data 0.83 0.09
*** 0.70 0.09 *** 0.67 0.27 ** 0.80 0.18 **
14. A.I collaborating software with other Apps. 0.80 0.29 **
0.82 0.05 *** 0.61 0.18 ** 0.69 0.12 ***
15. Deal with different versions of working data files 0.79 0.12 **
0.89 0.07 ** 0.74 0.07 ** 0.81 0.30 **
Experts
16. Need a technical person / professional on board 0.71 0.25 **
0.69 0.15 ** 0.67 0.37 ** 0.65 0.27 **
17. Requires frequent practical training on its usage 0.76 0.17 ***
0.74 0.12 *** 0.80 0.23 ** 0.79 0.17 ***
18. Handle relations with technicians and specialists 0.85 0.13 **
0.61 0.06 ** 0.72 0.18 ** 0.69 0.25 **
19. Check compliance and technical report on usage 0.80 0.32 ***
0.79 0.10 *** 0.62 0.34 ** 0.71 0.08 ***
20. Ensure periodic update of the digital tool 0.85 0.23 **
0.81 0.23 ** 0.59 0.12 ** 0.82 0.14 **
Cost (Fee)
21. A.I involves a lot of money to acquire it 0.80 0.28
*** 0.79 0.23 ** 0.89 0.39 ** 0.78 0.07 ***
22. A.I fee is reasonable / adequate 0.82 0.23
*** 0.80 0.49 *** 0.78 0.48 ** 0.59 0.32 ***
23. A.I functions justify its costs 0.68 0.18 **
0.83 0.18 *** 0.83 0.36 ** 0.67 0.31 ***
24. A.I fee on maintenance, web servers and services 0.71 0.12 *
0.89 0.27 * 0.76 0.23 ** 0.69 0.09 *
25. A.I cost of web design and/or software development 0.69 0.09
*** 0.72 0.12 ** 0.81 0.021 ** 0.71 0.26 ***
Privacy
26. Guidance towards collection, storing and processing 0.67 0.31 **
0.87 0.06 ** 0.69 0.19 ** 0.76 0.19 ***
27. Reports unauthorized entry of hackers 0.77 0.32
** 0.74 0.23 ** 0.83 0.28 ** 0.69 0.38 **
28. Access reviewed and consumer access rights updated 0.80 0.23
*** 0.80 0.11 ** 0.71 0.07 ** 0.55 0.29 **
29. Ability to grant access only authorized staff 0.85 0.21 *
0.74 0.08 ** 0.76 0.43 * 0.80 0.08 *
30. Routine review of stored data on business transactions 0.81 0.11
** 0.76 0.39 ** 0.70 0.09 ** 0.68 0.15 *
Value
31. There is value for money in terms of cost 0.84 0.18
*** 0.89 0.14 *** 0.85 0.29 ** 0.60 0.06 **
32. There is value for money in terms of task performed 0.89 0.12
*** 0.90 0.29 *** 0.82 0.11 ** 0.80 0.14 ***
33. It is worthwhile because its saves time 0.91 0.10
** 0.82 0.26 ** 0.79 0.37 ** 0.75 0.14 **
AE Benefits and Risks of Introducing Artificial Intelligence Into Trade and Commerce: The Case of Manufacturing Companies in West Africa
188 Amfiteatru Economic
34. Deliver good results in business transactions 0.85 0.14
*** 0.80 0.15 *** 0.73 0.19 ** 0.81 0.39 ***
35. Consumer and community value of its usage 0.79 0.09
** 0.89 0.12 ** 0.67 0.24 ** 0.83 0.06 **
Adoption
36. Intend to use it forever 0.80 0.13
** 0.87 0.25 ** 0.80 0.13 ** 0.87 0.14 **
37. Plan to recommend to others 0.79 0.10
*** 0.74 0.12 *** 0.78 0.09 ** 0.75 0.17 **
38. Intend to subscribe for latest tech. version 0.81 0.09
** 0.68 0.09 ** 0.82 0.42 ** 0.86 0.23 **
39. Laws and policies reasonable on the adoption process 0.90 0.21
*** 0.79 0.13 *** 0.71 0.08 ** 0.79 0.09 ***
40. Beneficial functions outperform the risk attached 0.86 0.13
*** 0.84 0.11 *** 0.76 0.15 ** 0.81 0.12 ***
F-Statistics (p-value) 71.05 (0.00)
81.43 (0.00) 68.82 (0.00) 73.86 (0.00)
R-squared 0.804
0.853 0.724 0.705
Adjusted R-squared 0.732
0.798 0.691 0.672
Durbin-Watson 2.09
1.98 1.86 2.05
*p < 0.1; **p < 0.05; ***p < 0.01.
Table no. 5. Overall Regression Results
Total Sample (N) = 2,903
Variable/Description
βeta Standard Error
Usefulness
1. A.I enables quicker completion of tasks 0.83
0.13 ***
2. A.I enhance my task effectively 0.91
0.08 ***
3. A.I makes it easier to do my task 0.90
0.23 ***
4. A.I improves my task performance 0.86
0.11 ***
5. A.I save time and effort in performing task 0.82
0.09 **
Trust
6. A.I has influence my business performance 0.80
0.21 **
7. A.I has impacted on my personal life 0.84
0.07 ***
8. A.I has change operations in the industry 0.93
0.13 **
9. A.I facilitate good decision making 0.72
0.10 **
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 189
10. A.I build confident in business transactions
0.81 0.13 ***
Data
11. A.I efficient in managing companies data 0.73
0.12 **
12. Convert raw data into structured electronic form 0.86
0.05 **
13. A.I back-up storage device on working data 0.83
0.17 ***
14. A.I collaborating software with other Apps. 0.86
0.21 **
15. Deal with different versions of working data files 0.74
0.14 **
Experts
16. Need a technical person / professional on board 0.84
0.15 **
17. Requires frequent practical training on its usage 0.79
0.27 ***
18. Handle relations with technicians and specialists 0.92
0.09 **
19. Check compliance and technical report on usage 0.73
0.20 ***
20. Ensure periodic update of the digital tool 0.88
0.07 **
Cost (Fee)
21. A.I involves a lot of money to acquire it 0.79
0.06 ***
22. A.I fee is reasonable / adequate 0.86
0.21 ***
23. A.I functions justify its costs 0.73
0.13 **
24. A.I fee on maintenance, web servers and services 0.82
0.11 **
25. A.I cost of web design and/or software development 0.79
0.04 ***
Privacy
26. Guidance towards collection, storing and processing 0.72
0.15 **
27. Reports unauthorized entry of hackers 0.69
0.21 **
28. Access reviewed and consumer access rights updated 0.88
0.29 ***
29. Ability to grant access only authorized staff 0.83
0.03 **
30. Routine review of stored data on business transactions 0.90
0.10 **
Value
31. There is value for money in terms of cost 0.76
0.23 ***
32. There is value for money in terms of task performed 0.80
0.08 ***
33. It is worthwhile because its saves time 0.86
0.20 **
AE Benefits and Risks of Introducing Artificial Intelligence Into Trade and Commerce: The Case of Manufacturing Companies in West Africa
190 Amfiteatru Economic
34. Deliver good results in business transactions 0.77
0.12 **
35. Consumer and community value of its usage 0.81
0.09 ***
Adoption
36. Intend to use it forever 0.84
0.17 **
37. Plan to recommend to others 0.76
0.22 **
38. Intend to subscribe for latest tech. version 0.91
0.18 **
39. Laws and policies reasonable on the adoption process 0.86
0.04 ***
40. Beneficial functions outperform the risk attached 0.90
0.10 ***
F-Statistics (p-value) 78.23
(0.00)
R-squared 0.842
Adjusted R-squared 0.838
Durbin-Watson 2.29
*p < 0.1; **p < 0.05; ***p < 0.01.
In support of above regression test conducted to examine the direct effects of the seven
hypothesis proposed including overall value of introducing artificial intelligence into
trade/commerce. Below table no. 6 result affirms that value is significant at critical vale of
2.493 (p-value=0.009). Similar with the adoption use of technological tool at a statistically
significant critical value of 3.072 (p-value=0.018). Likewise, all other benefits and sacrifice
variables as shown below (see Table no. 5). However, the study conclude at accepting the
seven hypothesis because perceived value reflects the overall comparison between benefit
and risk (sacrifice) in the use of artificial intelligence into trade/commerce using West-
African manufacturing and production companies as a case study.
Table no. 6: Pearson Chi-square
No Variable Critical Value Asymp. Sig. (2-sided)
1 Usefulness 6.018 0.090
2 Trust 0.931 0.015
3 Data 2.485 0.032
4 Experts 5.013 0.006
5 Cost 4.485 0.028
6 Privacy 3.102 0.011
7 Value 2.493 0.009
8 Adoption 3.072 0.018
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 191
Conclusions
This paper explores the benefits and risks of introducing A.I. into commerce using the
manufacturing and production companies in West-African as a case study. In order to
achieve the goals, we identified key influencing factors that affect the companies’ adoption
based on value-based adoption model (VAM) using seven key number of hypotheses. Also,
introduction of A.I. is a prerequisite for the adoption and proliferation of digital
technologies into Commerce. According to the results, the seven variables were found to be
significantly related to perceived value on the use of A.I. into trade/commerce. The findings
of this study shows that with A.I technologies, companies can smartly and efficiently scan a
lot of data. This business to customer (B2C) services help ease customer behavior by
offering relevant solutions for each consumer. Moreover, manufacturing companies are
now able to create shopper that assist customers online. Literally, this is similar to a
physical store in real time that assist customers to purchase their products. Also,
introducing A.I into commerce supports “round-the-clock” services. This means that there
is 24/7 shopping services providing customers with assistance via the buying process.
Furthermore, such technological introduction helps commerce industry predicts the
shopping patterns based on what customer buy and when they buy them. Such A.I digital
assistant, for instance, in the business-to-business (B2B) transactions are driving lots of
innovative solutions. For example, A.I enables supply chain automation that enables
effective management in respect to vendors, delivery schedules and market needs. On the
other hand, the study affirms the proposed framework since A.I systems have the ability to
learn and/or adapt as they make decisions, which in returns generates substantial economic
and social benefits. Additionally, the results show how businesses are implementing A.I to
improve retail standards, customer experience and revenue and fast delivery processing of
commodities. For instance, the emergence of COVID-19 broke most of the transportation
links and distribution in a global context as the world was hammered by one of the greatest
interruptions in modern history. Nevertheless, the implications of the COVID-19 pandemic
on A.I’s future present opportunities for A.I to mitigate such canker through the provision
of automated products manufacturing, distribution and sustainability. This, in turn, makes
market places and streets less crowded and also, better support measures such as social
distancing while performing business transactions online. This will go a longer way to
increase the reactivity and resilience of complex global products supply chain. Moreover,
A.I based predictive mechanism can help forecast customer demand, shortages and
bottlenecks before they occur. Such A.I tool when deployed assist firms with manufacturing
warehouses, distribution centers and consumer markets around the globe to predict pressure
points and boldly shift their human resources and inventory levels to meet market demands.
A.I does not require social distancing and may also offer attractive alternative for some
tasks that were previously undertaken by human workers.
In a nutshell, A.I. has been deployed to enhance human activities. Moreover, this empirical
study appreciates the significant role A.I. is playing as a leading mechanism in driving
innovative solutions and customer experiences in areas such as personalized shopping,
product recommendations, and inventory management.
The increasing penetration of A.I technologies into many aspects of business decision
making processes raises lots of concerns and ethical issues. However, there is the need for
A.I to observe every customer interaction related to the business. Future studies can be
extended by examining other factors that could influence the adoption of A.I. into other
AE Benefits and Risks of Introducing Artificial Intelligence Into Trade and Commerce: The Case of Manufacturing Companies in West Africa
192 Amfiteatru Economic
sectors of the economy and how to evaluate the rate of acceptance within the spheres of
exchanging products for business purposes.
Acknowledgements
This study was fully supported by Key Soft Science Projects of Guangdong Province (No.
2019B101001016)
References
Agrawal, A., Gans, J. and Goldfarb, A., 2018. Prediction machines: the simple economics
of artificial intelligence. Boston: Harvard Business Review Press.
Ahadiat, N., 2008. Technologies used in accounting education: A study of frequency of use
among faculty. Journal of Education for Business, 83(3), pp.123-134.
Alzahrani, J., 2019. The impact of e-commerce adoption on business strategy in Saudi
Arabian small and medium enterprises (SMEs). Review of Economics and Political
Science, [e-journal] 4(1), pp.73-88. https://doi.org/10.1108/REPS-10-2018-013
Armbrust, C., Braun, T., Föhst, T., Proetzsch, M., Renner, A., Schäfer, B. H. and Berns, K.,
2011. RAVON: The robust autonomous vehicle for off-road navigation. In:Y. Baudoin
and M.K. Habib eds., 2011. Using Robots in Hazardous Environments: Landmine
Detection, De-Mining and Other Applications. S.l: Woodhead Publishing, pp. 353-396.
Barrett, M., Branson, L., Carter, S., DeLeon, F., Ellis, J., Gundlach, C. and Lee, D., 2019.
Using Artificial Intelligence to Enhance Educational Opportunities and Student Services
in Higher Education. Inquiry: The Journal of the Virginia Community Colleges, 22(1),
p.11.
Benko, A. and Lányi, C.S., 2009. History of artificial intelligence. In: M. Khosrow-Pour
ed., 2009. Encyclopedia of Information Science and Technology. 2nd ed. S.l: IGI
Global, pp.1759-1762.
Blanchet, J., Kang, Y. and Murthy, K., 2019. Robust Wasserstein profile inference and
applications to machine learning. Journal of Applied Probability, 56(3), pp.830-857.
Bolloju, N., Khalifa, M. and Turban, E., 2002. Integrating knowledge management into
enterprise environments for the next generation decision support. Decision Support
Systems, 33(2), pp.163-176.
Calo, R., 2017. Artificial Intelligence policy: a primer and roadmap. UCDL Rev., 51, p.399.
Canbek, N.G. and Mutlu, M.E., 2016. On the track of artificial intelligence: Learning with
intelligent personal assistants. Journal of Human Sciences, 13(1), pp.592-601.
Cronbach, L.J., 1951. Coefficient alpha and the internal structure of
tests. Psychometrika, 16(3), pp.297-334.
Cunneen, M., Mullins, M. and Murphy, F., 2019. Artificial intelligence assistants and risk:
framing a connectivity risk narrative. AI & SOCIETY, 35, pp.625-634.
Dautenhahn, K., 2007. Socially intelligent robots: dimensions of human–robot
interaction. Philosophical transactions of the royal society B: Biological
sciences, 362(1480), pp.679-704.
Artificial Intelligence in Wholesale and Retail AE
Vol. 23 • No. 56 • February 2021 193
Davenport, T.H., 2018. From analytics to artificial intelligence. Journal of Business
Analytics, 1(2), pp.73-80.
Davis, F.D., Bagozzi, R.P. and Warshaw, P.R., 1989. User acceptance of computer technology:
a comparison of two theoretical models. Management Science, 35, pp.982-1003.
Deci, E.L., 1971. Effects of externally mediated rewards on intrinsic motivation. Journal of
personality and Social Psychology, 18(1), p.105.
Devaraj, S. and Kohli, R., 2003. Performance impacts of information technology: Is actual
usage the missing link? Management science, 49(3), pp.273-289.
Ekufu, T.K., 2012. Predicting cloud computing technology adoption by organizations: An
empirical integration of technology acceptance model and theory of planned behavior.
PhD. Capella University.
Foss, N.J., 2005. Strategy, economic organization, and the knowledge economy: the
coordination of firms and resources. Oxford: Oxford University Press.
Gefen, D., 2002. Reflections on the dimensions of trust and trustworthiness among online
consumers. ACM SIGMIS Database: the DATABASE for Advances in Information
Systems, 33(3), pp.38-53.
Golden, J.A., 2017. Deep learning algorithms for detection of lymph node metastases from
breast cancer: helping artificial intelligence be seen. Jama, 318(22), pp.2184-2186.
Haenlein, M. and Kaplan, A., 2019. A brief history of artificial intelligence: On the past,
present, and future of artificial intelligence. California management review, 61(4), pp.5-14.
Huang, M.H. and Rust, R.T., 2018. Artificial intelligence in service. Journal of Service
Research, 21(2), pp.155-172.
Jarvenpaa, S.L. and Shaw, T.R., 1998. Global virtual teams: Integrating models of
trust. Organizational Virtualness, pp.35-52.
Kim, H.W., Chan, H.C. and Gupta, S., 2007. Value-based adoption of mobile internet: an
empirical investigation. Decision support systems, 43(1), pp.111-126.
Lee, J., Davari, H., Singh, J. and Pandhare, V., 2018. Industrial Artificial Intelligence for
industry 4.0-based manufacturing systems. Manufacturing letters, 18, pp.20-23.
Lin, T.C., Wu, S., Hsu, J.S.C. and Chou, Y.C., 2012. The integration of value-based
adoption and expectation–confirmation models: An example of IPTV continuance
intention. Decision Support Systems, 54(1), pp.63-75.
McCorduck, P., Minsky, M., Selfridge, O.G. and Simon, H.A., 1977. History of Artificial
Intelligence. In: s.n., The 5th international joint conference on Artificial intelligence.
S.l., August 1977. S.l:S.n.
Moore, G.C. and Benbasat, I., 1991. Development of an instrument to measure the
perceptions of adopting an information technology innovation. Information systems
research, 2(3), pp.192-222.
Sanzogni, L., Guzman, G. and Busch, P., 2017. Artificial intelligence and knowledge
management: questioning the tacit dimension. Prometheus, 35(1), pp.37-56.
Sutton, J. and Trefler, D., 2016. Capabilities, wealth, and trade. Journal of Political
Economy, 124(3), pp.826-878.
AE Benefits and Risks of Introducing Artificial Intelligence Into Trade and Commerce: The Case of Manufacturing Companies in West Africa
194 Amfiteatru Economic
Syam, N. and Sharma, A., 2018. Waiting for a sales renaissance in the fourth industrial
revolution: Machine learning and artificial intelligence in sales research and
practice. Industrial Marketing Management, 69, pp.135-146.
Thaler, R., 1985. Mental accounting and consumer choice. Marketing science, 4(3),
pp.199-214.
Venkatesh, V., Morris, M.G., Davis, G.B. and Davis, F.D., 2003. User acceptance of
information technology: toward a unified view. MIS Quarterly, 27(3), pp.425-478.
Wiljer, D. and Hakim, Z., 2019. Developing an artificial intelligence–enabled health care
practice: rewiring health care professions for better care. Journal of medical imaging
and radiation sciences, 50(4), pp.S8-S14.
Zeithaml, V.A., 1988. Consumer perceptions of price, quality, and value: a means-end
model and synthesis of evidence. Journal of marketing, 52(3), pp.2-22.
Vol. 23 • No. 56 • February 2021 195
Contents
Economic Interferences
The Impact of COVID-19 on Romanian Tourism. An Explorative Case Study
on Prahova County, Romania ......................................................................................... 196
Christine Volkmann, Kim Oliver Tokarski, Violeta Mihaela Dincă and Anca Bogdan
On the Determinants of Fiscal Decentralization: Evidence From the EU ................... 206
Francisco J. Delgado
Food Quality Competition Among Companies and Government Food Safety
Supervision Under Asymmetric Product Substitution .................................................. 221
Ningzhou Shen, Yinghua Song, Dan Liu and Dalia Streimikiene
Are Positive and Negative Outcomes of Organizational Justice Conditioned
by Leader–Member Exchange? ...................................................................................... 240
Or Shkoler, Aharon Tziner, Cristinel Vasiliu and Claudiu-Nicolae Ghinea
AE The Impact of COVID-19 on Romanian Tourism . An Explorative Case Study on Prahova County, Romania
196 Amfiteatru Economic
THE IMPACT OF COVID-19 ON ROMANIAN TOURISM.
AN EXPLORATIVE CASE STUDY ON PRAHOVA COUNTY, ROMANIA
Christine Volkmann1, Kim Oliver Tokarski2, Violeta Mihaela Dincă3*
and Anca Bogdan4 1) University of Wuppertal, Wuppertal, Germany
2) Bern University of Applied Sciences, Bern, Switzerland 3)4) Bucharest University of Economic Studies, Bucharest, Romania
Please cite this article as:
Volkmann, C., Tokarski, K.O., Dincă, V.M, Bogdan,
A., 2021. The Impact of COVID-19 on Romanian
Tourism. An Explorative Case Study on Prahova
County, Romania. Amfiteatru Economic, 23(56),
pp. 196-205.
DOI: 10.24818/EA/2021/56/196
Article History
Received: 14 September 2020
Revised: 29 October 2020
Accepted: 30 November 2020
Abstract
Europe stands first in the world ranking of tourist destinations. Tourism is at the heart of a
huge business ecosystem that contributes substantially to prosperity and job creation in all
Member States of the European Union. The COVID-19 epidemic puts the EU tourism
industry under unprecedented pressure. It has led to the suspension of most domestic and
international travel, causing a significant reduction in revenue and creating liquidity
problems for all tour operators. Both travellers and businesses face uncertain prospects. The
goals of this research focused on the main directions: 1) to assess the main concerns for the
tourism industry in Romania in the autumn of 2020 and 2) to identify immediate measures
to support the Romanian tourism sector. A questionnaire was applied to acquire the
viewpoints of the sixteen representatives of eleven prestigious hotels in the mountain region
of Valea Prahovei in Romania. The outcomes show their considerations on lifting travel
restrictions, restoring traveller confidence and rethinking the Romanian tourism sector for
the future.
Keywords: tourism industry, hotel industry, Romania, COVID-19 pandemic.
JEL Classification: Z32; L83; R11
* Corresponding author, Violeta Mihaela Dinca – e-mail: [email protected].
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 197
Introduction
The COVID-19 pandemic has had a meaningful influence on the tourism industry as a
result of travel restrictions together with a collapse in demand from the part of potential
tourists. The travel businesses have been greatly hit by the spread of COVID-19, because a
lot of states have instituted travel constraints trying to keep the spread under control, which
lead to a stop of a whole sector (BBC, 2020; Brouder, 2020; Gössling, Scott and Hall,
2020). The United Nations World Tourism Organization appraised at the end of June 2020
that international tourist arrivals around the globe could sink by a third in 2020, triggering
an inherent financial harm of around 40 billion$ (UNTWO, 2020a). BBC also revealed that
in multiple areas of the world, organised trips decayed by 90% (BBC, 2020). This would
lead to a transformation of tourism, especially in a COVID-19 setting which triggers the
question if the tourism sector will be able to recover or if this might be the “end of tourism”
as we know it or a “re-discovery” of tourism (Brouder, 2018; Brouder, 2020;
Niewiadomski, 2020). Divergent and one-sided travel boundaries happened locally and
plenty tourist sites of visitation across the globe, like museums, amusement parks, and
sports venues stopped operating (Kliger and Silberzweig, 2020; The New York Times,
2020). Reflecting on these recent facts, the research questions of this article were the
following:
What are the implications of the COVID-19 crisis on the tourism sector for Romania?
What are the recommendations of the interviewed experts for the stakeholders of the
tourism industry in order to avert the worst effects and facilitate recovery?
1. Literature review
1.1. The downturn of tourism under COVID-19 at international level
The state of facts in September 2020 was that COVID-19 infected more than 10 million
citizens and provoked the death of 500,000 people globally (WHO, 2020a). On a
worldwide level, the spread does not display signs of slowing down and consequently, the
vast majority of states have locked their frontiers for foreign arrivals. The UN World
Tourism Organization communicated in August 2020 that 100 per cent of global
destinations implemented travel constraints and limitations, with international tourism
becoming almost entirely interrupted, and domestic tourism severely diminished because of
lockdown stipulations enforced in countless countries. Despite the fact that a few
destinations have begun gradually to open up, a large number of travellers are still hesitant
in coming back to international travel or can no longer meet the expenses given the
economic critical situation (UNWTO, 2020b).
Tourism is one of the quickest expanding economic areas and is a notable catalyser for
economic progress. In 2018 tourism receipts amounted to $1,480 billion, an increase by 4.4.
per cent, while tourism exports accounted for seven per cent of global trade in goods and
services (UNWTO, 2020c). In the same time tourism is a dominant source of jobs on a global
status, with a big share of the jobs being handled by women (54 per cent), which substantially
larger than in most other industries, and also by young workers, making the sector as an
inclusive one. There is also a meaningful quantity of indirect work engagement in construction
and infrastructure development, along with providing food, drink and souvenirs to tourists
(Glaser-Segura, Nistoreanu and Dincă, 2017; Vo, Chovancova and Tri, 2019).
AE The Impact of COVID-19 on Romanian Tourism . An Explorative Case Study on Prahova County, Romania
198 Amfiteatru Economic
A major problem within this current context is that in this industry many workers have
direct contact with tourists for example in travel agencies, airlines, ships, hotels,
restaurants, shopping centres and various tourist attractions and it is commonly concurred
that COVID-19 is easily transmissible (even though the fatality scale is small by contrast
with previous pandemics and its fatalities are being very often encountered at aged people
and those with a precarious medical history).
To illustrate the potential impact of the decline in the tourism sector, three scenarios were
simulated by the United Nations Conference on Trade and Development and described in
table no.1. The scenarios, Moderate (optimistic), Intermediate and Dramatic (pessimistic),
vary in the length of international tourism absence. The scenario Intermediate is closest to
the assessment of the UNWTO (2020b) that international tourist numbers could fall by 60
to 80 per cent in 2020 and with a reduction by 66 per cent of tourism expenditure
(UNCTAD, 2020).
Table no. 1. Potential impact of the decline in the tourism sector – three scenarios
from UNCTAD
Scenario type Characterisation
Moderate 1/3 of annual inbound tourism spending is eliminated in every country
This corresponds with 4 months impasse/ cessation of international tourism
or a 80% fall for 5 months
Intermediate 2/3 of annual inbound tourism spending is eliminated in every country
This corresponds with 8 months impasse/ cessation of international tourism
or a 80% fall for 10 months
Dramatic All annual inbound tourism spending is eliminated in every country
This corresponds with 12 months impasse/ cessation of international tourism
Source: UNCTAD, 2020. COVID-19 and tourism – assessing the economic consequences.
1.2. The problematic context for Romanian tourism under COVID-19
The tourism industry is, by definition, one of the most vulnerable industries when it comes
to threats related to an economic, military or medical crisis. COVID-19 has produced
dramatic effects for the tourism industry, and Romanian businesses have not been
bypassed. Romanian tourism agencies reported that this year their businesses dropped by
80-90% compared to previous years (Biz, 2020).
An important NGO within the Romanian tourism industry (the Alliance for Tourism) stated
that even though the contribution of tourism within the GDP for Romania suffered a severe
shrinkage in 2020, with the appropriate measures this shortcoming could be surpassed by
2025 (Incoming Romania, 2020) (Figure no. 1).
The measures set by the members within this NGO for tourism revival are: Supporting and
prioritizing the tourism sector within the national economy, digitization and implementing
new technologies in tourism, sustainability and development of sustainable tourism,
highlighting Romania's competitive advantages through smart promotion, tourism
development by encouraging public and private investments, supporting participatory
tourism through the use of public-private partnerships, reducing tax evasion in tourism and
administrative efficiency, and the forecast for the contribution of tourism within the
Romanian GDP if these measures are implemented can be seen above within figure no.1.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 199
Figure no. 1. Forecast of the tourism contribution within the Romanian GDP
Source: Incoming Romania, 2020. Solutii de Organizare si Sustinere a Turismului
Romanesc – Alianta Pentru Turism.
2. Research methodology
The aims of this research pursued two paths: 1) to identify the potential effects of the
tourism challenging period for Romania and 2) to examine what can the stakeholders from
the Romanian tourism industry do to mitigate the tough effects of lack of international
tourism during and after the COVID-19 pandemic.
These two aims of the paper were fulfilled by developing a qualitative research, the most
applicable technique for this topic, as qualitative research seeks to avoid making
generalizations or grand claims, and is often characterized by a high level of reflectivity
and sensitivity to power relations and ambiguity (Bregnholm Ren, 2016).
Within the specific methods of qualitative research, semi-structured interviews were
applied for this particular study. Semi-structured interviews are frequently put into practice
for qualitative research and are a regular qualitative data source in health services’ research.
This approach usually involves a dialogue between researcher and participant, overseen by
adaptable interview standards and expanded by follow-up questions, inquiries and
comments. This arrangement permits the researcher to gather open-ended data, to examine
participant thoughts, impressions and beliefs about a specific subject and to delve
profoundly into personal and at times sensitive matters (DeJonckheere and Vaughn, 2019).
Considering that our goal was to comprehend how the tourism particularly in Romania is and
will be shaped by COVID-19, we took into account that this method of the semi-structured
interviews was the most favorable. We had the chance to meet and interview sixteen
representatives of eleven renowned hotels from four tourism resorts within the Prahova County
in Romania: Sinaia (six participants from four hotels), Bușteni (four participants from three
hotels), Azuga (three participants from two hotels) and Slănic Prahova (three participants from
two hotels) and we gathered their ideas and know-how on how the Romanian tourism can
survive the COVID-19 pandemic and eventually be relaunched. The Prahova Valley was
chosen for the case study as it is one of the areas with the highest tourist potential in the
country owning some of the best tourism facilities in Romania.
AE The Impact of COVID-19 on Romanian Tourism . An Explorative Case Study on Prahova County, Romania
200 Amfiteatru Economic
Relying on the study’s research questions and on the objectives of this qualitative research,
we designed an interview agenda which emphasized the changes within the tourism sector
in Romania and its future challenges. For this particular research, four open questions were
crafted for the interviews so that the problems examined could be shaped: the state and the
provocations within the tourism area in Romania. These interviews with experts from the
hotel industry are suitable because they can offer a multilayered, deep apprehension and
perception into the turn of events from the Romanian tourism field.
Therefore, the interview guide was built pivoting on some notable topics: how can
Romanian tourism, struck hard by the COVID-19 pandemic, come back to normal and how
could this “new normal” look like and be sustained by the stakeholders involved.
The specialists who took part in our research by answering at the questions from the
interview guide (table no. 2) constitute an assembly of value of sixteen experts from
managerial staff of eleven hotels with history and tradition from four resorts of big
importance for the Romanian tourism.
Table no. 2. Interview guide applied in the study
No. Question
1. What are the most important effects of the COVID-19 pandemic in the field of tourism in
Romania?
2. What changes have the businesses you are connected with suffered within the summer of
2020 (hotels, tourism agencies) in terms of indicators, personnel, working points/ units?
3. How do you estimate that the consumption habits of tourists visiting Romania (foreign
and Romanians) will change in terms of tourism services following the COVID-19
pandemic?
4. What measures do you think could be the most effective from the government and state
institutions for the field of tourism to help your business and the industry as a whole?
All the experts who were included hold a more than twenty years’ experience not only in
the hotel industry but also within the tourism area as a whole, having tight connections with
important actors in the sector on both national and local levels such as: the National
Authority for Tourism (ANT), found under the subordination of the Ministry for Economy,
the National Association for Tourism Agencies (ANAT) along with national and local
tourism promotion offices, airline, transport and rent-a-car companies, restaurants, online
booking systems, insurance companies, educational institutions, the and other associations
with activity in tourism. Therefore all our sixteen participants in the study offered relevant
and significant input in order to tailor wider and extensive clarifications at the questions
which were addressed to them; also, because the study was very close related to the
COVID-19 pandemic all data was collected in June 2020.
3. Findings and discussions
The results divulge the opinions of the sixteen experts included in the study about the
current situation and the immediate impact that COVID-19 has on the tourism businesses,
be it travel agencies, HoReCa (hospitality industry) owners, tourism institutions or event
organization companies. Their answers are also gathered in short version within table no. 3
two pages later in this paper.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 201
With the help of our first question in the interview guide, the objective was to find out what are the consequences of the COVID-19 pandemic for the Romanian tourism. All the answers coming from the respondents centred around the idea that the global tourism industry has been devastated by the COVID-19 pandemic, with hundreds of billion $ in export losses in the first five months of the year and more than a hundred million jobs at risk. When taking the discussion at national level, some points of view were expressed by more than 80% of the respondents for example that after several weeks of lockdown, taken to curb the spread of the COVID-19 pandemic, Europe is gradually easing quarantine measures, trying to return to normal. Obviously, to a different normal, to a normality that, at least for a while, for a few months or even more, will be marked by the current pandemic and the crisis it has generated in the whole world. Tourism is one of the sectors hardest hit by the crisis – with a lower or higher share of GDP, varying from country to country. 70% of the respondents recognised that they were all in the phase in which, on the eve of the summer season (as mentioned the interviews were taken in June 2020), all the states of the world (Romania included naturally) were trying to relaunch their tourism, elaborating, cautiously, strategies capable of saving millions of jobs and offering optimal conditions, in terms of safety, to those eager for holidays. 10 out of our 16 respondents mentioned that all their partners from airlines to tour agencies offered the possibility to the clients to change travel dates or receive travel vouchers valid for one or two years. Five respondents brought into discussion the cultural events within tourism resorts – a major attraction for tourists in general and the fact that all cultural events have so far been cancelled, rescheduled or moved in open air spaces. Prestigious summer theatre festivals fell victim to the pandemic and took place in 2020 either in an online form either in open air (but with some limitations compared to proper closed theatre halls). As a conclusion the future for the tourism industry will not be easy for both tourists and tourism services’ providers and the holidays will be completely different in 2020 compared with previous years and even two-three years ahead.
Focusing on the second question addressed to our participants, which underlined the modifications that have the businesses with which the respondents are connected, suffered within the summer of 2020 (hotels, tourism agencies) in terms of indicators, personnel, working points/ units, it was evident that the tourism industry is, by definition, one of the most vulnerable industries when it comes to threats related to an economic, military or medical crisis. 14 respondents admitted that COVID-19 has produced dramatic effects for the tourism industry, and their business and those of their partners have not been bypassed, in April and May 2020 being recorded a drop between 75-90%. 10 respondents specified that some high-capacity accommodation units and numerous restaurants with no outdoor space have ceased operations (for example Hotel Caraiman from Sinaia part of the Palace-Caraiman complex was shut down in the summer of 2020 only the Palace Hotel being active). Regarding the employees within the hotels, the respondents recognised that around 30% of them were sent to technical unemployment as the number of tourists did not demand such a high number of staff. Some of the respondents (nine) also had information about the situation of the employees of their partners (travel agencies and companies operating in the tourism field) that in their case in the summer of 2020 they have chosen to resort to measures such as online working/ working from home, salary reductions and unfortunately they were forced to send some of the staff to technical unemployment. Within this question it was the first time when it was noticed that opinions were divided regarding the expenses with personnel as some respondents stated that 2020 was the moment when they decided to stop all investments in technology and to direct them to employee assistance and to sales, while another part of the respondents were adamant in stating their decision to focus their advertising on online promotion campaigns which demanded some investment.
AE The Impact of COVID-19 on Romanian Tourism . An Explorative Case Study on Prahova County, Romania
202 Amfiteatru Economic
The participants were also given a third question in the interview centring on their
estimation about the consumption habits of tourists coming to Romania: will they change in
terms of tourism services following the COVID-19 pandemic? Thirteen of the respondents
mentioned that the most dangerous thing that the potential tourists could feel is the fear of
traveling to areas other than comfort and they believed that tourists will think very carefully
about the future if they dare to travel because it will take a long time for the world to gain
confidence in traveling; twelve respondents stated that tourism is necessary not only in
terms of economic activity, but for human beings in general who miss relaxation -
especially after a period that has seriously tested the psyche. Tourism can be the way to
recharge batteries and overcome these difficult moments. Ten respondents think that the
most searched forms of tourism in 2020 will be those of family or individual tourism and
that people will look for destinations that are not very crowded, for green, sustainable
destinations, they will focus more on ecotourism and above all for secure tourism. The key
criteria in choosing this year's holiday destination will be first and foremost: health, non-
congested areas, quality, ecotourism, with holiday homes and apartments, guesthouses and
smaller hotels preferred. There could be also within the next years an increasing interest for
isolated tourism: Danube Delta, some parts of Transylvania, Bucovina, Maramureş which
could be important assets that Romania must take advantage of in this period.
When tackling the measures that could be the most effective from the government and state
institutions for the field of tourism to help the businesses and the industry as a whole (the
focus of the fourth interview question), the considerations coming from the respondents
were substantial and included the following insights: supporting the tourism workforce
after the end of the state of alert by prolonging the technical unemployment for the units /
activities that cannot be resumed on June 15, 2020, until the activities can be restarted;
conferring grants to tourism enterprises affected by COVID-19, as well as granting of
subsidized working capital loans and / or investment loans to SMEs and large tourism
companies; immediate payment of arrears to tourism companies (sick leave, VAT, overdue
bills); reduction of local taxes, proportional to the period in which the tourist units were
closed and postponement of these obligations until the end of the year, without penalties
(taxes and tax on buildings, taxes and tax on land, taxes on promotion and advertising,
taxes and taxes on means transport); establishment and approval of health safety standards
approved by industry, MoH and MEEMA applicable in the units of the tourism industry in
order to significantly reduce the risks of infection with COVID-19 virus, during vacation /
travel in Romania; partial or full support, by the state, of the additional costs generated by
the additional hygienic-sanitary measures that will have to be taken by the accommodation,
food, treatment, etc. units in order to resume the activity (sanitation, additional materials
and equipment, etc.); stimulating the tourist circulation by granting, in the shortest possible
time, holiday vouchers related to 2020 and maintaining them for the next years as well.
Table no. 3. Assembly of the answers at the four questions addressed
to the interviewed tourism experts Questions Answers
1. Mention the most
relevant effects of the
COVID-19 pandemic for
the field of tourism in
Romania
* the global tourism industry is in severe decline because of the
Covid-19 pandemic with important financial losses and jobs at risk
*around the world all stakeholders involved within the tourism
industry are trying to relaunch their businesses and find “a new
normal”
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 203
Questions Answers
2. Point out the changes
that the businesses you are
connected with suffered
within the summer of 2020
in terms of indicators,
personnel, working points/
units
*in the first half of 2020 the respondents’ businesses and those of
their partners within the tourism field recorded a drop of 75-90%
*the respondents admitted that they had to take measures as
sending their employees in technical unemployment, shutting down
units and focus on online working.
*half of the respondents targeted investments into digitalisation
3. Appraise how the
consumption habits of
tourists coming to Romania
will change in terms of
tourism services following
the COVID-19 pandemic
*according to our respondents the COVID-19 pandemic will shift
tourists’ preferences towards secure or known destinations, most
likely local (within their respective country) destinations
*there is also anticipated an increase in the interest for individual or
even isolated tourism and also green, sustainable tourism
4. Describe the measures to
be taken by the government
and state institutions for the
field of tourism to help
your business and the
industry as a whole in
Romania
*according grants to tourism companies affected by COVID-19
*immediate payment of arrears to tourism companies
*partial or full support, by the state, of the additional costs
generated by the additional hygienic-sanitary measures that will
have to be taken by the accommodation, food, treatment, etc. units
*reducing bureaucracy by digitizing operational processes for
tourism stakeholders
Source: arrangement of data obtained within the discussions
with the interviewed specialists
Many respondents noted that red tape is a major issue for the tourism stakeholders therefore
reducing bureaucracy by digitizing hotel operational processes could be bring significant
results while adaptation of labour legislation for seasonal work, remote work and
simplification of labour force reporting are also important measures to be taken in order to
develop more efficient tourism activities.
Conclusions
Among the most affected sectors of the economy by the crisis caused by COVID-19 in
Romania are tourism, HoReCa, passenger transport and event organization. Many
Romanian businesses within these economic areas developed in the period of March-June
2020 serious cash flow problems, and the risk of layoffs, insolvencies and bankruptcies
increased significantly, creating waterfall effects that affect not only the businesses
themselves, but also connected industries and all tourists as well.
The main objectives of this study were to clarify the ongoing difficulties faced by the
tourism industry in terms of the effects of the COVID-19 pandemic and what measures
could be implemented in order to improve the trend within the industry. A qualitative study
has been developed including interviews (direct or mostly per email) based on a
questionnaire with sixteen representatives from eleven renowned hotels from four tourism
resorts within the Prahova County in Romania.
When addressing the first research question of the study, regarding the implications of the
COVID-19 crisis on the tourism sector for Romania, based on the interviews with the
sixteen representatives of the eleven hotels, we reckoned that the period of March-June
2020, the COVID-19 crisis has taken a financial toll on tourism businesses, entrepreneurs
and workforce in Romania. The pandemic is putting the Romanian tourism ecosystem
AE The Impact of COVID-19 on Romanian Tourism . An Explorative Case Study on Prahova County, Romania
204 Amfiteatru Economic
under unprecedented pressure and as result of travel and other restrictions, Romanian
tourism reached a gradual halt during the first two quarters of 2020. The opinions of our
respondents estimated the decline of their businesses to be from 70% to 90%, depending on
the length of the health crisis and on the pace of recovery. They also stated that revenue
losses for other stakeholders within their network is of around 80% for tour operators and
travel agencies, 70% for long-distance rail and around 95% for cruises and airlines. The
effects are most severe for small companies: lacking liquidity and facing uncertainty, they
struggle to stay afloat, access funding and maintain their employees and talent and therefore
they all need urgent action and emergency funding to bridge the period until tourism flows
will come back.
With regard to the second research question of the study which referred the
recommendations of the interviewed experts for the stakeholders of the tourism industry in
order to avert the worst effects and facilitate recovery, we can underline that the summer of
2020 and the following months will not be lost completely for the vibrant Romanian
tourism ecosystem, but they will all be dominated by a cautious attitude from all parts
involved. The tourism services’ providers themselves should focus more on family tourism,
individual tourism, health safety, non-congested areas, quality, sustainability and
ecotourism. However all the respondents agreed that the Romanian tourism system needs a
reset backed by the public institutions in the sector. In order to rearrange the entire field of
tourism, several generic lines of action are needed: increasing administrative efficiency,
promotion investment for tourism, labour and education, digitalization and innovation,
sustainability in tourism, but also actions specifically aimed at the main sectors of tourism:
accommodation structures, restaurants, recreation facilities, tour operators and agencies,
tourism guides, related services and tourist transport.
This made us conclude that tourism in Romania one major bet lost in 2020 and needs a
rapid and complex revival supported by not only tourism owners, operators and employees,
but also by tourists, suppliers, politicians and officials.
References
BBC, 2020. Business News. [online] BBC News. Available at: <https://www.bbc.com/
news/business> [Accessed 1 August 2020].
BBC, 2020. Coronavirus: Six months after pandemic declared, where are global hotspots?.
[online] BBC. Available at: <https://www.bbc.com/news/world-51235105> [Accessed
16 August 2020].
Biz, 2020. Cum a pus coronavirusul la pământ turismul și cum reacționează marii jucători
în domeniu. [online] Revista Biz. Available at: <https://www.revistabiz.ro/impactul-
covid-19-in-turism/> [Accessed 7 August 2020].
Bregnholm Ren, C., 2016. Qualitative research, tourism. In: Jafari, J. and Ziao, H., 2016.
Encyclopedia of Tourism. [online] Springer Publishing Company. Available at: <https://vbn.
aau.dk/en/publications/qualitative-research-tourism> [Accessed 14 August 2020].
Brouder, P., 2018. The end of tourism? A Gibson-Graham inspired reflection on the
tourism economy. Tourism Geographies, 20(5), pp. 916-918.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 205
Brouder, P., 2020. Reset redux: possible evolutionary pathways towards the transformation
of tourism in a COVID-19 world. Tourism Geographies, 22 (3), pp.484-490.
DeJonckheere, M. and Vaughn, L.M., 2019. Semistructured interviewing in primary care
research: a balance of relationship and rigour. Family Medicine and Community Health,
7(2), pp.1-8.
Glaser-Segura, D., Nistoreanu, P. and Dincă, V.M., 2018. Considerations on Becoming a World
Heritage Site ‒ A Quantitative Approach. Amfiteatru Economic, 20(47), pp. 202-216.
Gössling, S., Scott, D. and Hall, C.M., 2020. Pandemics, tourism and global change: a rapid
assessment of COVID-19. Journal of Sustainable Tourism, 29(1), pp. 1-20.
DOI: 10.1080/09669582.2020.1758708
Incoming Romania, 2020. Solutii de Organizare si Sustinere a Turismului Romanesc –
Alianta Pentru Turism. [online] Incoming Romania. Available at:
<https://incomingromania.org/industry/solutii-organizare-sustinere-turismului-
romanesc-alianta-pentru-turism/> [Accessed 9 August 2020].
Kliger, A.S. and Silberzweig, J., 2020. Mitigating Risk of COVID-19 in Dialysis Facilities.
American Society of Nephrology, 15(may), pp. 707-709.
Niewiadomski, P., 2020. COVID-19: from temporary de-globalisation to a re-discovery of
tourism?. Tourism Geographies, 22(3), pp. 651-656.
The New York Times, 2020. Coronavirus Travel Restrictions, Across the Globe. [online]
Available at: <https://www.nytimes.com/article/coronavirus-travel-restrictions.html>
[Accessed 17 August 2020].
UNCTAD, 2020. Covid-19 and tourism - assessing the economic consequences. [online]
UNCTAD. Available at: <https://unctad.org/en/PublicationsLibrary/ditcinf2020d3_
en.pdf> [Accessed 5 August 2020].
UNWTO, 2020a. Covid-19: Putting people first. [online] Available at:
<https://www.unwto.org/tourism-covid-19> [Accessed 2 August 2020].
UNWTO, 2020b. 100% of global destinations now have Covid-19 travel restrictions,
UNTWO reports. [online] Available at: <https://www.unwto.org/news/covid-19-travel-
restrictions> [Accessed 18 August 2020].
UNWTO, 2020c. International tourism continues to outpace the global economy. [online]
Available at: <https://www.e-unwto.org/doi/pdf/10.18111/9789284421152> [Accessed
27 July 2020].
Vo, N., Chovancová, M. and Tri, H. 2019. A major boost to the website performance of up-
scale hotels in Vietnam, Management & Marketing. Challenges for the Knowledge
Society, 14(1), pp.14-30.
WHO, 2020a. Coronavirus Disease (COVID-19) Dashboard. [online] Available at:
<https://covid19.who.int/> [Accessed 24 July 2020].
WHO, 2020b. Coronavirus disease 2019 (COVID-19) Situation Report – 72. [online]
Available at: <https://apps.who.int/iris/bitstream/handle/10665/331685/nCoVsitrep
01Apr2020-eng.pdf> [Accessed 22 September 2020].
AE On the Determinants of Fiscal Decentralization: Evidence From the EU
206 Amfiteatru Economic
ON THE DETERMINANTS OF FISCAL DECENTRALIZATION:
EVIDENCE FROM THE EU
Francisco J. Delgado
University of Oviedo and GEN, Spain
Please cite this article as:
Delgado, F.J., 2021. On the Determinants of Fiscal
Decentralization: Evidence From the EU. Amfiteatru
Economic, 23(56), pp. 206-220.
DOI: 10.24818/EA/2021/56/206
Article History
Received: 18 September 2020
Revised: 28 October 2020
Accepted: 27 November 2020
Abstract
We empirically analyze the determinants of the fiscal decentralization in the European
Union. Our approach consists on a quantile regression for the period 2005-2017. The
results show the differences in the impact of explanatory variables on the fiscal
decentralization by quantiles. Specifically, while GDP per capita or corruption are not
significant in a linear modelling, both are relevant in the quantile approach with different
effects along the distribution of fiscal decentralization. And other variables as population,
density or inequality do not have neither the same impact among quantiles, denoting the
limitations of using linear approaches for complex issues as fiscal decentralization.
Keywords: fiscal decentralization, federalism, European Union, quantile regression.
JEL Classification: H11, H77
Contact author: [email protected]
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 207
Introduction
The allocation of expenditure and revenue among the different levels of government,
central and subcentral -local and regional or state in some countries-, is a major issue in
Public Economics and, specifically, in Fiscal Federalism. The classical public functions of
efficiency -allocation-, redistribution -equity- and stabilization (Musgrave, 1959) are
permanently under the debate of the extent of fiscal decentralization, with large differences
across nations, arranged as federal or unitary countries. That extent of fiscal
decentralization depends on a wide array of factors, namely socioeconomic, institutional
and, obviously, historical, in each territory. It should be also noted that the interest of
focusing on fiscal decentralization processes is even greater in recent decades, with the
need of undertaking fiscal adjustments to control the public debt, especially after the Great
Recession, the integration processes as the eurozone, or the secession movements in
countries as United Kingdom or Spain, among others.
Beyond the link between decentralization and economic growth, a major issue in this
literature (Wasylenko, 1987; Martínez-Vázquez and McNab, 2003; Sato and Yamashige,
2005; Brueckner, 2006; Bodman, 2011; Chu and Yang, 2012; Baskaran and Feld, 2013;
Yang, 2019; or Canavire-Bacarreza et al., 2020), inequality (Sepulveda and Martínez-
Vázquez, 2011; Liu et al., 2017), welfare (Aslim and Neyapti, 2017), public sector
efficiency (Adam et al., 2014), public sector employment (Martínez-Vázquez and Yao,
2009), government size (Cassette and Paty, 2010), tax incentives (Li, 2015), corruption
(Arikan, 2004; Alfano et al., 2019) or even CO2 emissions (Cheng et al., 2020), the
determinants of fiscal decentralization have attracted limited attention in the literature*,
reviewed in the next section. Due to the potentially relevant impacts of fiscal
decentralization, the study of its determining factors deserves new empirical analysis with
different approaches.
This paper contributes to the fiscal decentralization literature with the first study with
quantile regression, to deal with potential nonlinearities and primarily to explore different
impact of determining factors along the distribution of fiscal decentralization. In addition,
we analyze the European case, not studied before with this aim†, especially interesting
because it is an integrated economic area, although Member States enjoy a large degree of
autonomy within their territories. Concretetly, we consider 26 European countries and for
the period 2005-2017.
The rest of the paper is organized as follows. Section 1 reviews the literature on
determinants of fiscal decentralization. Section 2 describes the empirical strategy, based on
quantile regression. Section 3 presents data, including a sigma convergence analysis of the
fiscal decentralization measures, and main results. Finally, we offer the main conclusions
and potential extensions of the work. The annex includes data for the initial and final years
of the period (2005 and 2017).
* See Martínez-Vázquez et al. (2017) for a recent survey on the effects of fiscal decentralization. † Jílek (2015) studied the tax decentralization in OECD-Europe countries and only for the local level
of government, for the period 1995-2013.
AE On the Determinants of Fiscal Decentralization: Evidence From the EU
208 Amfiteatru Economic
1. Theoretical background and literature review
The theoretical background is based on seminal works on Fiscal Federalism and the role of
levels of government, as Tiebout (1956)‡, Oates (1972, 2005) and Zodrow and
Mieszkowski (1986). Theoretically, subcentral levels of government should play a relevant
role in the allocative function, basically in terms of local public goods§ with the
Decentralization Theorem (Oates, 1972) as main reference, while their function in the
redistribution of income and stabilization of the economy would be less pronounced. In this
manner, the optimal degree of fiscal decentralization has attracted theoretical research since
those seminal works, as Janeba and Wilson (2011) or more recently Aslim and Neyapti
(2017) and Bellofatto and Besfamille (2018).
And specifically, some papers addressed theoretically the determinants of fiscal
decentralization, as Panizza (1999) or Arzaghi and Henderson (2005). Concretely, Panizza
(1999), in his theoretical model, concluded that the level of fiscal centralization was inversely
correlated with country size, income per capita, tastes differentiation and level of democracy.
Subsequently, Arzaghi and Henderson (2005) offered a theoretical model based on the
proposal by Panizza (1999) and also the literature on secession, where fiscal decentralization in
a country was higher with higher income, larger population, larger in spatial terms, greater
degree of local democratic culture and population concentrated in the hinterland.
As stated above, the determinants of fiscal decentralization have been analyzed in relatively
few empirical papers to date. Table no. 1 summarizes the papers on this topic. It should be
noted that the empirical evidence, short, is some contradictory about the significance of
some variables and their impact (positive or negative), reaching mixed results depending
the model, determining factors, countries and period considered. We will comment on these
results in Section 3 with the results of the study. Hence, our empirical evidence for the
European case will help to better understand the determining factors of the distribution of
expenditure and revenue among government levels.
Table no. 1: Literature review on determinants of fiscal decentralization
Paper Data Main results
Bahl and Nath (1986) 57 countries
1973
Positive relation with level of economic
development (GDP per capita and
urbanization) and population
Panizza (1999) 55 countries
1975, 1980, 1985
Negative effect of country size, income per
capita, ethnic fractionalization and level of
democracy
Cerniglia (2003) OECD countries Positive relation of area, population, degree
of urbanization and income per capita
‡ Brueckner (2004) carried out a numerical simulation to explore the fiscal decentralization in terms
of Tiebout versus tax competition, taking into account the contrary postulations of both approaches.
He concluded that under certain conditions, namely the curvature of the production function and the
dispersion of preferences are high, the decentralization is desirable. § See Besley and Coate (2003) for a political economy approach to the trade-off between centralized
and decentralized provision of local public goods.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 209
Paper Data Main results
Letelier (2005) 64 countries Negative impact of urbanization and
positive relationship with income per capita,
being stronger for high-income countries
Treisman (2006) 66 countries
1993-1995
Territorially larger, but not necessarily more
populous, countries were more fiscally
decentralized; in addition, the economic
development led to greater expenditure
decentralization and the federal states were
more decentralized
Bodman and Hodge
(2010)
53 countries
1981-1998
Positive relationship with income, but for
the middle‐ and lower‐ income nations,
higher income is found to be associated with
less decentralization
Lessmann
and Markwardt (2010)
64 countries
Decentralization counteracts corruption in
countries with high degrees of freedom of
the press, whereas countries without
effective monitoring suffer from
decentralization
Wu and Wang (2013) China
1995-2006
Negative effect of density, and non-
significant impact of GDP per capita and
openness
Letelier-Saavedra
and Saez-Lozano
(2015)
45 countries
1972-2008
Fiscal decentralization does not exhibit the
same pattern across specific government
functions, considering 6 functions
Canavire-Bacarreza
et al. (2016)
91 countries
1960-2007
Geographical fragmentation and area are
significantly and positively related to fiscal
decentralization
2. Empirical strategy
We employ a quantile regression approach to capture different patterns along the
distribution of fiscal decentralization. Contrary to linear regression, which summarizes the
average relationship between the regressors and the dependent variable, this semiparametric
approximation, proposed by Koenker and Basset (1978) and revised in Buchinsky (1998),
Koenker and Hallock (2001), Koenker (2017) and Waldmann (2018), minimizes the
deviations in absolute value with asymmetric weighting, instead of minimizing the squares
of the errors as in Ordinary Least Squares (OLS).
In this way, in the quantile regression approach, with the 0.05, 0.25, 0.50, 0.75 and 0.95
quantiles considered, the estimated marginal effects from the estimates of β would indicate
how the 5, 25, 50, 75 and 95 per cent conditional quantile would be affected at all x values.
In methodological terms, the quantile regression estimator can be more efficient than OLS
if errors deviate from normality and, in addition, the quantile estimators are less sensitive to
outliers. Besides, quantile regression provides a richer characterization of the data and is
invariant to monotonic transformations.
AE On the Determinants of Fiscal Decentralization: Evidence From the EU
210 Amfiteatru Economic
Regarding the fiscal decentralization measures**, we consider both the expenditure and
revenue sides (FD Expenditure and FD Revenue), taking the percentage of non-central
levels of government expenditure or revenue over the total government as indicators, the
most employed measures in the literature of fiscal decentralization. The data have been
obtained from Eurostat††.
As explanatory variables, and following the literature on this topic, we consider the level of
development through the GDP per capita, population, density, inequality, corruption and a
dummy variable for federal versus unitary countries. The model also includes year
dummies.
GDP per capita: the literature does not predict a clear relationship with fiscal
decentralization, although mostly positive: while Bahl and Nath (1986), Letelier (2005),
Treisman (2006), Martinez-Vazquez and Timofeev (2009) and Bodman and Hodge (2010)
found a positive relation between economic development and fiscal decentralization, Oates
(1972) and Panizza (1999) concluded a negative relation, while Wu and Wang (2013) did
not reach a significant impact. Source: Eurostat.
Population: although there are reasons for both a positive as a negative effect, we
expect a positive relationship, according to Litvack and Oates (1971); as population grows,
the rising costs of congestion at the local level of government will tend to increase the non-
central government’s expenditures relative to the central government’s ones. Data are in
millions of inhabitants. Source: Eurostat.
Density: again, we find arguments in favor of both positive as negative impact on
fiscal decentralization. Our expectation, following Letelier (2005), is a negative
relationship. On one hand, a lower density will lower public spending, reducing the
government’s marginal benefit of centralization; on the other hand, as the median voter’s
marginal utility is decreasing in government expenditures, a lower government budget
involves a higher marginal utility of public goods against private consumption. And the
assumption is a more significant negative effect of more centralization on the median
voter’s demand for spending, raising the marginal cost of centralization. Source: Eurostat.
Inequality‡‡: we include the Gini coefficient, reflecting the income inequality, but
not in wealth. The expected sign is negative, in line with Sepulveda and Martínez-Vázquez
(2011), who distinguished direct and indirect effects of fiscal decentralization on income
inequality, derived from changes in the implementation of public policies or in the behavior
of relevant economic agents, and those observed after the decentralization process has
interacted with the socioeconomic framework, respectively. They found empirically that
** For a discussion about fiscal decentralization measures, see Stegarescu (2005), Martínez-Vázquez
and Timofeev (2009), Dziobek et al. (2011) or more recently Martínez-Vázquez et al. (2017). †† Government revenue, expenditure and main aggregates (gov_10a_main), available at Eurostat (2020). ‡‡ In a related interesting study, Sacchi and Salotti (2014) investigated the effects of fiscal
decentralization on income inequality for 23 OECD countries in 1971-2000, concluding that a
higher degree of tax decentralization is associated with higher household income inequality within a
country. And Kyriacou et al. (2017), for a sample of 23 OECD countries for the period 1984-2005,
found that fiscal decentralization, together with measures to improve the quality of government,
was an effective strategy for reducing regional inequalities. But they stated that these results have
been obtained for rich and advanced economies, and it would be useful to analyze the case of less
developed countries.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 211
this impact is negative if the general government represents a significant share of the
economy, above 20 per cent GDP. Source: Eurostat.
Corruption: we consider the Corruption Perceptions Index by Transparency
International, an indicator ranging from 0 (highly corrupt) to 100 (very clean). We would
expect a negative relationship with fiscal decentralization, according to the hypothesis of
Arikan (2004) and Lessmann and Markwardt (2010). Arikan (2004) developed a theoretical
model where, as the number of competing jurisdictions rises, the level of corrupt earnings,
or tax revenue appropriated by bureaucrats, falls; and found some evidence of this negative
relationship in a cross-section data set of 40 countries. In Lessmann and Markwardt (2010),
they analyzed the impact of fiscal decentralization on corruption taking into account the
degree of freedom of the press, and they did not identify a robust impact of decentralization
on corruption, but negative in countries with an eff ective monitoring through a free and
independent press.
Federal: this dummy variable assigns value 1 to the three federal countries, Belgium,
Austria, Germany, and also to Spain, with three levels of government -central, regional and
local- despite it is not formally or politically a federal state, and one of the most
decentralized country in the world, even including a fourth level of government represented
by the provinces, with limited competences. Of course, the expected sign is positive
3. Data and results
We study the European Union, concretely 26 countries due to unavailability of data for
Croatia and Romania, for the period 2005-2017. The main statistics are reported in Table
no. 2. We observe a great gap in fiscal decentralization, both in expenditure as in revenue,
ranging from near zero in Malta to around 72 per cent in Germany. Other few decentralized
countries are Ireland and United Kingdom, with indicators under 10 per cent, and among
the highly decentralized nations we must also mention France, Finland and Spain, with
measures above 50 per cent§§. It should be noted that these three last countries are not
federal in political terms, but their degrees of fiscal decentralization are larger than federal
countries as Belgium or Austria.
Table no. 2: Summary statistics
Variable Mean Std. Dev. Min Max
FD Expenditure 32.58 15.81 0.04 71.84
FD Revenue 34.52 16.10 0.22 71.94
GDPpc 25,961 15,841 4,190 84,420
Population 18.41 23.49 0.40 82.52
Density 177.10 250.20 15.49 1,457.00
Inequality 29.72 3.87 22.70 40.20
Corruption 65.36 16.41 33.00 96.00
Federal (dummy) 0.15 0.36 0.00 1.00
Source: own elaboration from Eurostat and Transparency International
§§ In a recent study, Blanco et al. (2020) analyze the convergence of fiscal decentralization in the
European Union. In addition, Finzgar and Oplotnik (2013) revise the fiscal decentralization systems
in the EU.
AE On the Determinants of Fiscal Decentralization: Evidence From the EU
212 Amfiteatru Economic
With the aim at deepening the evolution of fiscal decentralization measures in the period,
we carry out a sigma convergence analysis. This approach consists on the measurement of
the dispersion of the variable through the coefficient of variation, evidencing a sigma
convergence process if that dispersion diminishes over time, and sigma divergence if the
dispersion increases. We plot this sigma convergence measure along with the average fiscal
decentralization to better understand the evolution in our sample (Figure no. 1).
With regard to expenditure, we can observe a growing trend in the (unweighted) average
after 2010 and, especially, after 2013. And, however, in the dispersion, we differentiate two
main stages: a general growing trend until 2013, followed by a decline to end the period
above the initial level of dispersion. Hence, the expenditure does not exhibit a sigma
convergence process in the EU.
In relation to the revenue, the path of the average is, to some extent, contrary to the case of
expenditure: a decreasing trend since 2009, broken in 2015, ending the period practically at the
same level of 2005. Meanwhile, the behaviour of the dispersion is similar to the observed for
the expenditure, and we can not conclude a sigma convergence neither in the revenue.
1.a. Expenditure
1.b. Revenue
Figure no. 1: Sigma convergence of fiscal decentralization
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 213
The main results are summarized in Table no. 3, concretely 3.a for expenditure and 3.b for
revenue, being the results virtually identical in both cases. Hence, we will comment only on
the expenditure case. The linear results, reported only for benchmarking, show non
significant results for GDP per capita nor corruption, significant and positive for population
and federal, and significant and negative for density and inequality. It should be noted that
most of the signs are the expected ones. But the quantile results allows a richer analysis of
the impacts along the distribution of fiscal decentralization.
Regarding GDP per capita, our results show a significant and negative effect for all
quantiles except 0.25, and it is stronger in the extremes of the distribution, 0.05 and 0.95,
namely the lowest and highest decentralized countries. This negative impact is in line with
Panizza (1999), as stated in the review reported in previous section.
For the population, we conclude a significant and positive effect on all quantiles, result in
line with Bahl and Nath (1986). In this case, the impact is weaker in the extremes, 0.05 and
0.95, that is, for lowest or highest-decentralized countries.
In the case of density, the effect is negative in all quantiles, and this impact is stronger as
we move foreward the distribution of fiscal decentralization.
With regard to inequality, as expected, we find a negative relation with fiscal
decentralization, but it is bigger in the first quantiles and finally non significant at 0.95.
Hence, the negative impact of inequality is larger for low-decentralized countries, and this
effect dissapears for high-decentralized cases.
With respect to corruption, the results reveal that the relationship is significant and negative
at the first quantiles, low-decentralized countries, turning into significant and positive in the
last ones, 0.75 and 0.95, namely high-decentralized cases.
Finally, the federal dummy presents a significant and positive coefficient in all quantiles, as
expected.
The data and results achieved in this paper bring to light a huge variety of degrees of fiscal
decentralization across European countries and their determining factors, with impacts
varying along the distribution of fiscal decentralization, namely, the effects are not uniform
for low, medium or high-decentralized countries. It should be also noted that, among the
most decentralized European countries, we find federal and unitary cases, denoting also the
relevance of historical issues to explain fiscal decentralization in Europe. We must also take
into account that the integration*** and harmonization processes in the European Union do
not include rules or recommendations about the fiscal decentralization within the Member
States. In another perspective, it should be remarked that the proper European Union, as a
whole, faces the centralization versus descentralization issue, in terms of fiscal policy,
fiscal discipline and structural reforms (Wyplosz, 2015).
*** In a related topic, Ermini and Santolini (2014) studied the effects of globalization on fiscal
decentralization for OECD countries, concluding a positive impact of the overall index of
globalization, concretely the KOF Globalisation Index, on both tax revenue and expenditure
decentralization.
AE On the Determinants of Fiscal Decentralization: Evidence From the EU
214 Amfiteatru Economic
Table no. 3: Results from the quantile regression
3.a. Expenditure Quantiles
Linear 0.05 0.25 0.50 0.75 0.95
GDPpc -0.00008 -0.00036*** -0.00009 -0.00008*** -0.00021*** -0.00029***
(0.00006) (0.00004) (0.00006) (0.00003) (0.00003) (0.00004)
Population 0.28755*** 0.13272*** 0.38226*** 0.33659*** 0.32694*** 0.28469***
(0.02742) (0.01763) (0.02712) (0.01507) (0.01500) (0.02173)
Density −0.02405*** -0.01747*** -0.01739*** -0.02370*** -0.02437*** -0.02512***
(0.00242) (0.00155) (0.00239) (0.00133) (0.00132) (0.00192)
Inequality −1.18137*** -1.33762*** -0.91074*** -1.03553*** -0.55402*** 0.02353
(0.17731) (0.11398) (0.17532) (0.09744) (0.09700) (0.14048)
Corruption −0.01038 -0.27420*** -0.15590*** 0.13449*** 0.37579*** 0.41299***
(0.05695) (0.03661) (0.05631) (0.03129) (0.03116) (0.04512)
Federal 14.7996*** 23.3013*** 14.8889*** 13.6451*** 9.30624*** 9.40833***
(1.78230) (1.14573) (1.76237) (0.97944) (0.97511) (1.41207)
3.b. Revenue Quantiles
Linear 0.05 0.25 0.50 0.75 0.95
GDPpc -0.00004 -0.00023*** 0.00010 -0.00001 -0.00015*** -0.00026***
(0.00001) (0.00001) (0.00006) (0.00004) (0.00002) (0.00002)
Population 0.29508*** 0.11778*** 0.35496*** 0.37330*** 0.31689*** 0.30110***
(0.02824) (0.00733) (0.03043) (0.01791) (0.01092) (0.00907)
Density −0.02597*** -0.01831*** -0.02071*** -0.02507*** -0.02806*** -0.02991***
(0.00249) (0.00064) (0.00269) (0.00158) (0.00096) (0.00080)
Inequality −1.24355*** -1.06082*** -1.20627*** -0.92315*** -0.73916*** 0.03652
(0.18260) (0.04744) (0.19674) (0.11581) (0.07062) (0.05864)
Corruption −0.08135 -0.35670*** -0.20084*** 0.05146 0.28492*** 0.40426***
(0.05865) (0.01524) (0.06319) (0.03720) (0.02268) (0.01883)
Federal 14.16920*** 21.7734*** 15.6258*** 11.0865*** 9.38635*** 7.88633***
(1.83550) (0.47690) (1.97762) (1.16414) (0.70994) (0.58948)
Notes: ***, **, * denotes statistical significance at the 1 per cent, 5 per cent and 10 per cent
levels, respectively. Standard errors in parentheses.
Conclusions
Fiscal decentralization, in the expenditure and revenue sides, have attracted researchers in
the last decades to investigate issues as the relationship with economic growth, inequality,
public efficiency or government size, but few papers have addressed the determinants of
fiscal decentralization to date.
In this paper, we study the determining factors of fiscal decentralization in the European
Union for the period 2005-2017, considering 26 Member States due to data unavailability
for Croatia and Romania. Specifically, and in order to deal with potential nonlinearities and
to enrich the analysis along the distribution of fiscal decentralization, we employ a quantile
regression approach.
The results show a positive relationship between fiscal decentralization and population. In
addition, the impact is negative for GDP per capita, density -with larger negative impact in
high-decentralized countries relative to low-decentralized nations- and inequality -although
the impact is non-significant in high-decentralization countries. Finally, the evidence for
corruption is mixed, being negative for the low-decentralized countries and positive in
high-decentralized cases. Hence, we observe how the effects vary at different parts of the
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 215
fiscal decentralization distribution. It should be also noted that other factors, mainly
historical issues, help to explain the variety in the degree of fiscal decentralization in the
European Union Member States, taking into account that the integration and harmonization
processes do not contain rules or recommendations about fiscal decentralization, prevailing
the autonomy of the countries in this matter.
Regarding possible extensions of our study, we could extend the analysis for alternative
measures of fiscal decentralization, beyond the two most commonly used in the literature
followed in this work, as the Regional Authority Index -although this is available only for
some years-, or with some disaggregation in those measures. In addition, it should be
interesting to explore the effects of Great Recession on the degree of fiscal decentralization.
And finally, to investigate the merging of this literature of fiscal decentralization
determinants with club convergence analysis to explain the composition of clusters.
Acknowledgements
I gratefully acknowledge the funding from the Spanish Ministry of Economy, Industry and
Competitiveness, project CSO2017-85024-C2-2-P.
References
Adam, A., Delis, M.D. and Kammas, P., 2014. Fiscal decentralization and public sector
efficiency: evidence from OECD countries. Economics of Governance, 15(1), pp. 17-
49.
Alfano, M.R., Baraldi, A.L. and Cantabene, C., 2019. The effect of fiscal decentralization
on corruption: a non‐ linear hypothesis. German Economic Review, 20(1), pp. 105-128.
Arikan, G.G., 2004. Fiscal decentralization: a remedy for corruption? International Tax and
Public Finance, 11(2), pp. 175-195.
Arzaghi, M. and Henderson, J.V., 2005. Why countries are fiscally decentralizing. Journal
of Public Economics, 89(7), pp. 1157-1189.
Aslim, E.G. and Neyapti, B., 2017. Optimal fiscal decentralization: redistribution and
welfare implications. Economic Modelling, 61, pp. 224-234.
Bahl, R.W. and Nath, S., 1986. Public expenditure decentralization in developing countries.
Environment and Planning C: Government and Policy, 4(4), pp. 405-418.
Baskaran, T. and Feld, L.P., 2013. Fiscal decentralization and economic growth in OECD
countries: is there a relationship?. Public Finance Review, 41(4), pp. 421-445.
Bellofatto, A.A. and Besfamille, M., 2018. Regional state capacity and the optimal degree
of fiscal decentralization. Journal of Public Economics, 159, pp. 225-243.
Besley, T. and Coate, S., 2003. Centralized versus decentralized provision of local public
goods: a political economy approach. Journal of Public Economics, 87(12), pp. 2611-2637.
Blanco, F.A., Delgado, F.J. and Presno, M.J., 2020. Fiscal decentralization policies in the
EU: a comparative analysis through a club convergence analysis. Journal of
Comparative Policy Analysis: Research and Practice, 22(3), pp. 226-249.
Bodman, P., 2011. Fiscal decentralization and economic growth in the OECD. Applied
Economics, 43(23), pp. 3021-3035.
AE On the Determinants of Fiscal Decentralization: Evidence From the EU
216 Amfiteatru Economic
Bodman, P. and Hodge, A., 2010. What drives fiscal decentralisation? Further assessing the
role of income. Fiscal Studies, 31(3), pp. 373-404.
Brueckner, J.K., 2004. Fiscal decentralization with distortionary taxation: Tiebout vs. tax
competition. International Tax and Public Finance, 11(2), pp. 133-153.
Brueckner, J.K., 2006. Fiscal federalism and economic growth. Journal of Public
Economics, 90, pp. 2107-2120.
Buchinsky, M., 1998. Recent advances in quantile regression models: a practical guideline
for empirical research. Journal of Human Resources, 33(1), pp. 88-126.
Canavire-Bacarreza, G., Martinez-Vazquez, J. and Yedgenov, B., 2016. Reexamining the
determinants of fiscal decentralization: what is the role of geography? Journal of
Economic Geography, 17(6), pp. 1209-1249.
Canavire-Bacarreza, G., Martinez-Vazquez, J. and Yedgenov, B., 2020. Identifying and
disentangling the impact of fiscal decentralization on economic growth. World
Development, [e-journal] 127. https://doi.org/10.1016/j.worlddev.2019.104742
Cassette, A. and Paty, S., 2010. Fiscal decentralization and the size of government: a
European country empirical analysis. Public Choice, 143, pp. 173-189.
Cerniglia, F., 2003. Decentralization in the public sector: quantitative aspects in federal and
unitary countries. Journal of Policy Modeling, 25(8), pp. 749-776.
Cheng, S., Fan, W., Chen, J., Meng, F., Liu, G., Song, M. and Yang, Z., 2020. The impact
of fiscal decentralization on CO2 emissions in China. Energy, [e-journal] 192. DOI:
10.1016/j.energy.2019.116685.
Chu, A.C. and Yang, C.C., 2012. Fiscal centralization versus decentralization: Growth and
welfare effects of spillovers, Leviathan taxation, and capital mobility. Journal of Urban
Economics, 71(2), pp. 177-188.
Dziobek, C., Gutierrez, C. and Kufa, P., 2011. Measuring fiscal decentralization –
Exploring the IMF’s databases. IMF Working Paper 11/126.
Ermini, B. and Santolini, R., 2014. Does globalization matter on fiscal decentralization?
New evidence from the OECD. Global Economic Review, 43(2), pp. 153-183.
Eurostat, 2020. Government revenue, expenditure and main aggregates. [online] Available
at: < https://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=gov_10a_main&lang=
en> [Accessed 25 November 2020].
Finzgar, M. and Oplotnik, Z.J., 2013. Comparison of fiscal decentralization systems in EU-
27 according to selected criteria. Lex Localis, 11(3), pp. 651-672.
Janeba, E. and Wilson, J.D., 2011. Optimal fiscal federalism in the presence of tax
competition. Journal of Public Economics, 95(11-12), pp. 1302-1311.
Jílek, M., 2015. Factors of tax decentralization in OECD-Europe countries. European
Financial and Accounting Journal, 10(2), pp. 33-49.
Koenker, R., 2017. Quantile regression: 40 years on. Annual Review of Economics, 9,
pp. 155-176.
Koenker, R. and Basset, G., 1978. Regression quantiles. Econometrica, 46(1), pp. 33-50.
Koenker, R. and Hallock, K.F., 2001. Quantile regression. An introduction. Journal of
Economic Perspectives, 15(4), pp. 143-156.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 217
Kyriacou, A.P., Muinelo-Gallo, L. and Roca-Sagalés, O., 2017. Regional inequalities, fiscal
decentralization and government quality. Regional Studies, 51(6), pp. 945-957.
Lessmann, C. and Markwardt, G., 2010. One size fits all? Decentralization, corruption, and
the monitoring of bureaucrats. World Development, 38(4), pp. 631-646.
Letelier, L., 2005. Explaining fiscal decentralization. Public Finance Review, 33(2),
pp. 155-183.
Letelier-Saavedra, L. and Saez-Lozano, J.L., 2015. Fiscal decentralization in specific areas
of government: an empirical evaluation using country panel data. Environment and
Planning C: Politics and Space, 33(6), pp. 1344-1360.
Li, Q., 2015. Fiscal decentralization and tax incentives in the developing world. Review of
International Political Economy, 23(2), pp. 232-260.
Litvack, J. and Oates, W., 1971. Group size and the output of public goods: theory and
application to state-local finance in the United States. Public Finance, 25(2), pp. 42-58.
Liu, Y., Martinez-Vazquez, J. and Wu, A.M., 2017. Fiscal decentralization, equalization,
and intra-provincial inequality in China. International Tax and Public Finance, 24,
pp. 248-281.
Martínez-Vázquez, J., Lago-Peñas, S. and Sacchi, A., 2017. The impact of fiscal
decentralization: a survey. Journal of Economic Surveys, 31(4), pp. 1095-1129.
Martínez-Vázquez, J. and McNab, R.M., 2003. Fiscal decentralization and economic
growth. World Development, 31(9), pp. 1597-1616.
Martínez-Vázquez, J. and Timofeev, A., 2009. Decentralization measures revisited.
Working Paper 09-13. Georgia State University.
Martínez-Vázquez, J. and Yao, M., 2009. Fiscal decentralization and public sector
employment. Public Finance Review, 37(5), pp. 539-571.
Mullen, J.K., 1980. The role of income in explaining state-local fiscal decentralization.
Public Finance, 35(2), pp. 300-308.
Musgrave, R.A., 1959. The Theory of Public Finance. New York: McGraw-Hill.
Oates, W.E., 1972. Fiscal Federalism. New York: Harcourt Brace Jovanovich.
Oates, W.E., 2005. Toward a second-generation theory of fiscal federalism. International
Tax and Public Finance, 12, pp. 349-373.
Panizza, U., 1999. On the determinants of fiscal centralization: Theory and evidence.
Journal of Public Economics, 74(1), pp. 97-139.
Sacchi, A. and Salotti, S., 2014. The effects of fiscal decentralization on household income
inequality: some empirical evidence. Spatial Economic Analysis, 9(2), pp. 202-222.
Sato, M. and Yamashige, S., 2005. Decentralization and economic development: an
evolutionary approach. Journal of Public Economic Theory, 7(3), pp. 497-520.
Sepulveda, C.F. and Martínez-Vázquez, J., 2011. The consequences of fiscal
decentralization on poverty and income equality. Environment and Planning C: Politics
and Space, 29(2), pp. 321-343.
Stegarescu, D., 2005. Public sector decentralisation: measurement concepts and recent
international trends. Fiscal Studies, 26(3), pp. 301-333.
Tiebout, C.M., 1956. A pure theory of local expenditures. Journal of Political Economy,
64(5), pp. 416-424.
AE On the Determinants of Fiscal Decentralization: Evidence From the EU
218 Amfiteatru Economic
Treisman, D., 2006. Explaining fiscal decentralisation: geography, colonial history,
economic development, and political institutions. Commonwealth & Comparative
Politics, 44(3), pp. 289-325.
Waldmann, E., 2018. Quantile regression: a short story on how and why. Statistical
Modelling, 18(3-4), pp. 203-218.
Wasylenko, M., 1987. Fiscal decentralization and economic development. Public
Budgeting & Finance, 7(4), pp. 57-71.
Wu, A.M. and Wang, W., 2013. Determinants of expenditure decentralization: evidence
from China. World Development, 46, pp. 176-184.
Wyplosz, C., 2015. The centralization-decentralization issue. Discussion Paper 014.
European Commission.
Yang, S., 2019. Fiscal decentralization or centralization: diverging paths of Chinese cities.
China & World Economy, 27(3), pp. 102-125.
Zodrow, G. and Mieszkowski, P., 1986. Pigou, Tiebout, property taxation, and the
underprovision of local public goods. Journal of Urban Economics, 19(3), pp. 356-370.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 219
Annex
Table A.1: Data - 2005
Country FD Exp FD Rev GDPpc Popul Density Ineq. Corrup.
Belgium 48.68 50.42 35,250 11.35 371.85 26.0 75
Bulgaria 30.94 30.48 6,310 7.10 64.03 40.2 43
Czechia 27.15 28.56 17,200 10.58 134.14 24.5 57
Denmark 26.22 25.83 47,360 5.75 133.40 27.6 88
Germany 71.36 71.59 35,420 82.52 231.13 29.1 81
Estonia 14.53 14.79 14,440 1.32 29.09 31.6 71
Ireland 4.69 5.00 54,240 4.78 68.08 30.6 74
Greece 24.09 27.68 17,410 10.77 81.61 33.4 48
Spain 54.41 55.54 24,410 46.53 92.17 34.1 57
France 58.75 62.36 32,370 66.81 121.65 28.8 70
Italy 37.62 40.30 26,490 60.59 201.07 32.7 50
Cyprus 23.00 24.63 23,120 0.85 149.02 30.8 57
Latvia 40.41 41.54 11,560 1.95 30.19 34.5 58
Lithuania 34.30 35.82 12,720 2.85 43.61 37.6 59
Luxembourg 27.71 31.19 82,550 0.59 227.44 30.9 82
Hungary 27.41 28.79 11,930 9.80 105.32 28.1 45
Malta 0.04 0.22 20,910 0.46 1456.64 28.2 56
Netherlands 40.28 39.76 40,730 17.08 411.34 27.1 82
Austria 34.53 35.28 37,090 8.77 104.61 27.9 75
Poland 39.29 46.17 11,820 37.97 121.44 29.2 60
Portugal 24.67 29.48 17,650 10.31 112.07 33.5 63
Slovenia 40.02 40.39 19,430 2.07 101.90 23.7 61
Slovakia 38.31 40.04 14,970 5.44 111.28 23.2 50
Finland 50.96 53.70 36,310 5.50 16.27 25.3 85
Sweden 40.33 38.81 43,350 10.00 24.09 28.0 84
United Kingdom 8.56 8.07 32,460 65.84 268.95 33.1 82
AE On the Determinants of Fiscal Decentralization: Evidence From the EU
220 Amfiteatru Economic
Table A.2: Data - 2017
Country FD Exp FD Rev GDPpc Popul Density Ineq. Corrup.
Belgium 42.40 44.83 32,200 10.45 342.17 28.0 74
Bulgaria 33.62 31.73 4,190 7.69 69.32 31.2 40
Czechia 26.36 28.14 13,570 10.20 129.32 26.0 43
Denmark 32.60 29.18 44,400 5.41 125.58 23.9 95
Germany 69.59 71.94 29,730 82.50 231.07 26.1 82
Estonia 14.43 14.27 11,110 1.36 30.05 34.1 64
Ireland 7.60 7.82 39,470 4.11 58.50 31.9 74
Greece 23.54 29.93 20,910 10.97 83.14 33.2 43
Spain 54.06 54.23 23,420 43.30 85.77 32.2 70
France 55.51 57.99 30,320 62.77 114.30 27.7 75
Italy 41.55 44.43 28,090 57.87 192.06 32.7 50
Cyprus 14.64 22.52 23,050 0.73 127.80 28.7 57
Latvia 39.15 42.76 8,170 2.25 34.83 36.2 42
Lithuania 26.72 28.54 7,950 3.36 51.38 36.3 48
Luxembourg 28.20 30.77 76,460 0.46 177.60 26.5 85
Hungary 34.60 39.32 9,910 10.10 108.54 27.6 50
Malta 0.27 0.35 14,790 0.40 1274.27 27.0 66
Netherlands 40.30 39.64 36,570 16.31 392.66 26.9 86
Austria 31.95 33.39 33,710 8.20 97.80 26.3 87
Poland 42.31 46.73 7,510 38.17 122.08 35.6 34
Portugal 24.30 27.87 16,600 10.49 114.08 38.1 65
Slovenia 35.53 38.12 16,570 2.00 98.53 23.8 61
Slovakia 35.20 37.24 9,960 5.37 109.99 26.2 43
Finland 47.87 50.46 34,250 5.24 15.49 26.0 96
Sweden 39.27 40.43 37,770 9.01 21.72 23.4 92
United Kingdom 8.43 8.05 30,160 60.18 245.82 34.6 86
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 221
FOOD QUALITY COMPETITION AMONG COMPANIES
AND GOVERNMENT FOOD SAFETY SUPERVISION
UNDER ASYMMETRIC PRODUCT SUBSTITUTION
Ningzhou Shen1*, Yinghua Song2, Dan Liu3 and Dalia Streimikiene4 1)2)3) China Research Center for Emergency Management, Wuhan University
of Technology and School of Safety Science and Emergency Management, Wuhan University of Technology, Wuhan, China
4) Lithuanian Institute of Agrarian Economics, Vilnius, Lithuania
Please cite this article as:
Shen, N., Song, Y., Liu, D. and Streimikiene, D., 2021.
Food Quality Competition Among Companies and
Government Food Safety Supervision Under Asymmetric
Product Substitution. Amfiteatru Economic, 23(56),
pp. 221-239.
DOI: 10.24818/EA/2021/56/221
Article History
Received: 11 September
2020
Revised: 1 November 2020
Accepted: 25 November
2020
Abstract
The frequent exposures of food safety events in recent years have aroused extensive social
concerns. Food quality and safety are hot topics in the food engineering field. In a market
with mutual competitions, the products of different enterprises are substitutive, and
enterprises have to achieve reasonable yield and to ensure product quality to maximize their
profits. However, the limited production resources of enterprises affect their strategies in
yield and quality. To explore enterprises’ decisions in yield and food quality under no
resource constraint as well as with yield and quality under resource constraints, a Cournot
model with differential product substitution was constructed by using game theory, and the
effects of government monitoring on food quality decision were further investigated.
Results show that enterprises with strong product substitutability increase the yield and
quality of their products. They can gain higher profits under no resource constraint and
under resource constraint, but their profits are lower under quality constraint compared to
the profits of enterprises with low product substitutability. When the equilibrium quality of
enterprises is lower than the lowest quality standard requirements of the government, a high
penalty ( β ) or a relatively low government supervision difficulty ( c ) urges enterprises to
improve product quality. Under this circumstance, two enterprises produce according to the
lowest quality standards. On the contrary, enterprises take certain risks to produce low-
quality products to increase profits. The conclusions provide a theoretical basis to formulate
food safety regulation policies from the perspective of product substitutability.
Keywords: food safety; Cournot competition; asymmetric product substitution; game theory
JEL Classification: I20, I23, O15, P46
* Corresponding author, Ningzhou Shen – e-mail: [email protected]
AE Food Quality Competition Among Companies and Government Food Safety Supervision Under Asymmetric Product Substitution
222 Amfiteatru Economic
Introduction
The research field of food engineering is very extensive, involving microorganism,
chemistry, technological process, public health, and risk assessment among other areas
(Rodriguez-Parada, et al., 2018; Neumann-Langdon et al., 2019; Tomerlin et al., 2019;
Aray et al., 2020; Datta et al., 2020). Food safety problems have attracted the extensive
attention of the whole society. In particular, food quality and safety has become a very
important research direction in the food engineering field (Petrovic et al., 2018; Pineda-
Escobar, 2019; Luo et al., 2020). Food safety scandals have been reported continuously in
various media in recent years (Yan, 2012; Liu, Pieniak and Verbeke, 2013; Liu and Ma,
2016). For example, Burger King, the international fast food giant, was exposed for using
expired ingredients in 2020. In 2008, Sanlu was exposed for adding melamine to its milk
powder (Pei et al., 2011). In 2005, KFC, another international fast food giant, was exposed
for the illegal use of food additives. Exposures of a series of food safety problems have not
only greatly damaged consumers’ trust in the food industry but also caused a crisis of
public confidence over food safety. Therefore, food safety and quality in food engineering
has become a problem that governments and consumers are concerned about more and
more (Wang et al., 2008; Liu, 2010; Peng et al., 2015).
Factors that influence food safety mainly come from two aspects. On the one hand,
consumers are concerned about price and food quality during food purchase, which forces
competing enterprises in the market to gain advantages through quality control (Yang and
Nie, 2016; Han and Li, 2017). On the other hand, governments supervise the product
quality of enterprises in consideration of public safety in order to guarantee that the food
quality meets specific standard (Song et al., 2018). However, previous studies demonstrate
that the same type of products from different brands are substitutive to some extent
(Besanko et al., 2005), and enterprises consider the substitutability of their competitors’
products when making decisions regarding yield and quality of products (Shaffer and
Zettelmeyer, 2004; Zheng et al., 2020). Enterprises in the market influence consumers’
cognition of and preferences for products through advertisement and other means, resulting
in different brand influences and substitutability of enterprise products (Shaffer and
Zettelmeyer, 2004). As a result, the differences in product substitutability can directly
influence enterprises’ decision regarding yield and quality. However, existing studies
neglect the differences among enterprises in product substitutability. The effects of
differences in product substitutability among enterprises on their decision regarding product
quality under double oligopolistic competition were analyzed on the basis of the Cournot
competition model. Moreover, enterprise decision in production resources when there are
yield and quality constraints was discussed. Accordingly, the effects of government
supervision on the decision of enterprises were further analyzed.
The remainder of this study is organized as follows. Section 1 provides the literature
review. Section 2 introduces the research methodology. Cournot competition models under
no resource constraint and under yield and quality resource constraints were constructed,
respectively. On this basis, the government–enterprise game model under government
supervision was constructed. Section 3 introduces the analysis of results. The enterprise
balances among yield, quality, and profits under three situations were analyzed. The yield–
quality balance of enterprises under government supervision and equilibrium of
government supervision input was also investigated. Section 4 provides the discussions.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 223
Management and policy enlightenments were disclosed according to the result analysis.
The last section draws the main conclusions of this study.
1. Literature review
For food safety and quality in food engineering, foods should meet both yield and quality
requirements to ensure consumers gain enough standard-quality foods (Pinstrup-Andersen,
2009; Yang and Nie, 2016). Many scholars are concerned about food safety problems in
food engineering and have studied such concern from different perspectives. Pinstrup-
Andersen (2009) discussed the definition and measurement standards of food safety. Tirado
et al. (2010) reviewed the potential influences of climatic changes on food contamination
and food safety as well as analyzed food safety and relevant strategies in different stages of
the food supply chain. Liu and Ma (2016) constructed a hierarchical analysis model of
cities in China to investigate food safety scandals, media exposure, and public safety
concerns as well as disclose the influences of food safety scandals and media exposure on
food safety risks.
The above studies mainly discussed food safety from the perspectives of the natural
environment, government, and consumers but neglected the role of enterprises in food
safety. Hence, some scholars focus on the food quality management of enterprises in the
food supply chain (Rong et al., 2011; Chen et al., 2014; Migliore et al., 2015). Van Der
Vorst et al. (2009) studied the increasing demands of consumers in the food supply chain
for food quality and sustainability and proposed a new method to integrate logistics,
sustainability, and food quality analysis. Wang et al. (2015) constructed the game models of
enterprises under three cooperation situations to analyze the quality and safety inputs of
enterprises and their effects on enterprise profits and food safety. Meanwhile, an
evolutionary game model was built to investigate factors that influence the cooperation
strategies of enterprises through theoretical and simulation analyses. Rouvière (2016)
discussed the relationship between enterprise scale and preventive measures of enterprise
food safety and found that small companies often made more efforts than do large
companies. Parker et al. (2016) studied the food safety problems at farms and found that
good agricultural practice knowledge has no significant differences among farms with
different scales. Planters in small-scale production believe that food safety standards cannot
adapt to local agricultural conditions and their agricultural scales. Chen et al. (2017) studied
the impacts of enterprise social responsibility on food safety, and It is generally believed
that positive price control and quality control lower the benefits of monopolists, consumer
surplus, and social welfare (Nguyen et al., 2018; Hatami and Firoozi, 2019; Mauricio et al.,
2019; Lu et al., 2020a, 2020b). Given incomplete information, monopolists may exaggerate
product quality and enterprises may lower their exaggeration degree of quality due to
quality control. Han and Li (2017) constructed a food safety evolutionary game model
based on the improved prospect theory and found that food safety is difficult to utilize as an
evolutionary stabilization strategy. Luo et al. (2018) constructed a game model among
enterprises, consumers, and government supervision agencies as well as analyzed the
effects of cost for food information searching, the subjective perception of consumers to
foods, and the authentication effect of government supervision departments on food safety
risks. Song et al. (2018) constructed an evolutionary game model of food safety
information disclosure and analyzed the main influencing factors of information disclosure
between government and enterprises. Xu et al. (2020) analyzed the quality decisions of
AE Food Quality Competition Among Companies and Government Food Safety Supervision Under Asymmetric Product Substitution
224 Amfiteatru Economic
subjects in the supply chain of agricultural products and discovered that a decentralization
decision is more beneficial to improve the quality of agricultural products compared to a
centralization decision.
However, these studies ignored product substitutability in the competition of enterprises
(Vives, 2008; Yang and Nie, 2016). Chen et al. (2018) considered product
substitutability among different enterprises when they were studying product quality
decisions of enterprises in a Cournot competition. Unfortunately, they ignored product
differences caused by factors such as the technological innovation of enterprises, further
resulting in differences of product substitutability (Vives, 2008; Cunha and Mota, 2020).
As a result, differences of product substitutability among different enterprises were
taken into account and a Cournot model of double oligopolistic competition was built to
analyze the yield and quality decisions of enterprises under no resource constraint and
under yield and quality resource constraints. Moreover, a government–enterprise game
model under government supervision was built to analyze the yield and quality
decisions of enterprises under government supervision and government supervision
input (Bojanić, 2015; Civera et al., 2019).
2. Methodology
2.1. Cournot competition model under no resource constraint
Given a double oligopolistic competition model under government supervision, there are
two competing enterprises (i and j) on the market and they acquire the maximum benefits
through a reasonable allocation of production capacity, that is, yield and quality of
products. Since two enterprises produce the same types of products, the products are
substitutable to some extent. The differences between two enterprises in product
substitutability caused by technological factors are likewise taken into account. It is
supposed that the enterprise with a stronger competitive edge has stronger product
substitutability than its competitor. In this case, the product demand function (Chen and
Nie, 2014; Nie, 2014; Chen et al., 2018) of enterprise i can be expressed as
1- - i i i jP q x x (1)
where α refers to the market scale of products, ix is the yield of enterprise i ,
iq denotes
the product quality of enterprise i , jx is the yield of enterprise j , and 1 0,1γ expresses
product substitutability of enterprise j . Similarly, the product demand function of
enterprise j can be gained as
2- - j j j iP q x x (2)
where jq denotes the product quality of enterprise j , 2 0,1γ expresses product
substitutability of enterprise i .Without loss of generality, we suppose that enterprise i is
more advantageous than enterprise j and its product substitutability is stronger: 2 1γ γ and
1 2, 0,1γ γ .
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 225
On the double oligopolistic market, enterprises determine product quality in the first stage
and determine product yield in the second stage. The revenue functions of enterprises i and
j are
2 2
1,
max2 2
i i
i ii i i j i i i
x q
q xπ α q x γ x x q x
2 2
2,
max2 2
j j
j j
j j j i j j jx q
q xπ α q x γ x x q x (3)
2.2. Cournot competition model under yield resource constraint
Under the yield capacity constraint, the production input resource constraint of the
enterprise leads to the limited yield of product. The enterprise will further consider the
influences of production input resource constraint on yield when it is making decisions
regarding yield. At this moment, the model is
2 2
1,
max2 2
i i
i ii i i j i i i
x q
q xπ α q x γ x x q x
2 2
2,
max2 2
j j
j j
j j j i j j jx q
q xπ α q x γ x x q x
S.T. i jθ x x R (4)
where R is the resource constraint related to yield, and θ is the yield conversion efficiency
of production resources. A high value of θ means low yield conversion efficiency of
production resources, while a low value of θ means high yield conversion efficiency of
production resources.
2.3. Cournot competition model under quality resource constraint
When there is a quality resource constraint, the production resource constraint can lead to
the limited product quality of the enterprise. The enterprise will further consider the
influences of resource constraint on quality when it is making a quality decision. In this
case, the Cournot competition model is as follows:
2 2
1,
max2 2
i i
i ii i i j i i i
x q
q xπ α q x γ x x q x
2 2
2,
max2 2
j j
j j
j j j i j j jx q
q xπ α q x γ x x q x
S.T. i jθ q q R (5)
AE Food Quality Competition Among Companies and Government Food Safety Supervision Under Asymmetric Product Substitution
226 Amfiteatru Economic
Similarly, R refers to quality-related resource constraint, and θ refers to the quality
conversion efficiency of production resources. A high value of θ implies the low quality
conversion efficiency of production resources, while a low value of θ indicates the high
quality conversion efficiency of production resources.
2.4. Modeling enterprise quality decision under government supervision and optimal
government supervision input
Yield and quality decisions of an enterprise under government quality supervision were
further taken into account on the basis of the Cournot competition model. Moreover,
product quality and yield of an enterprise are beyond the resource constraint. In this model,
the government will input certain costs to supervise the product quality of enterprises. With
the increase of supervision costs, the probability of the government detecting low-quality
products increases accordingly. The government will confiscate all profits of the enterprise
and impose a certain penalty for the discovered low-quality products. At this moment, the
utility function of government is
2
2 G
c cπ δ q π β
c (6)
where 1,
0,
q qδ q
q q
. q denotes the product quality of enterprise; q is the lowest product
quality required by the government; π is the profits of the enterprise; c is the supervision
input of the government; and c is the upper limit of government input, that is, the input
needed for the government to discover product quality problems timely and accurately.
When government input reaches the upper limit, the government will discover low-quality
products timely at the probability of 1 and impose a penalty on the enterprise. A higher
value of c reflects the bigger difficulties of government supervision. 2
2
c refers to the cost
for such input, and β refers to the government penalty on the enterprise. In this case, the
utility function of enterprise i is
2 2
1
2 2
1
12 2
12 2
i ii i i i j i i i
i ii i i j i i i
q xπ δ q α q x γ x x q x
q xc cδ q α q x γ x x q x β
c c
(7)
When the equilibrium product quality of enterprise i and enterprise j meet,
2
1 2 1 2 1
2 2 3 3
1 2 1 2 1 2
9 18 3 9
324 189 27
EQ
i
α γ γ γ γ γq q
γ γ γ γ γ γ
2
1 2 1 2 2
2 2 3 3
1 2 1 2 1 2
9 18 3 9
324 189 27
EQ
j
α γ γ γ γ γq q
γ γ γ γ γ γ
(8)
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 227
At this moment, the equilibrium product qualities of both enterprises under double
oligopolistic competition meet the lowest quality requirements. For the government, no
supervision is needed for enterprises in double oligopolistic competition, and thus the
optimal government supervision input is * 0c .
When the equilibrium product qualities of enterprises i and j meet,
2
1 2 1 2 1
2 2 3 3
1 2 1 2 1 2
9 18 3 9
324 189 27
EQ
i
α γ γ γ γ γq q
γ γ γ γ γ γ
2
1 2 1 2 2
2 2 3 3
1 2 1 2 1 2
9 18 3 9
324 189 27
EQ
j
α γ γ γ γ γq q
γ γ γ γ γ γ
(9)
The equilibrium product qualities of enterprises i and j are
1 1*
2 2
1 2 1 2
9 3
18 54
EQ
i
αγ qγ αq
γ γ γ γ
* EQ
jq q (10)
Under this circumstance, enterprise j increases product quality to make the equilibrium
quality of enterprise i lower than the minimum quality requirements. Enterprise i also
faces the options of equilibrium quality and lowest quality. Therefore, when the product
quality of enterprise j is lower than the lowest quality standards or product quality of two
enterprises that cannot meet the lowest quality standards, the double oligopolistic
competition model is
2 2
1,
max2 2
i i
i ii i i j i i i
x q
q xπ α q x γ x x q x
2 2
2,
max2 2
j j
j j
j j j i j j jx q
q xπ α q x γ x x q x
S.T. , i jq q q (11)
3. Analysis of results
3.1. Cournot competition model analysis under no resource constraint
According to the first-order conditions, the equilibrium product qualities of enterprises i
and j in a Cournot competition when there is no resource constraint are
2
1 2 1 2 1
2 2 3 3
1 2 1 2 1 2
9 18 3 9
324 189 27
EQ
i
α γ γ γ γ γq
γ γ γ γ γ γ
2
1 2 1 2 2
2 2 3 3
1 2 1 2 1 2
9 18 3 9
324 189 27
EQ
j
α γ γ γ γ γq
γ γ γ γ γ γ
(12)
AE Food Quality Competition Among Companies and Government Food Safety Supervision Under Asymmetric Product Substitution
228 Amfiteatru Economic
At this moment, the equilibrium product yields of enterprises i and j are
2
1 2 1 2 1 2 1
2 2 3 3
1 2 1 2 1 2
9 18 3 9
324 189 27
EQ
i
γ γ α γ γ γ γ γx
γ γ γ γ γ γ
2
1 2 1 2 1 2 2
2 2 3 3
1 2 1 2 1 2
9 18 3 9
324 189 27
EQ
j
γ γ α γ γ γ γ γx
γ γ γ γ γ γ
(13)
Hence, the equilibrium earnings of enterprises i and j are as follows:
22 2 2 2
1 2 1 2 1 2 1 2 1
22 2 3 3
1 2 1 2 1 2
3 18 54 18 3 9
2 324 189 27
EQ
i
γ γ γ γ α γ γ γ γ γπ
γ γ γ γ γ γ
22 2 2 2
1 2 1 2 1 2 1 2 2
22 2 3 3
1 2 1 2 1 2
3 18 54 18 3 9
2 324 189 27
EQ
j
γ γ γ γ α γ γ γ γ γπ
γ γ γ γ γ γ
(14)
Proposition 1: Under the oligopolistic competition, the advantageous party generates more
high-quality products and gains higher profits. In other words, there are EQ EQ
i jq q ,
EQ EQ
i jx x , and EQ EQ
i jπ π .
Proposition 1 shows equilibrium quality, yield, and profit under the double oligopolistic
competition. The enterprise with stronger product substitutability has higher product yield
and quality compared to the enterprise with weaker product substitutability. The stronger
product substitutability propels the enterprise to occupy market shares of the competitor by
increasing product yield and forcing the competitor to lower product yield. Meanwhile, it
offsets profit loss caused by increasing yield as a result of the price drop by improving
product quality.
3.2. Cournot competition model analysis under yield resource constraint
According to the first-order condition, the equilibrium product qualities of enterprises i
and j in a Cournot competition when there are yield resource constraints are
2 2
2 1 1 1 2 2 2*
1 2 1 2 1 2
2 6 2 2 19 11 36 3 3 8 12
5 8 3
i
γ R αθ R γ γ R αθ R γ αθ R γ γ αθ R γ αθ Rq
θ γ γ γ γ γ γ
2 2
1 2 2 2 1 1 1*
1 2 1 2 1 2
2 6 2 2 19 11 36 3 3 8 12
5 8 3
j
γ R αθ R γ γ R αθ R γ αθ R γ γ αθ R γ αθ Rq
θ γ γ γ γ γ γ
(15)
In this case, the equilibrium product yields of enterprises i and j are
2
1 2 1 2*
1 2 1 2
8 3 12
8 3
i
R γ γ γ γx
θ γ γ γ γ
2
2 1 2 1*
1 2 1 2
8 3 12
8 3
j
R γ γ γ γx
θ γ γ γ γ
(16)
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 229
Hence, the equilibrium earnings of enterprises i and j are
*2 *2
* * * * * * *
12 2
i ii i i j i i i
q xπ α q x γ x x q x
*2 *2
* * * * * * *
22 2
j j
j j j i j j j
q xπ α q x γ x x q x
(17)
Proposition 2: Under the oligopolistic competition, the advantageous party generates more
high-quality products when there is a production capacity constraint. In other words, * *i jq q
and * *i jx x . In this case, the resource constraint is a tight constraint, which means
* * i j
Rx x
θ.
Proposition 2 shows the equilibrium quality, yield, and profits under a yield capacity
constraint. Similar with Proposition 1, the enterprise with stronger product substitutability
has higher product yield and quality compared to the enterprise with weaker product
substitutability. In this case, the yield competition urges competing enterprises to make full
use of available yield resources under the Cournot competition.
Figure no. 1 shows that with the increase of 1γ , the profits of enterprise i decrease
gradually while the profits of enterprise j increase gradually, thus narrowing the profit
difference between the two enterprises. This outcome indicates that when the products of
enterprise i have stronger substitutability than those of enterprise j , the former can occupy
the market shares of the latter by increasing the yield and quality of its products, thus
enabling it to gain higher earnings.
Figure no. 1: Enterprise profit curve in double oligopolistic competition under yield
resource constraint ( 1α , 2 1γ , 1.5θ , and 1R )
AE Food Quality Competition Among Companies and Government Food Safety Supervision Under Asymmetric Product Substitution
230 Amfiteatru Economic
3.3. Cournot competition model analysis under quality resources constraint
According to the first-order condition, the equilibrium product qualities of enterprises i
and j in a Cournot competition when there is a quality capacity constraint are as follows:
2 2** 1 2 1 2 1 2 1
2 2
1 2 1 2 1 2
9 9 18 9 54
2 36 9 9 108
i
Rγ γ αθγ αθγ γ γ R γ R Rq
θ γ γ γ γ γ γ
2 2** 1 2 1 2 1 2 2
2 2
1 2 1 2 1 2
9 9 18 9 54
2 36 9 9 108
j
Rγ γ αθγ αθγ γ γ R γ R Rq
θ γ γ γ γ γ γ
(18)
At this moment, the equilibrium product yields of enterprises i and j are
2
1 2 1 2 1**
2 2
1 2 1 2 1 2
2 3 9 18
2 36 9 9 108
i
αθ R γ γ γ γ γx
θ γ γ γ γ γ γ
2
1 2 1 2 2**
2 2
1 2 1 2 1 2
2 3 9 18
2 36 9 9 108
j
αθ R γ γ γ γ γx
θ γ γ γ γ γ γ
(19)
Therefore, the equilibrium earnings of enterprises i and j are
* 2 * 2
* * * * * * *
12 2
* ** * * * * * *i i
i i i j i i i
q xπ α q x γ x x q x
* 2 * 2
* * * * * * *
22 2
* *
j j* * * * * * *
j j j i j j j
q xπ α q x γ x x q x
(20)
Proposition 3: Under oligopolistic competition, the advantageous party will produce more
high-quality products when there is a quality capacity constraint: ** **i jq q and ** **i jx x . At
this moment, the resource constraint is a tight constraint and ** ** i j
Rq q
θ.
Proposition 3 shows the equilibrium quality, yield, and profits when there is a quality
capacity constraint. Similar with Propositions 1 and 2, the enterprise with stronger product
substitutability has higher product yield and quality compared to the enterprise with weaker
product substitutability. On the one hand, strong product substitutability drives the
enterprise to use its substitutability advantages to expand yield. On the other hand, the price
drop caused by increasing yield is offset by increasing product quality. Nevertheless, the
enterprise with stronger product substitutability may not always gain higher profits due to
the quality resource constraint. In this case, the relationship between the two enterprises in
terms of earnings has the following three situations.
Figure no. 2 reveals that the profits of enterprise i are always lower than those of enterprise
j , and the profit difference between the two enterprises increases initially and then
decreases with the increase of 1γ . Both enterprises are in the deficit state. Owing to
differences in product substitutability, enterprise i will increase its yield and quality. The
enterprise with stronger product substitutability suffers more losses as a result of the
production resource constraint over product quality. The profit difference between the two
enterprises is relatively small when 1γ is very large or very small.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 231
Figure no. 2: Enterprise profit curve in double oligopolistic competition under quality
resource constraint ( 1α , 2 1γ , 1.1θ , and 1R )
Figure no. 3 shows that the profit of enterprise i is higher than that of enterprise j when 1γ
is relatively small. With the increase of 1γ , the profit of enterprise j is higher compared to
that of enterprise i . The profit of enterprise i presents a V-shaped variation with the
increase of 1γ . The profit of enterprise j is negatively related with
1γ . Both enterprises are
in deficit states. With the increase θ , production resources have stronger constraint over
product quality and restrict enterprise j , which has weaker product substitutability, to
offset early substitutability-induced loss by increasing product quality. This condition
brings higher profits to enterprise i , which has stronger product substitutability, by
increasing the yield. As 1γ increases, the substitutability advantages of enterprise i are
weakened. Therefore, enterprise j can gain relatively higher profits even though enterprise
i has higher yield and quality.
Figure no. 3: Enterprise profit curve in double oligopolistic competition under quality
resource constraint ( 1α , 2 1γ , 1.2θ , and 1R )
AE Food Quality Competition Among Companies and Government Food Safety Supervision Under Asymmetric Product Substitution
232 Amfiteatru Economic
Figure no. 4 shows that the profits of enterprise i are always higher than those of enterprise j .
With the increase of 1γ , the profit of enterprise i decreases, while the profit of enterprise j
increases gradually, thus gradually narrowing the profit difference between the two
enterprises. This outcome reflects that when θ is relatively high, production resource has a
very strong constraint over product quality and restricts enterprise j , which has weaker
substitutability, from offsetting loss caused by substitutability by improving product
quality. On the contrary, enterprise i can easily gain higher benefits by increasing product
yield.
Figure no. 4: Enterprise profit curve in double oligopolistic competition under quality
resource constraint ( 1α , 2 1γ , 1.8θ , and 1R )
3.4. Enterprise quality decision analysis under government supervision and optimal
government supervision input
Given government supervision, the equilibrium product qualities of enterprises i and j in
Cournot competition are
* * EQ EQ
i jq q q (21)
Equilibrium yields of enterprises i and j are
1 1*
1 2
3 3
9
EQ
i
q α αγ γ qx
γ γ
2 2*
1 2
3 3
9
EQ
j
q α αγ γ qx
γ γ (22)
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 233
Equilibrium profits of enterprises i and j are
2 2 22 2 2 2 2
1 2 1 2*
2
1 2
3 6 3 18 18 36 18 27 54 54
2 9
EQ
i
γ q αq α γ γ q αq α γ α αq qπ
γ γ
2 2 22 2 2 2 2
2 1 2 1*
2
1 2
3 6 3 18 18 36 18 27 54 54
2 9
EQ
j
γ q αq α γ γ q αq α γ α αq qπ
γ γ
(23)
Proposition 4: Given government supervision, both enterprises i and j produce products
according to the lowest quality requirements: * * EQ EQ
i jq q q . The advantageous party will
produce more products: * *EQ EQ
i jx x . In this case, the advantageous party possesses lower
profits: * *EQ EQ
i jπ π .
Proposition 4 shows the equilibrium quality, yield, and profit under government
supervision. At this circumstance, both enterprises produce products according to the
lowest quality requirements of the government to avoid relevant penalty risks. The
enterprise with stronger product substitutability has higher product yield but lower profits
compared to the enterprise with weak product substitutability. This result demonstrates that
government supervision forces both enterprises to improve product quality, which incurs
higher quality costs. This adverse factor influences the enterprise with stronger product
substitutability more, so the enterprise encounters difficulty in gaining advantages through
yield competition.
Meanwhile, it can be known from EQ EQ
i jπ π that
*1
EQ EQ
i i
c cπ β π
c c
*1
EQ EQ
j j
c cπ β π
c c (24)
Whether enterprises i and j are intended to improve product quality is determined by
enterprise j . When *
EQ EQ
j j
EQ
j
π πc
π βc
, the enterprise takes certain risks to produce low-quality
products. When *
EQ EQ
j j
EQ
j
π πc
π βc
, enterprises produce products according to the lowest quality
standards.
For the government,
2
max2
EQ
G jc
c cπ π β
c (25)
The optimal supervision input is
*
EQ
jπ βc
c (26)
AE Food Quality Competition Among Companies and Government Food Safety Supervision Under Asymmetric Product Substitution
234 Amfiteatru Economic
Therefore, enterprises will take certain risks to produce low-quality products when
2 2
* EQ EQ EQ
j j jπ β c π π , and they will produce products according to the lowest quality
standards when 2 2
* EQ EQ EQ
j j jπ β c π π . Obviously, the higher value of β is beneficial for
enterprises to produce products according to quality standards. On the contrary, the higher
value of c will make enterprises take certain risks to produce low-quality products. When
the product quality of an enterprise is lower than the lowest quality standards, the
government imposes a high penalty to the enterprise for disqualified products to force it to
improve product quality. Under this circumstance, the enterprise will produce products
according to the lowest quality standards to avoid the penalty from the government. When
there is great difficulty for government supervision, the enterprise decides to gain higher
benefits with disqualified products.
4. Discussions
Consumers are apt to buy products manufactured by enterprises with great brand influence.
Nevertheless, food quality safety accidents in leading food production enterprises have
been exposed frequently in recent years, causing a crisis of consumer trust in them.
Consumers do not choose products of large enterprises blindly. On the basis of the Cournot
competition, this study investigated the influences of product substitutability differences of
enterprises on their yield and quality decisions and interpreted the causes of quality safety
accidents in large enterprises. Different from previous studies (Migliore et al., 2015),
enterprises in double oligopolistic competition do not adopt the same yield and quality
decisions when they have different product substitutability. The enterprise with stronger
product substitutability will increase product yield and quality to occupy the market shares
of the enterprise with weaker product substitutability and gain higher profits. Hence,
government should strengthen quality supervision over small-sized enterprises with weak
product substitutability to assure their products meet quality standards.
However, large enterprises have to choose either product quality or profits when small-
sized enterprises choose to produce low-quality products. At this moment, large enterprises
may produce low-quality products to gain high profits under government supervision.
When the government faces great challenges in supervision or imposes a light penalty,
these enterprises are more likely to have quality accidents. Under this circumstance, the
government lowers the product quality safety risk of enterprises and propels enterprises to
produce products according to the lowest quality standards by increasing supervision input
or formulating stricter penalty measures. The four propositions give some enlightenment.
First, Propositions 1-3 demonstrate that to occupy the market, food enterprise with stronger
product substitutability has a stronger motivation to improve quality. In the situation
without food safety regulation policies, the product substitutability of food enterprises is
generally proportional to quality. In other words, the stronger the product substitutability,
the higher the product quality, whereas the weaker the product substitutability, the lower
the product quality. As a result, enterprises with strong product differences more easily
damage consumer welfare through quality standards (Garella and Petrakis, 2008). When the
implementation of government regulations on food safety has some resource constraints,
the government can observe the law and use product substitutability as an index to allocate
resources for implementing food safety regulations and maximize the implementation
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 235
effect. To pursue the maximum effect of food safety regulations, weaker product
substitutability requires more resources to implement food safety regulations.
Second, Proposition 4 reflects that to realize the lowest quality requirements regulated by the
government, food enterprises with strong product substitutability will produce more products
but gain lower profits. Meanwhile, food enterprises with weak product substitutability will
produce fewer products but gain more profits. Therefore, product substitutability determines
the profits of food enterprises with different product substitutability when they observe the
same lowest quality standards. The enterprise with stronger product substitutability has lower
profits. In the long run, capitals tend to be input into high-profit fields, indicating that
enterprises with weaker product substitutability will expel enterprises with stronger product
substitutability and influence the supply of the whole food market, finally influencing the
overall social welfare. This finding reveals that when the government formulates food safety
regulation policies, it should pay attention to the following two aspects. On the one hand, the
influences of product substitutability on food enterprises should be considered. On the other
hand, the market distortion effect wherein capitals prefer food enterprises with weaker
product substitutability in the long run because the same quality standards are observed
should be taken into account (Marette, 2007).
Third, on the other hand, according to the conclusion of the theoretical model, in order to
improve the quality of products without damaging the profits of enterprises, the
government can set higher quality standards for products with strong substitutability, and
the government can formulate qualified quality standards for products with low
substitutability, but at the same time strengthen supervision and inspection. Enterprises
with strong product substitutability have greater market competition pressure, more intense
quality competition, and sufficient number of production enterprises. Improving product
quality standards can help to eliminate enterprises with ordinary product quality, and ensure
that enterprises can increase enterprise profits. Enterprises with weak product
substitutability have low market competition pressure, low quality competition intensity,
and insufficient number of production enterprises. Setting up qualified but not high
standard product quality standards helps to ensure the number of enterprises. Under the
condition of strengthening supervision and inspection, ensuring the quality of enterprise
products and ensuring sufficient output can improve the profits of enterprises.
Conclusions
This study investigated the yield and quality decisions of food enterprises under no
resource constraint and under yield and quality resource constructs. Equilibrium yield and
quality of enterprises are analyzed by establishing Cournot models. The conclusions could
be drawn: (1) Under all conditions, enterprises with stronger product substitutability own
higher yield and ensure higher quality of food products than do enterprises with weaker
food product’s substitutability. (2) When there is no quality resource constraint but there is
yield resource constraint, enterprises with stronger product substitutability gain higher
profits. (3) In the situation of quality resource constraint, enterprises with stronger product
substitutability have lower profit than do enterprises with weaker product substitutability.
(4) Given government supervision, enterprises are urged to improve product quality due to
the high penalty or lower government supervision difficulty when the equilibrium quality of
an enterprise is lower than the lowest quality standards set by the government. At this
AE Food Quality Competition Among Companies and Government Food Safety Supervision Under Asymmetric Product Substitution
236 Amfiteatru Economic
moment, both enterprises produce products according to the lowest quality standards. (5)
On the contrary, low penalty or great government supervision difficulty forces enterprises
to take certain risks to produce low-quality products and gain higher profits.
The main policy implications of conducted study is to improve governance of food quality
undertaken by the government in order to avoid situation that companies try to increase
profits by lowering food production quality. State policies for securing food quality should
ensure that under government supervision, enterprises are urged to improve product quality
due to the high penalty or lower government supervision difficulty when the equilibrium
quality of an enterprise is lower than the lowest quality standards set by the government.
However, under government supervision, the punishment imposed by government will
increase expenditures and lower the profits of food enterprises. Government can reward or
subsidize food enterprises which comply with the food quality regulation. Compared with
punishment, the reward or subsidy will encourage enterprises to produce qualified food
products without damage to the profit of enterprises, although it will also raise government
expenditures. Therefore, government should pay attention to the food enterprises with
higher product substitutability and reward or subsidize them, which will protect them from
the damage of unqualified food products with weaker product substitutability.
This study investigates the equilibrium yield and quality of food products for enterprises
under a Cournot competition. However, some aspects still require further explorations. On
the one hand, different positions of enterprises on the market are ignored. The
advantageous party often occupies the dominant role on the market and becomes the
pioneer of the Steinberg game. On the other hand, this study considers the substitutability
of product yield to other products but ignores the impacts of product quality on product
substitutability. These limitations of current research are necessary to address in future
research.
Acknowledgement
This study was funded by the National Social Science Foundation of China (grant number
15ZDB168). We thank D. Fang, W. Lyu and M. Li for their advice.
References
Aray, Y., Veselova, A., Knatko, D. and Levchenko, A., 2020. Drivers for adoption of
sustainability initiatives in supply chains of large Russian firms under environmental
uncertainty. Corporate Governance: The International Journal of Business in Society.
Ahead-of-print. https://doi.org/10.1108/CG-02-2020-0048.
Besanko, D, Dubé, J. and Gupta, S., 2005. Own-brand and cross-brand retail pass-through.
Marketing Science, 24(1), pp.123-137.
Bojanić, B., 2015. Analysis of supervision of the quality management system from the
highest management of an organization. Tehnicki Glasnik-Technical Journal, 9(3),
pp.321-326.
Chen, C., Zhang, J. and Delaurentis, T., 2014. Quality control in food supply chain
management: An analytical model and case study of the adulterated milk incident in
China. International Journal of Production Economics, [e-journal] 152, pp.188-199.
DOI: 10.1016/j.ijpe.2013.12.016.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 237
Chen, Y. and Nie, P., 2014. Duopoly innovation under product externalities. Economic
Research-Ekonomska Istrazivanja, 27(1), pp.232-243.
Chen, Y., Huang, S., Mishra, A. K. and Wang, X. H., 2018. Effects of input capacity
constraints on food quality and regulation mechanism design for food safety
management. Ecological Modelling, [e-journal] 385(C), pp.89-95. DOI:
10.1016/j.ecolmodel.2018.03.011.
Chen, Y., Nie, P. and Yang, Y., 2017. Effects of corporate social responsibility on food
safety. Agricultural Economics-Zemedelska Ekonomika, 63(12), pp.539-547.
Civera, C., De Colle, S. and Casalegno, C., 2019. Stakeholder engagement through
empowerment: The case of coffee farmers. Business Ethics: A European Review, 28(2),
pp.156-174.
Cunha, M. and Mota, F., 2020. Coordinated effects of corporate social responsibility.
Journal of Industry, Competition and Trade, [e-journal] 20, pp.617-641.
DOI: 10.1007/s10842-020-00344-2.
Datta, A. K., Ukidwe, M. S. and Way, D. G., 2020. Simulation‐ based enhancement of
learning: the case of food safety. Journal of Food Science Education, 19(3), pp.192-211.
Garella, P. G. and Petrakis, E., 2008. Minimum quality standards and consumers'
information. Economic Theory, 36(2), pp.283-302.
Han, F. and Li, H., 2017. Food safety evolutionary game simulation model based on
improved prospect theory. Journal of Interdisciplinary Mathematics, [e-journal] 20(6-
7), pp.1349-1354. https://doi.org/10.1080/09720502.2017.1384216
Hatami, A. and Firoozi, N., 2019. A dynamic stakeholder model: an other‐ oriented ethical
approach. Business Ethics: A European Review, 28(3), pp.349-360.
Liu, P. and Ma, L., 2016. Food scandals, media exposure, and citizens’ safety concerns: A
multilevel analysis across Chinese cities. Food Policy, [e-journal] 63, pp.102-111.
DOI: 10.1016/j.foodpol.2016.07.005
Liu, P., 2010. Tracing and periodizing China’s food safety regulation: A study on China's
food safety regime change. Regulation & Governance, 4(2), pp.244-260.
Liu, R., Pieniak, Z. and Verbeke, W., 2013. Consumers’ attitudes and behaviour towards
safe food in China: a review. Food Control, 33(1), pp.93-104.
Lu, J. T., Ren, L. C., Yao, S. Q., Qiao, J. Y., Mikalauskiene, A. and Streimikis, J., 2020.
Exploring the relationship between corporate social responsibility and firm
competitiveness. Economic Research-Ekonomska Istraživanja, 33(1), pp.1621-1646.
Lu, J. T., Ren, L. C., Zhang, C., Liang, M. S., Abrham, J. and Streimikis, J., 2020.
Assessment of Corporate Social Responsibility performance and state promotion
policies: a case study of The Baltic States. Journal of Business Economics and
Management, 21(4), pp.1203-1224.
Luo, J., Ma, B., Zhao, Y. and Chen, T., 2018. Evolution model of health food safety risk
based on prospect theory. Journal of Healthcare Engineering, 2018, pp.1-12.
DOI: 10.1155/2018/8769563.
Luo, X., Han, Y., Chen, X., Tang, W., Yue, T. and Li, Z., 2020. Carbon dots derived
fluorescent nanosensors as versatile tools for food quality and safety assessment: A
review. Trends in Food Science and Technology, 95, pp.149-161.
Marette, S., 2007. Minimum safety standard, consumers' information and competition.
AE Food Quality Competition Among Companies and Government Food Safety Supervision Under Asymmetric Product Substitution
238 Amfiteatru Economic
Journal of Regulatory Economics, 32(3), pp.259-285.
Mauricio, A., Latapí, A, Lára J. and Brynhildur D., 2019. A literature review of the history
and evolution of corporate social responsibility. International Journal of Corporate
Social Responsibility, 4(1), pp.1-23.
Migliore, G., Schifani, G. and Cembalo, L., 2015. Opening the black box of food quality in
the short supply chain: Effects of conventions of quality on consumer choice. Food
Quality and Preference, [e-journal] 39, pp.141-146. DOI: 10.1016/j.foodqual.
2014.07.006.
Neumann-Langdon, P., Oviedo-Silva, C., Suazo-Schwencke, A., Ramis-Lanyon, F. and
Delgado-Neira, P., 2019. Technological and management aspects of the anaerobic co-
digestion of sewage sludge with vegetable and organic wastes. Dyna, 94(5), pp.574-578.
Nguyen, M., Bensemann, J. and Kelly, S., 2018. Corporate social responsibility (CSR) in
vietnam: a conceptual framework. International Journal of Corporate Social
Responsibility, 3(1), pp.1-12.
Nie, P., 2014. Effects of capacity constraints on mixed duopoly. Journal of Economics,
112(3), pp.283-294.
Parker, J. S., DeNiro, J., Ivey, M. L. and Doohan, D., 2016. Are small and medium scale
produce farms inherent food safety risks? Journal of Rural Studies, [e-journal] 44,
pp.250-260. DOI: 10.1016/j.jrurstud.2016.02.005.
Pei, X., Tandon, A., Alldrick, A., Giorgi, L., Huang, W. and Yang, R., 2011. The China
melamine milk scandal and its implications for food safety regulation. Food Policy,
36(3), pp.412-420.
Peng, Y., Li, J., Xia, H., Qi, S. and Li, J., 2015. The effects of food safety issues released
by we media on consumers’ awareness and purchasing behavior: A case study in China.
Food Policy, [e-journal] 51(C), pp.44-52. DOI: 10.1016/j.foodpol.2014.12.010.
Petrovic, D., Jurisic, M., Tadic, V., Plascak, I. and Barac, Z., 2018. Different sensor
systems for the application of variable rate technology in permanent crops. Tehnicki
Glasnik-Technical Journal, 12(3), pp.188-195.
Pineda-Escobar, M. A., 2019. Moving the 2030 agenda forward: SDG implementation in
Colombia. Corporate Governance: The International Journal of Business in Society, 19
(1), pp.176-188.
Pinstrup-Andersen, P., 2009. Food security: definition and measurement. Food Security,
1(1), pp.5-7.
Rodriguez-Parada, L., Pardo-Vicente, M. A. and Mayuet-Ares, P. F., 2018. Digitalization
fresh food using 3D scanning for custom packaging design. Dyna, 93(6), pp.681-688.
Rong, A., Akkerman, R. and Grunow, M., 2011. An optimization approach for managing
fresh food quality throughout the supply chain. International Journal of Production
Economics, 131(1), pp.421-429.
Rouvière, E., 2016. Small is beautiful: firm size, prevention and food safety. Food Policy,
63, pp.12-22.
Shaffer, G. and Zettelmeyer, F., 2004. Advertising in a distribution channel. Marketing
Science, 23(4), pp.619-628.
Song, Y., Shen, N. and Liu, D., 2018. Evolutionary game and intelligent simulation of food
safety information disclosure oriented to traceability system. Journal of Intelligent and
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 239
Fuzzy Systems, 35(3), pp.2657-2665.
Tirado, M. C., Clarke, R., Jaykus, L. A., Mcquatters-gollop, A. and Frank, J. M., 2010.
Climate change and food safety: A review. Food Research International, 43(7),
pp.1745-1765.
Tomerlin, R., Tomisa, M. and Vusic, D., 2019. The influence of printing, lamination and
high pressure processing on spot color characterization. Tehnicki Glasnik-Technical
Journal, 13(3), pp.218-225.
Van Der Vorst, J. G., Tromp, S. O. and Van Der Zee, D. J., 2009. Simulation modelling for
food supply chain redesign; integrated decision making on product quality,
sustainability and logistics. International Journal of Production Research, 47(23),
pp.6611-6631.
Vives, X., 2008. Innovation and competitive pressure. The Journal of Industrial
Economics, 56(3), pp.419-469.
Wang, J., Chen, T. and Wang, J., 2015. Research on cooperation strategy of enterprises’
quality and safety in food supply chain. Discrete Dynamics in Nature and Society, [e-
journal] 2015. https://doi.org/10.1155/2015/301245
Wang, Z., Mao, Y. and Gale, F., 2008. Chinese consumer demand for food safety attributes
in milk products. Food Policy, 33(1), pp.27-36.
Xu, J., Yao, G. and Dai, P., 2020. Quality decision-making behavior of bodies participating
in the agri-Foods e-supply chain. Sustainability, 12(5), pp.1874.
Yan, Y., 2012. Food Safety and Social Risk in Contemporary China. The Journal of Asian
Studies, 71(3), pp.705-729.
Yang, Y. and Nie, P., 2016. Asymmetric competition in food industry with product
substitutability. Agricultural Economics-Zemedelska Ekonomika, 62(7), pp.324-333.
Zheng, Q., Pan, X. A. and Vakharia, A. J., 2020. Common retailer channel revisited: the role of
supply network size. Production and Operations Management, 29(9), pp.2175-2181.
AE Are Positive and Negative Outcomes of Organizational Justice Conditioned by Leader–Member Exchange?
240 Amfiteatru Economic
ARE POSITIVE AND NEGATIVE OUTCOMES OF ORGANIZATIONAL
JUSTICE CONDITIONED BY LEADER–MEMBER EXCHANGE?
Or Shkoler1, Aharon Tziner2,3, Cristinel Vasiliu4
and Claudiu-Nicolae Ghinea5 1) Independent researcher, Israel
2)Peres Academic Center, Rehovot, Israel 3)Netanya Academic College, Netanya, Israel
4) University of Economic Studies, Bucharest, Romania 5)S.C. Therme Nord București SRL, Otopeni, Romania
Please cite this article as:
Shkoler, O., Tziner, A., Vasiliu, C. and Ghinea,
C.N., 2021. Are Positive and Negative Outcomes of
Organizational Justice Conditioned by Leader–
Member Exchange? Amfiteatru Economic, 23(56),
pp. 240-258.
DOI: 10.24818/EA/2021/56/240
Article History
Received: 22 September 2020
Revised: 29 October 2020
Accepted: 15 December 2020
Abstract
The workplace is complex, comprising many entities (abstract and tangible) – affective
states, attitudes, and perceptions, but also workers and managers themselves and their
behaviors. Understanding the link between them is vital for organizational prosperity. In the
current paper, the perceptions of organizational justice are investigated as a precursor to
two important outcomes – organizational citizenship behavior and counterproductive work
behavior. To that end, a two-step research study was conducted to test a moderated-
mediation model. First, a pilot study of 93 Romanian employees was undertaken, followed
by a larger study consisting of 3293 Romanian workers. There were distinct differences
between the two studies. Implications, limitations, and suggestions for future research are
discussed.
Keywords: counterproductive work behavior (CWB), leader–member exchange (LMX),
moderated-mediation, organizational citizenship behavior (OCB), organizational justice,
work motivation.
JEL Classification: D23, M54, O15
Corresponding author, Or Shkoler – e-mail: [email protected]
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 241
Introduction
When managers are learning about employees, they need to understand what types of
perceptions, feelings, and reactions they should elicit from personnel under their direction.
In the research presented in this paper, we focus on connections between a delimited,
parsimonious set of attitudes – perceptions of organizational justice and organizational
citizenship behavior – in conjunction with the dynamic personal states of leader–member
exchange (LMX), motivation, and workplace misbehavior. These attitudes and personal
states have consistently been shown to explain great variability in critical outcomes such as
turnover (Bernerth and Walker, 2012), work performance (Wang, et al., 2010), and burnout
(Faragher, Cass and Cooper, 2013). The model we tested in this research is thus articulated
in Figure no. 1 and includes two central attitudes – organizational justice (comprising
procedural, interactional, and distributive justice) and organizational citizenship behavior
(OCB) – and three critical personal states of LMX, work motivation, and workplace
misbehavior (counterproductive work behavior - CWB).
Contribution and Focus of the Current Research
The main goal of the current research is to examine all associations between the variables
displayed in Figure 1. Indeed, most of the dyadic relationships between the variables in
Figure 1 (e.g., organizational justice and OCB; work motivation and OCB or CWB, etc.)
have been studied in the past (e.g., Eskew, 1993; Karriker and Williams, 2009; Al-A’wasa,
2018; Ugaddan and Park, 2019). However, to the best of our knowledge, no study has
scrutinized the network of interrelationships between the variables as comprised by the
model in Figure 1 using a moderated-mediation approach.
Furthermore, most of these associations were studied in Western countries such as the
USA, Australia, Canada, the UK, and to a (very) lesser extent, in Eastern countries or post-
communist countries. As such, we chose to conduct the current research in Romania. We
identified Romania as a relatively virgin (and fertile) field of research on human resources
management (Buzea, 2014). As an ex-communist country (in Central and Eastern Europe;
CEE), Romania joined the European Union only in 2007. “The greater explanatory power
of the contextual paradigm in such cases (namely CEE) at least is manifest; the poverty of
attempts to explain developments there by contrasting them with the universalistic
conception of HRM is clear” (Mayrhofer, Brewster and Morley, 2000, p. 12). Thus, as
implied by the contingency perspective, human resource strategies and managerial practices
will be more or less effective according to critical contingencies in the environment (Delery
and Doty, 1996), such as the Romanian culture.
We now proceed with a review of the literature in support of our model in Figure 1.
1. Perceived Organizational Justice
An important antecedent variable within this analysis is perceived organizational justice –
that is, the degree to which employees think or feel they are provided with apposite, just
and considerate treatment, accurate and sufficient information, and rewards and resources
(Cohen-Charash and Spector, 2001; Colquitt, et al., 2001; Ambrose and Schminke, 2009).
These perceptions are a product of overall impressions based on the consequences of
arbitrary organizational events and employees’ own assessments of specific components of
AE Are Positive and Negative Outcomes of Organizational Justice Conditioned by Leader–Member Exchange?
242 Amfiteatru Economic
the organization, including managers and work colleagues (Hollensbe, Khazanchi and
Masterson, 2008).
Typically, organizational justice comprises procedural, interactional, and distributive justice
(for further reading, see Niehoff and Moorman, 1993; Cohen-Charash and Spector, 2001;
Colquitt, et al., 2001). Organizational justice has been researched extensively in the past, but
most studies have emphasized its role as a predictor of work outcomes and not as a possible
outcome in its own right (e.g., Brienza and Bobocel, 2017; Shkoler and Tziner, 2017).
2. Work Motivation
Work motivation is another variable we investigated with regard to the relationship
between predictors and outcomes. Work motivation may be understood as the
psychological dynamism that engenders complex cycles of thoughts and behavior directed
towards a goal (Tziner, Fein and Oren, 2012). Motivation is what energizes us to persevere
until goals are attained. Scholars of work motivation try to ascertain the processes by which
an individual’s internal, psychological forces – combined with external, environmental
forces – influence the persistence, direction, and intensity of an individual’s behavior aimed
at reaching that goal (Kanfer, Frese and Johnson, 2017). However, Pinder (2014, p. 11)
provides another, and currently the most accepted, working definition of work motivation:
“Work motivation is a set of energetic forces that originate both within individuals, as well
as beyond an individual’s being, to initiate work-related behavior and to determine its form,
direction, intensity, and duration.” In that vein, work motivation emanates from the
interaction between the external organizational and societal environments and a person’s
characteristics (Latham and Pinder, 2005). In sum, motivation may be regarded as the
impetus that drives one to participate in an activity, and we consider the perceptions of
organizational justice as an individual antecedent to motivation in the present model.
Organizational Justice and Work Motivation
Organizational justice (distributive, procedural, interactional), that is, employee perceptions
of fairness in the workplace, may have an impact on the employees’ drive to work. For
example, a worker who perceives that he or she is being treated fairly (e.g., reward/bonus
distribution, the fairness of managerial decisions – the manner in which they were reached
and how the immediate manager has proceeded in this regard, etc.), he or she would feel
obliged to reciprocate the fair treatment received (Gouldner, 1960; Blau, 1964). Hence, the
balance between an employee’s input at work (e.g., expertise, knowledge, effort invested)
and what he or she receives in return (e.g., monetary compensation, good working
conditions, job prestige, challenging work) will be (e.g., Adams, 1965) maintained. Thus,
we hypothesize that:
H1: Organizational justice perceptions (distributive, procedural, interactional) positively
correlate with work motivation.
3. Organizational Citizenship Behavior
Researchers argue that organizations benefit from employees who are prepared to
contribute beyond their formal job duties (Organ, Podsakoff and MacKenzie, 2006) – in
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 243
other words, when they demonstrate organizational citizenship behavior (OCB). OCB is
individual behavior that is discretionary, not overtly acknowledged by the formal reward
system, and that promotes the effective performance of an organization (Organ, Podsakoff
and MacKenzie, 2006). In today’s increasingly dynamic and competitive organizational
environment, OCB is a greatly valued contribution. It is therefore no surprise that attention
to OCB has been increasing, with Podsakoff et al.’s (2009) meta-analysis noting the
publication of more than 400 articles on OCB and related constructs since 2000.
Organizational Justice and OCB
As stated in section 2.3, positive perceptions of fairness may induce greater work drive.
However, this is mostly an attitudinal outcome aspect of such perceptions. The different
perceptions of justice in the workplace (distributive, procedural, interactional) may also
promote de facto action by the employee. As mentioned, positive perceptions are likely to
be reciprocated by positive action (Gouldner, 1960; Blau, 1964), and that means that the
worker would put extra in extra effort at work to “compensate” the good treatment he or
she perceives. Hence, we hypothesize:
H2: Organizational justice perceptions (distributive, procedural, interactional) positively
correlate with OCB.
4. Workplace Misbehavior
In recent years, counterproductive work behavior (Cohen-Charash and Mueller, 2007) and
workplace misbehavior (Cohen-Charash and Mueller, 2007; Dilchert, et al., 2007;
Bodankin and Tziner, 2009) have received considerable attention from researchers, as these
manifestations have significant psychological, sociological, and economic implications for
the working environment (Aubé, et al., 2009; Bodankin and Tziner, 2009).
Counterproductive behavior and misbehavior might be directed towards the organization or
its workers and management, and hence are costly for both the organization and the
individual (Bennett and Robinson, 2003). These behaviors almost always infringe upon
important organizational norms and cause damage to an organization’s objectives,
procedures, productivity, profitability, and employees themselves (Vardi and Weitz, 2002;
Spector, et al., 2006; Aubé, et al., 2009). Work misbehavior includes employees’ reducing
or withdrawing their input to balance the social exchange process (Greenberg and Scott,
1996); feeling negatively towards the organization; feeling less motivated; exhibiting
distrust (toward to the manager and/or the organization); and even retaliating against the
organization (Skarlicki and Folger, 1997), which might manifest as harassment, theft or
sabotage (Bennett and Robinson, 2000; Spector, et al., 2006). Hence, work misbehavior is
hypothesized as negatively associated with job satisfaction.
Organizational Justice and CWB
Just as positive perceptions of fairness may promote increased citizenship behavior (section
2.5), the opposite is also true. In other words, should an employee perceive that the
distributive, procedural, and/or interactional aspects of justice at his or her workplace are
negative, this might prompt the worker to engage in negative behaviors. The worker would
do so in order to resume a balance between what he or she receives from the organization
and what he or she gives in return (Gouldner, 1960; Blau, 1964; Adams, 1965), and this
AE Are Positive and Negative Outcomes of Organizational Justice Conditioned by Leader–Member Exchange?
244 Amfiteatru Economic
may manifest in reducing his or her work output and performance, and even in destructive
behavior. Consequently, we hypothesize that:
H3: Organizational justice perceptions (distributive, procedural, interactional) negatively
correlate with CWB.
Organizational Justice, Work Motivation, OCB and CWB
As previously articulated, organizational justice perceptions may induce increased work
motivation, but also can promote positive/negative behaviors (OCB and CWB,
respectively). This leads us to predict that work motivation acts as a mediational
mechanism in our model, meaning that justice perceptions may affect the worker’s
motivation to work, which in turn may elicit (increased) positive or negative behaviors at
work, regardless of the direct effect justice may have on said outcomes. As such, we
hypothesize the following:
H4: Work motivation mediates the relationships between organizational justice perceptions
(distributive, procedural, interactional) and CWB.
H5: Work motivation mediates the relationships between organizational justice perceptions
(distributive, procedural, interactional) and OCB.
5. Buffering Effect – Leader–Member Exchange (LMX)
The theory of leader–member exchange argues that in dyadic relationships, managers tend
to use different approaches for each of their employees (Graen and Uhl-Bien, 1995). In
turn, each relationship or management style provokes different attitudes in subordinates,
which drives the latter to behave differently from each other (Ilies, Nahrgang and
Morgeson, 2007). Capitalizing on SET (Blau, 1964) and reciprocity theory (Gouldner,
1960), subordinates in good/bad relations with their supervisor or manager (that is,
high/low LMX) feel obligated/reluctant to reciprocate (Adams, 1965).
Thus, LMX is one of the pivotal constituents of the workplace social network (Cole,
Schaninger Jr. and Harris, 2002), and underlines the essential role that managers play in
influencing their employees’ performance by providing them with support and other
resources (Hobfoll, 1989; Zagenczyk, et al., 2015), which ultimately reduces their physical
and emotional exhaustion – the core elements of work burnout (e.g., Huang, et al., 2010).
In spite of a plethora of research on LMX, to the best of our knowledge, less is known
about the effects of individuals’ dispositional differences (e.g., Maslyn, Schyns and Farmer,
2017) and the effects of cultural and demographic parameters on leader–member
interrelations (Zagenczyk, et al., 2015).
This intimate nature of LMX may have a more profound impact on the daily work routines
of employees. The dyadic relationship between a worker and his or her manager may have
various effects (e.g., increasing/decreasing organizational support, rewards, commitment,
and so on). These can, to a certain extent, affect previously conceived associations. For
example, a good relationship with the manager (i.e., high LMX) is conducive to positive
perceptions of justice and, thus, given a situation where the employee already has good
relations with the manager, this may enhance the positive effect that justice perceptions
have on work motivation. As another example, good relations with the manager may act as
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 245
a buffer, or as a kind of shock absorber, which will mitigate the negative effect that justice
perceptions have on CWB. As such, we hypothesize:
H6: LMX moderates the relationships in the model, as a general conditional factor.
6. Hypotheses Summary and Research Model
To conclude, the model in Figure 1 summarizes all the predicted relationships between the
variables of investigation articulated so far.
As was elaborated earlier, the literature review has led us to conceive the following
hypotheses in a comprehensive moderated-mediation model:
H1: Organizational justice perceptions (distributive, procedural, interactional) positively
correlate with work motivation.
H2: Organizational justice perceptions (distributive, procedural, interactional) positively
correlate with OCB.
H3: Organizational justice perceptions (distributive, procedural, interactional) negatively
correlate with CWB.
H4: Work motivation mediates the relationships between organizational justice perceptions
(distributive, procedural, interactional) and CWB.
H5: Work motivation mediates the relationships between organizational justice perceptions
(distributive, procedural, interactional) and OCB.
H6: LMX moderates the relationships in the model, as a general conditional factor.
Figure no. 1. Model for the current research
Note: D_Justice = distributive justice. P_Justice = procedural justice. I_Justice =
interactional justice. LMX = leader–member exchange. CWB = counterproductive work
behavior. OCB = organizational citizenship behavior.
AE Are Positive and Negative Outcomes of Organizational Justice Conditioned by Leader–Member Exchange?
246 Amfiteatru Economic
Study 1
Method
This study was a pilot to test the model presented above on a small-scale sample in order to
obtain a first estimation of the relationships depicted in our model.
Participants
There were 93 subjects in the study, 41.9% males and 58.1% females aged 19-57 years
(M = 33.44, SD = 9.52).
Measures
To ensure quality control of the questionnaires, the measures were translated into
Romanian and then translated (a completed translation) back into the original language
(i.e., English). The new translation was then compared with the original text, reconciling
any meaningful differences between the two.
Organizational justice was measured using Niehoff and Moorman’s (1993) Justice Scale,
comprising 20 Likert-type items between 1 (completely disagree) and 6 (completely agree).
The measure is divided into three different perceptions of justice: (1) Distributive: for
instance, “I consider my work load to be quite fair” ( = .74, M = 3.92, SD = 0.87);
(2) Procedural: for instance, “All job decisions are applied consistently across all affected
employees” ( = .92, M = 3.78, SD = 1.15); and (3) Interactional: for instance, “When
decisions are made about my job, the general manager treats me with respect and dignity”
( = .96, M = 4.54, SD = 1.06).
Work motivation was gauged using the Work Extrinsic and Intrinsic Motivation Scale
(WEIMS; Tremblay et al., 2009), comprising 18 Likert-type items ranging from 1 (does not
correspond at all) to 6 (corresponds exactly): for instance, “The reason for being involved
in my job is the satisfaction I experience when I am successful at doing difficult tasks”
( = .86, M = 4.36, SD = 0.68).
Leader–member exchange was gauged using the Leader–Member Exchange Multi-
Dimensional Measure (LMX-MDM; Liden and Maslyn, 1998), which includes 12 Likert-
type items between 1 (strongly disagree) and 6 (strongly agree): for instance, “My
supervisor would defend me to another in the organization if I made an honest mistake”
( = .78, M = 4.49, SD = 0.68).
Counterproductive work behavior was gauged with Bennett and Robinson’s (2000)
Interpersonal and Organizational Deviance Scale (IODS), comprising 19 Likert-type items
between 1 (never) and 6 (every day): for example, “I deliberately worked slower than I
could” ( = .93, M = 2.50, SD = 1.11).
Organizational citizenship behavior was gauged with a scale derived from the work of
Williams and Anderson (1991), comprising 14 Likert-type items between 1 (strongly
disagree) and 6 (strongly agree): for instance, “I help others who have been absent” ( =
.75, M = 4.36, SD = 0.69).
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 247
Procedure
The items of the questionnaire were initially written in English and then translated into
Romanian, utilizing the back-translation procedure (Brislin, 1980). Amendments to items
were made if needed to ensure semantic equivalence. Only then was the questionnaire
administered to participants.
The field research was based on the administration of the translated questionnaires by
students who participated as research assistants. The participation of the respondents in the
questionnaire was voluntary (i.e., informed consent). In the questionnaire, the participants
were assured of our respect for the principle of data confidentiality and anonymity
throughout the entire collection, processing, storage, dissemination, and archiving flow.
Thus, it is impossible to identify the respondents whatsoever. There are no questions in the
survey regarding the names, e-mail addresses, telephone numbers or other personal data of
the respondents. In this way, the information was treated responsibly according to
European Union legislation in the field of personal data and ethical standards.
Results
Common-method bias (CMB). To evaluate the extent to which variable intercorrelations might
be an artifact of common method variance (CMV), as suggested by Podsakoff et al. (2003) we
utilized two methods: (a) Harman’s single-factor model (where all items are loaded into one
common/marker factor); and (b) a common latent factor (CLF) model (where all items are
loaded into both their expected factors and one latent common method factor).
Harman’s single-factor method accounted for only 22.64% of the explained variance: 2(3,070)
= 7,771.49, p = .000, 2/df = 2.53, CFI = .55, NFI = .61, GFI = .24, SRMR = .19, RMSEA
(90% CI) = .29 (.11-.35), p-close = .000. Furthermore, the CLF method of analysis produced
20.39% of the explained variance: 2(2,990) = 6,758.87, p = .000, 2/df = 2.26, CFI = .63, NFI
= .71, GFI = .28, SRMR = .16, RMSEA (90% CI) = .18 (.10-.27), p-close = .000. Although the
results do not completely exclude the possibility of same-source bias (CMV), according to
Podsakoff et al. (2003), less than 50% (R2 < .50) of the explained variance accounted for by the
first emerging factor indicates that CMB is an improbable explanation of our findings, in
conjunction with the bad model fit for each analysis.
Table 1 displays the between-variable zero-order correlational relationships in the research.
Table no. 1. Pearson correlation matrix
1 2 3 4 5 6
1. D_Justice
2. P_Justice .56***
3. I_Justice .27** .73***
4. Motivation .23* .13 .29**
5. LMX .33** .55*** .60*** .42***
6. CWB .38*** .37*** .12 .02 .11
7. OCB –.16 .002 .09 .37* .19* –.35***
Note: *p < .05, **p < .01, ***p < .001. D_Justice = distributive justice; P_Justice =
procedural justice; I_Justice = interactional justice; LMX = leader–member exchange;
CWB = counterproductive work behavior; OCB = organizational citizenship behavior.
AE Are Positive and Negative Outcomes of Organizational Justice Conditioned by Leader–Member Exchange?
248 Amfiteatru Economic
To test the model, an SEM (structural equation modeling) analysis with multiple-group
analysis was employed using the IBM AMOS (v. 23) software package. The model did not
have a good fit: 2(df) = 59.51(13), p = .000, 2/df = 4.58, SRMR = .15, GFI = 0.87, CFI =
.74, NFI = .71, NNFI = .39, RMSEA (90% CI) = .19 (.15-.25), p-close = .000. Table 2
displays the results from the path analysis, while LMX is a moderator (“Low LMX” = data
below or equal to the LMX’s median, while “High LMX” = data above the LMX’s
median), and Z-tests in order to discern whether the differences in estimators between the
two LMX groups are statistically significant. Table 3 depicts the indirect effects analysis
for the mediation effects, as per the hypotheses. Figure 2 portrays the results in Table 2 on a
path diagram.
Table no. 2. SEM path results with standardized regression coefficients
and difference tests
Low LMX High LMX Difference Test
Path β Sig. β Sig. Z-score
D_Justice Motivation .46 .000 .18 .233 –2.59***
P_Justice Motivation –.22 .235 –.63 .000 –0.90
I_Justice Motivation –.01 .949 .82 .000 3.65***
Motivation CWB –.08 .608 .15 .396 0.95
Motivation OCB .51 .000 .82 .000 3.48***
D_Justice CWB .18 .263 .16 .387 0.11
D_Justice OCB –.47 .001 –.22 .171 1.37
P_Justice CWB .36 .075 .47 .044 1.18
P_Justice OCB .32 .080 .22 .271 –0.03
I_Justice CWB –.27 .179 –.19 .363 –0.49
I_Justice OCB .01 .944 –.64 .000 –3.24***
Note: *p < .05, **p < .01, ***p < .001. D_Justice = distributive justice; P_Justice =
procedural justice; I_Justice = interactional justice; LMX = leader–member exchange;
CWB = counterproductive work behavior; OCB = organizational citizenship behavior.
Table no. 3. Mediation (indirect) effects analyses
Low LMX High LMX
Paths LL UL Sig. LL UL Sig.
D_Justice Motivation OCB –0.25 0.23 .890 0.28 1.23 .001
D_Justice Motivation CWB –0.08 0.09 .756 –0.14 0.48 .229
P_Justice Motivation OCB –0.47 0.10 .249 –1.23 –0.22 .001
P_Justice Motivation CWB –0.04 0.28 .508 –0.31 0.15 .256
I_Justice Motivation OCB –0.01 0.56 .051 –0.10 0.70 .274
I_Justice Motivation CWB –0.25 0.10 .631 –0.02 0.19 .214
Note: Analyses used bootstrapping (95% bias-corrected, 5000 resamples). LL = lower limit
of the CI; UL = upper limit of the CI; D_Justice = distributive justice; P_Justice =
procedural justice; I_Justice = interactional justice; LMX = leader–member exchange;
CWB = counterproductive work behavior; OCB = organizational citizenship behavior.
As can be seen in Table 2, taking into account the group comparison (Low LMX vs. High
LMX), there several differences in the correlational relationships between the variables.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 249
Combined with the Z-tests for the differences, this indicates that LMX is indeed a
moderator in this context. However, we can also see some counterintuitive associations
(e.g., negative links between distributive justice and OCB, or a positive link between
procedural justice and CWB). In addition, Table 3 has shown that motivation is indeed a
mediator, but only between distributive and procedural justice perceptions and OCB (for
the High LMX group only).
Figure no. 2. Path diagram with SEM results (Study 1)
Note: Data outside parenthesis = Low LMX group. Data inside parenthesis = High LMX
group. D_Justice = distributive justice. P_Justice = procedural justice. I_Justice =
interactional justice. LMX = leader–member exchange. CWB = counterproductive work
behaviors. OCB = organizational citizenship behaviors.
The information gathered from these analyses (i.e., low model fit, peculiar correlations,
insignificant mediation effects) made it necessary to replicate the study with a larger
sample size.
Study 2
Method
Participants
There were 3293 subjects in the study, 40% males and 60% females between the ages of:
18-25 (53.6%), 26-35 (23.2%), 36-45 (12.3%), 46-55 (9.1%), 56-65 (1.6%), and 65+
(0.2%). The participants had either completed high school education (31.2%), tertiary/post-
secondary studies (7.7%), holding/studying for a bachelor’s degree (41.4%),
holding/studying for a master’s degree (19.4%), or holding/studying for a PhD (0.3%).
Regarding their work, most were in a managerial position (83.4%), which included: head of
office/team (15.7%), head of department (6.8%), or director/executive manager (3.4%)
while the rest in this managerial group (74.2%) were not responsible for the work of other
people. The tenure ranged between: 0–5 years (66.1%), 6-10 (14.5%), 11-15 (7.5%), 16-20
(4.6%), 21-25 (2.8%), and 25+ (4.4%).
Measures
All the measures in Study 1 were replicated in this study. Descriptive statistics are included
in Table 4.
AE Are Positive and Negative Outcomes of Organizational Justice Conditioned by Leader–Member Exchange?
250 Amfiteatru Economic
Procedure
Replication of the procedure employed in Study 1.
Results and Discussion
Common-method bias (CMB). To evaluate the extent to which variable intercorrelations might
be an artifact of common method variance (CMV), as suggested by Podsakoff et al. (2003), we
utilized two methods: (a) Harman’s single-factor model (where all items are loaded into one
common/marker factor); and (b) a common latent factor (CLF) model (where all items are
loaded into both their expected factors and one latent common method factor).
Harman’s single-factor method accounted for only 25.49% of the explained variance:
2(3,070) = 9,433.57, p = .000, 2/df = 3.07, CFI = .67, NFI = .66, GFI = .31, SRMR = .15,
RMSEA (90% CI) = .24 (.17-.29), p-close = .000. Further, the CLF method of analysis
produced 23.17% of the explained variance: 2(2,990) = 7,115.33, p = .000, 2/df = 2.38,
CFI = .70, NFI = .69, GFI = .47, SRMR = .12, RMSEA (90% CI) = .14 (.05-.21), p-close =
.000. As with the pilot study, while these results do not entirely exclude the possibility of
same-source bias (i.e., CMV), according to Podsakoff et al. (2003) less than 50% (R2 <
0.50) of the explained variance accounted for by the first emerging factor indicates that
CMB is an improbable explanation of our findings, in conjunction with the bad model fit
for each analysis.
Table 4 displays the between-variable bivariate zero-order correlational relationships in the
research.
In order to test the model, an SEM with multiple-group analysis was employed using the
IBM AMOS (v. 23) software package. The model boasted fit in the absolute sense: 2(df) =
22.34(11), p = .022, 2/df = 2.03, SRMR = .02, GFI = .99, CFI = .98, NFI = .98, NNFI =
.97, RMSEA (90% CI) = .05 (.04-.06), p-close = .478. Table 5 displays the results of the
path analysis, while LMX is a moderator (“Low LMX” = data below or equal to the LMX’s
median, while “High LMX” = data above the LMX’s median), and Z-tests in order to
discern whether the differences in estimators between the two LMX groups are statistically
significant. Table 6 depicts the indirect effects analysis for the mediation effects, as per the
hypotheses. Figure 3 portrays the results in Table 5 on a path diagram.
Table no. 4. Pearson correlation matrix
1 2 3 4 5 6 7 M SD
1. D_Justice (.83) 4.40 0.93
2. P_Justice .84 (.88) 4.43 0.97
3. I_Justice .87 .88 (.89) 4.27 0.90
4. Motivation .53 .56 .54 (.91) 4.04 0.83
5. LMX .55 .53 .58 .31 (.85) 4.12 0.91
6. CWB –.28 –.27 –.23 –.15 –.12 (.95) 2.10 0.98
7. OCB .34 .33 .35 .27 .33 –.15 (.83) 3.72 0.77
Note: All the correlations are significant at p < .001. Data in bold and parentheses are the
reliability coefficients (Cronbach’s alphas). D_Justice = distributive justice; P_Justice =
procedural justice; I_Justice = interactional justice; LMX = leader–member exchange;
CWB = counterproductive work behavior; OCB = organizational citizenship behavior.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 251
Table no. 5. SEM path results with standardized regression
coefficients and difference tests
Low LMX High LMX Difference test
Path β Sig. β Sig. Z-score
D_Justice Motivation .17 .000 .13 .001 –0.33
P_Justice Motivation .28 .000 .31 .000 1.27
I_Justice Motivation .12 .007 .12 .015 0.13
Motivation CWB –.03 .311 .03 .206 1.61
Motivation OCB .13 .000 .09 .000 –0.83
D_Justice CWB –.22 .000 –.24 .000 –0.68
D_Justice OCB .14 .002 .06 .193 –0.96
P_Justice CWB –.22 .000 –.23 .000 –0.69
P_Justice OCB .05 .311 –.04 .479 –1.20
I_Justice CWB .22 .000 –.16 .003 –2.49**
I_Justice OCB .05 .321 .19 .000 2.17*
Note: *p < .05, **p < .01, ***p < .001. D_Justice = distributive justice; P_Justice =
procedural justice; I_Justice = interactional justice; LMX = leader–member exchange;
CWB = counterproductive work behavior; OCB = organizational citizenship behavior.
Table no. 6. Mediation (indirect) effects analyses
Low LMX High LMX
Paths LL UL Sig. LL UL Sig.
D_Justice Motivation OCB 0.00 0.04 .005 0.00 0.03 .010
D_Justice Motivation CWB –0.02 0.00 .234 0.00 0.02 .160
P_Justice Motivation OCB 0.02 0.06 .000 0.01 0.05 .002
P_Justice Motivation CWB –0.03 0.01 .310 –0.01 0.03 .207
I_Justice Motivation OCB 0.01 0.04 .000 0.00 0.03 .002
I_Justice Motivation CWB –0.02 0.01 .293 0.00 0.02 .145
Note: Analyses used bootstrapping (95% bias-corrected, 5000 resamples). LL = lower limit
of the CI; UL = upper limit of the CI; D_Justice = distributive justice; P_Justice =
procedural justice; I_Justice = interactional justice; LMX = leader–member exchange;
CWB = counterproductive work behavior; OCB = organizational citizenship behavior.
Figure no. 3. Path diagram with SEM results (Study 2)
AE Are Positive and Negative Outcomes of Organizational Justice Conditioned by Leader–Member Exchange?
252 Amfiteatru Economic
Note: Data outside parenthesis = Low LMX group. Data inside parenthesis = High LMX
group. D_Justice = distributive justice; P_Justice = procedural justice; I_Justice =
interactional justice; LMX = leader–member exchange; CWB = counterproductive work
behavior; OCB = organizational citizenship behavior.
As can be seen in Table 5, taking into account the group comparison (Low LMX vs. High
LMX) there is only one statistically significant difference in the correlational relationships
between the variables. This indicates that LMX is not actually a moderator, as was
previously conceived and in total contrast to Study 1’s findings. Noteworthy, the
counterintuitive associations in Study 1 (e.g., negative links between distributive justice and
OCB, or a positive link between procedural justice and CWB) are now rendered logical,
save one (positive link between interactional justice and CWB, only in the Low LMX
group).
In addition, Table 6 has shown that motivation is indeed a mediator, but only between
distributive, procedural, and interactional justice perceptions and OCB (for both LMX
groups). No mediation effect was found when considering CWB as the criterion.
To conclude the findings, Table 7 presents the summary of results from the analyses made
in both Study 1 and Study 2.
Table no. 7. Summary of results from hypotheses testing Study 1 Study 2
Hypothesis/Path
Low-LMX High-LMX Low-LMX High-
LMX
D_Justice Motivation Supported N.S. Supported Supported
P_Justice Motivation N.S. Supported Supported Supported
I_Justice Motivation N.S. Supported Supported Supported
D_Justice OCB Supported N.S. Supported N.S.
P_Justice OCB N.S. N.S. Supported N.S.
I_Justice OCB N.S. Supported Supported Supported
D_Justice CWB N.S. N.S. Supported Supported
P_Justice CWB N.S. Supported Supported Supported
I_Justice CWB N.S. N.S. Supported Supported
D_Justice Motivation OCB N.S. Supported Supported Supported
P_Justice Motivation OCB N.S. Supported Supported Supported
I_Justice Motivation OCB N.S. N.S. Supported Supported
D_Justice Motivation CWB N.S. N.S. N.S. N.S.
P_Justice Motivation CWB N.S. N.S. N.S. N.S.
I_Justice Motivation CWB N.S. N.S. N.S. N.S.
LMX = Moderator Supported N.S.
Note: N.S. = not-supported; D_Justice = distributive justice; P_Justice = procedural justice;
I_Justice = interactional justice; CWB = counterproductive work behavior; OCB =
organizational citizenship behavior; LMX = leader–member exchange.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 253
Conclusions
The aim of the current paper was to elucidate (1) the relationship between organizational
justice (as reflected by its three dimensions: distributive, procedural, and interactional) and
positive (i.e., OCB) and negative (i.e., CWB) outcomes; (2) the mediational mechanism of
work motivation in said association; and (3) the moderation effect of LMX on the whole
research model (as outlined in Figure 1). In order to do so, we employed a pilot study (i.e.,
Study 1) and a follow-up study (i.e., Study 2) with a significantly larger sample size.
The results show distinct differences between the two samples (Study 1 vs. Study 2). In the
pilot study, most of our hypotheses were corroborated: (1) organizational justice
(distributive, procedural, interactional) negatively correlates with CWB, and positively with
OCB (H1 and H2); (2) work motivation mediated only two of these relationships
(procedural/distributive justice motivation OCB) (H4); and (3) the LMX level, as a
moderator, seemed to be a conditional factor on the overall model (H5). However, in our
larger sample (Study 2) we revealed a better and more sensible correlative constellation
among the variables. The mediation of work motivation (H3 and H4) has been bolstered,
and the moderation of LMX (H5) has been completely rejected.
There are some implications to our research. First, we can learn from the differences
between the two studies. Employing a pilot before conducting the full-scale research may
be useful, to a certain extent. As in the current research, a pilot portrays the network of
relationships, enabling us to view the general picture. However, evidently, the model
resulted somewhat differently in Study 2, with the larger sample size. As Study 2 is more
representative of the population, one may assume the associations found in it better
resemble reality. For example, LMX was found to be a moderator in Study 1, but not at all
in Study 2. Also, mediation analyses produced better findings in Study 2. This connects to
the central limit theorem, which estimates that a larger the sample size in a given set (i.e., n
∞) approximates to a normal distribution (e.g., Rosenblatt, 1956). We, therefore,
recommend the use of larger sample sizes as humanly possible, especially in cross-sectional
research such as this.
These results manifested in a certain cultural context (i.e., Romania), and might not be
relevant to other cultures and/or places. As such, we suggest replicating the study in other
countries, similar to or different from Romanian cultural values to extend the external
validity of the research. Replications have been employed successfully in well-established
sciences such as physics, chemistry and biology to ensure veracity of findings. Therefore,
we recommend that this approach be applied to behavioral sciences as well. Thus,
replication of investigations such as the current one should be embarked on despite the
support our hypotheses in this study have gained (Tziner, in press).
We can see that, finally, LMX did not moderate any of the relationships in the model, as
hypothesized; therefore, the exchanges between managers and their subordinates do not act
as a conditional factor. Thus, (1) there may be an untapped cognitive process of attribution
that should be explored in the future (perhaps the employees do not attribute lack of
fairness to their immediate manager); and (2) future studies should consider other potential
moderators such as ethical climate in the workplace, the size of the
organization/department, Big Five personality factors, and job autonomy.
In addition to these, the associations discovered in our research may be useful for
organizations. We recommend that organizations create a just and fair work environment as
AE Are Positive and Negative Outcomes of Organizational Justice Conditioned by Leader–Member Exchange?
254 Amfiteatru Economic
work environment can lead to increased work motivation and organizational citizenship
behavior (when perceived as positive), or counterproductive work behavior (when
perceived as negative). We also recommend monitoring the motivation of the employees, as
this acts as a partial mediational mechanism to OCB (i.e., organizational justice
motivation OCB), and as such, increasing it may result in increased OCB.
Using a self-report CWB questionnaire comprising items of a “judgmental” nature about
the employee’s conduct at work might have impacted the results, as the effects on CWB as
an outcome are weak or non-significant, as opposed to OCB. The questionnaire may be
perceived as a “critical voice,” thus making it difficult for the respondent to report his or
her own negative behaviors (as well as towards others). Items such as “I have taken
property from work without permission” or “I have fabricated a receipt in order to get
remuneration for work expenses” may be difficult to answer honestly. Individuals may find
it difficult to admit to behaviors such as disparagement of others or theft, even to
themselves, and under anonymity. Chernyak-Hai and Tziner’s (2014) study, which revealed
a low average CWB (M = 2.64 on a scale of 1-6), similar to our results (M = 2.50 and M =
2.10 on a scale of 1-6, for Study 1 and Study 2, respectively), provides empiric support.
Furthermore, the use of a self-reported measure of CWB might be considered a limitation
of this analysis. This deficiency might be remedied by supplementing self-reporting
measures with other-reported measures (e.g., by supervisors and co-workers) of CWB, the
latter being thought of as comparatively objective. Nevertheless, as CWB is difficult to
observe, the inter-rater reliability of other-reported measures of CWB is typically low
(Berry, Carpenter and Barratt, 2012).
In our model, we did not consider any proper individual differences such as emotional
intelligence or the Big Five personality factors as predictors, as organizational justice
perceptions may be considered as an attitudinal individual difference. We might have been
able to draw deeper conclusions from doing so (e.g., Staw and Cohen-Charash, 2005).
References
Adams, J.S., 1965. Inequity in social exchange. Advances in Experimental Social
Psychology, 2, pp.267-299. https://doi.org/10.1016/S0065-2601(08)60108-2.
Al-A’wasa, S.I.S., 2018. The impact of organizational justice on the counterproductive
work behavior (CWB): a field study conducted in the Jordan Customs Department
(JCD). International Journal of Business and Social Science, 9(1), pp.27-38.
Ambrose, M.L. and Schminke, M., 2009. The role of overall justice judgments in
organizational justice research: a test of mediation. Journal of Applied Psychology,
94(2), pp.491-500. DOI: 10.1037/a0013203.
Aubé, C., Rousseau, V., Mama, C. and Morin, E.M., 2009. Counterproductive behaviors
and psychological well-being: the moderating effect of task interdependence. Journal of
Business and Psychology, 24(3), pp.351-361. DOI: 10.1007/s10869-009-9113-5.
Bennett, R.J., and Robinson, S.L., 2000. Development of a measure of workplace deviance.
Journal of Applied Psychology, 85(3), pp.349-360. https://doi.org/10.1037/0021-
9010.85.3.349.
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 255
Bennett, R.J. and Robinson, S.L., 2003. The past, present, and future of workplace deviance
research. In: J. Greenberg, ed., 2003. Organizational Behavior: The State of the Science.
Mahwah, NJ: Lawrence Erlbaum Associates. pp.247-281.
Bernerth, J. and Walker, H.J., 2012. Reexamining the workplace justice to outcome
relationship: does frame of reference matter? Journal of Management Studies, 49(5),
pp.945-969. https://doi.org/10.1111/j.1467-6486.2010.00977.x.
Berry, C.M., Carpenter, N.C. and Barratt, C.L., 2012. Do other-reports of
counterproductive work behavior provide an incremental contribution over self-reports?
A meta-analytic comparison. Journal of Applied Psychology, 97(3), pp.613-636. doi:
10.1037/a0026739.
Blau, P.M., 1964. Exchange and Power in Social Life. New York: John Wiley.
Bodankin, M. and Tziner, A., 2009. Constructive deviance, destructive deviance and
personality: how do they interrelate? Amfiteatru Economic, 11(26), pp.549-564.
Brienza, J.P. and Bobocel, D.R., 2017. Employee age alters the effects of justice on
emotional exhaustion and organizational deviance. Frontiers in Psychology, 8, Article
no. 479. https://doi.org/10.3389/fpsyg.2017.00479.
Brislin, R.W., 1980. Translation and content analysis of oral and written material. In: H.C.
Triandis and J.W. Berry, eds. Handbook of Cross-Cultural Psychology. Boston, MA:
Allyn and Bacon. pp.389-444.
Buzea, C., 2014. Equity theory constructs in a Romanian cultural context. Human Resource
Development Quarterly, 25, pp.421-439. https://doi.org/10.1002/hrdq.21184.
Chernyak-Hai, L. and Tziner, A., 2014. Relationships between counterproductive work
behavior, perceived justice and climate, occupational status, and leader–member
exchange. Journal of Work and Organizational Psychology - Revista de Psicologia del
Trabajo y de Las Organizaciones, 30(1), pp.1-12. https://doi.org/10.5093/tr2014a1.
Cohen-Charash, Y. and Mueller, J.S., 2007. Does perceived unfairness exacerbate or
mitigate interpersonal counterproductive work behaviors related to envy? Journal of
Applied Psychology, 92(3), pp.666-680. DOI: 10.1037/0021-9010.92.3.666.
Cohen-Charash, Y., and Spector, P.E., 2001. The role of justice in organizations: a meta-
analysis. Organizational Behavior and Human Decision Processes, 86(2), pp.278-321.
DOI: 10.1006/obhd.2001.2958.
Cole, M.S., Schaninger Jr, W.S. and Harris, S.G., 2002. The workplace social exchange
network: a multilevel, conceptual examination. Group and Organization Management,
27, pp.142-167. https://doi.org/10.1177/1059601102027001008.
Colquitt, J.A., Conlon, D.E., Wesson, M.J., Porter, C.O. and Ng, K.Y., 2001. Justice at the
millennium: a meta-analytic review of 25 years of organizational justice research.
Journal of Applied Psychology, 86(3), pp.425-445. DOI: 10.1037//0021-9010.86.3.425.
Delery, J.E. and Doty, D.H., 1996. Modes of theorizing in strategic human resource
management: tests of universalistic, contingency, and configurational performance
predictions. Academy of Management Journal, 39(4), pp.802-835.
https://doi.org/10.5465/256713.
Dilchert, S., Ones, D.S., Davis, R.D. and Rostow, C.D., 2007. Cognitive ability predicts
objectively measured counterproductive work behaviors. Journal of Applied
Psychology, 92(3), pp.616-627. https://doi.org/10.1037/0021-9010.92.3.616.
AE Are Positive and Negative Outcomes of Organizational Justice Conditioned by Leader–Member Exchange?
256 Amfiteatru Economic
Eskew, D.E., 1993. The role of organizational justice in organizational citizenship
behavior. Employee Responsibilities and Rights Journal, 6, pp.185-194.
DOI:10.1007/BF01419443.
Faragher, E.B., Cass, M. and Cooper, C.L., 2013. The relationship between job satisfaction
and health: a meta-analysis. In: Cooper C.L. (eds) From Stress to Wellbeing, Vol. 1.
London: Palgrave Macmillan, pp.254-271. https://doi.org/10.1057/9781137310651_12.
Gouldner, A.W., 1960. The norm of reciprocity: a preliminary statement. American
Sociological Review, 25(2), pp.161-178. https://doi.org/10.2307/2092623.
Graen, G.B. and Uhl-Bien, M., 1995. Relationship-based approach to leadership:
development of leader–member exchange (LMX) theory of leadership over 25 years:
applying a multi-level multi-domain perspective. Leadership Quarterly, 6(2),
pp.219-247. https://doi.org/10.1016/1048-9843(95)90036-5.
Greenberg, J. and Scott, K.S., 1996. Why do workers bite the hand that feeds them?
Employee theft as a social exchange process. In: B.M. Staw and L.L. Cummings, eds.
1996. Research in organizational behavior: An annual series of analytical essays and
critical reviews, Vol. 18. London: Elsevier Science/JAI Press. pp.111-156.
Hobfoll, S.E., 1989. Conservation of resources: a new attempt at conceptualizing
stress. American Psychologist, 44(3), pp.513-524. https://doi.org/10.1037/0003-
066X.44.3.513.
Hollensbe, E.C., Khazanchi, S. and Masterson, S.S., 2008. How do I assess if my
supervisor and organization are fair? Identifying the rules underlying entity-based
justice perceptions. Academy of Management Journal, 51(6), pp.1099-1116.
https://doi.org/10.5465/amj.2008.35732600.
Huang, X., Chan, S.C., Lam, W. and Nan, X., 2010. The joint effect of leader–member
exchange and emotional intelligence on burnout and work performance in call centers in
China. The International Journal of Human Resource Management, 21(7), pp.1124-
1144. https://doi.org/10.1080/09585191003783553.
Ilies, R., Nahrgang, J.D. and Morgeson, F.D., 2007. Leader–member exchange and
citizenship behaviors: a meta-analysis. Journal of Applied Psychology, 92(1),
pp.269-277. https://doi.org/10.1037/0021-9010.92.1.269
Kanfer, R., Frese, M. and Johnson, R.E., 2017. Motivation related to work: a century of
progress. Journal of Applied Psychology, 102(3), pp.338-355. DOI:
10.1037/apl0000133
Karriker, J.H. and Williams, M.L., 2009. Organizational justice and organizational
citizenship behavior: a mediated multifoci model. Journal of Management, 35(1),
pp.112-135. DOI:10.1177/0149206307309265
Latham, G.P. and Pinder, C.C., 2005. Work motivation theory and research at the dawn of
the twenty-first century. Annual Review of Psychology, 56(1), pp.485-516. DOI:
10.1146/annurev.psych.55.090902.142105
Liden, R.C. and Maslyn, J.M., 1998. Multidimensionality of leader–member exchange: an
empirical assessment through scale development. Journal of Management, 24(1),
pp.43-72. https://doi.org/10.1016/S0149-2063(99)80053-1
Maslyn, J.M., Schyns, B. and Farmer, S.M., 2017. Attachment style and leader–member
exchange: the role of effort to build high quality relationships. Leadership and
Economic Interferences AE
Vol. 23 • No. 56 • February 2021 257
Organization Development Journal, 38(3), pp.450-462. https://doi.org/10.1108/LODJ-
01-2016-0023
Mayrhofer, W., Brewster, C. and Morley, M., 2000. The concept of strategic European
human resource management. In: Brewster C., Mayrhofer W., Morley M., eds. 2000.
New Challenges for European Human Resource Management. London: Palgrave
Macmillan, pp.3-33.
Niehoff, B.P. and Moorman, R.H., 1993. Justice as a mediator of the relationship between
methods of monitoring and organizational citizenship behavior. Academy of
Management Journal, 36(3), pp.527-556. https://doi.org/10.5465/256591
Organ, D.W., Podsakoff, Ph.M. and MacKenzie, S.B., 2006. Organizational Citizenship
Behavior: Its Nature, Antecedents, and Consequences. Thousand Oaks, CA: Sage
Publications.
Pinder, C.C., 2014. Work Motivation in Organizational Behavior, 2nd edition. New York:
Psychology Press.
Podsakoff, P.M., MacKenzie, S.B., Lee, J.Y. and Podsakoff, N.P., 2003. Common method
biases in behavioral research: a critical review of the literature and recommended
remedies. Journal of Applied Psychology, 88(5), pp.879-903. DOI: 10.1037/0021-
9010.88.5.879
Podsakoff, N.P., Whiting, S.W., Podsakoff, P.M. and Blume, B.D., 2009. Individual- and
organizational-level consequences of organizational citizenship behaviors: a meta-
analysis. Journal of Applied Psychology, 94(1), pp.122-141. DOI: 10.1037/a0013079
Rosenblatt, M., 1956. A central limit theorem and a strong mixing condition. Proceedings
of the National Academy of Sciences (PNAS) of the United States of America, 42(1),
pp.43-47. doi: 10.1073/pnas.42.1.43
Shkoler, O. and Tziner, A., 2017. The mediating and moderating role of burnout and
emotional intelligence in the relationship between organizational justice and work
misbehavior. Journal of Work and Organizational Psychology - Revista de Psicologia
del Trabajo y de las Organizaciones, 33(2), pp.157-164. https://doi.org/10.1016/
j.rpto.2017.05.002
Skarlicki, D.P. and Folger, R., 1997. Retaliation in the workplace: the roles of distributive,
procedural, and interactional justice. Journal of Applied Psychology, 82(3), pp.434-443.
https://doi.org/10.1037/0021-9010.82.3.434
Spector, P.E., Fox, S., Penney, L.M., Bruursema, K., Goh, A. and Kessler, S., 2006. The
dimensionality of counterproductivity: are all counterproductive behaviors created
equal? Journal of Vocational Behavior, 68(3), pp.446-460. https://doi.org/
10.1016/j.jvb.2005.10.005
Staw, B.M. and Cohen‐ Charash, Y., 2005. The dispositional approach to job satisfaction:
more than a mirage, but not yet an oasis. Journal of Organizational Behavior, 26(1),
pp.59-78. DOI: 10.1002/job.299
Tremblay, M.A., Blanchard, C.M., Taylor, S., Pelletier, L.G. and Villeneuve, M., 2009.
Work extrinsic and intrinsic motivation scale: Its value for organizational psychology
research. Canadian Journal of Behavioural Science (Revue canadienne des sciences du
comportement), 41(1), pp.213-226. https://doi.org/10.1037/a0018176
AE Are Positive and Negative Outcomes of Organizational Justice Conditioned by Leader–Member Exchange?
258 Amfiteatru Economic
Tziner, A., in press. Facet methodology and analysis: mining the unconquered land in
behavioral sciences research. In: P.M.W. Hackett, ed. Mereologies, Ontologies and
Facets: The Categorical Structure of Reality. New York: Lexington. pp.161-200.
Tziner, A., Fein, E. and Oren, L., 2012. Human motivation and its outcomes. In: C.L.
Cooper and J.C. Quick, eds. 2012. Downsizing: Is Less Still More?. New York:
Cambridge University Press. pp.103-133.
Ugaddan, R.G. and Park, S.M., 2019. Do trustful leadership, organizational justice, and
motivation influence whistle-blowing intention? Evidence from federal
employees. Public Personnel Management, 48(1), pp.56-81. https://doi.org/10.1177/
0091026018783009
Vardi, Y. and Weitz, E., 2002. Using the theory of reasoned action to predict organizational
misbehavior. Psychological Reports, 91(3), pp.1027-1040. https://doi.org/10.2466/
pr0.2002.91.3f.1027
Wang, X., Liao, J., Xia, D. and Chang, T., 2010. The impact of organizational justice on
work performance: mediating effects of organizational commitment and leader–member
exchange. International Journal of Manpower, 31(6), pp.660-677.
Williams, L.J. and Anderson, S.E., 1991. Job satisfaction and organizational commitment
as predictors of organizational citizenship and in-role behaviors. Journal of
Management, 17, pp.601-617. http://dx.doi.org/10.1177/014920639101700305
Zagenczyk, T.J., Purvis, R.L., Shoss, M.K., Scott, K.L. and Cruz, K.S., 2015. Social
influence and leader perceptions: multiplex social network ties and similarity in leader–
member exchange. Journal of Business and Psychology, 30(1), pp.105-117.
https://doi.org/10.1007/s10869-013-9332-7
Vol. 23 • No. 56 • February 2021 259
Contents
Amfiteatru Economic recommends
Immigrant Entrepreneurship in Romania: Drawing Best Practices
From Middle Eastern Immigrant Entrepreneurs’ Experiences .................................. 260
Ozgur Ozmen, Raluca Mariana Grosu and Mariana Dragusin
The Governance Impact on the Romanian Trade Flows. An Augmented
Gravity Model................................................................................................................... 276
Anca Tamaș and Dumitru Miron
Socio-Economic and Macro-Financial Determinants and Spatial Effects
on European Private Health Insurance Markets .......................................................... 290
Gabriela Mihaela Mureșan, Cristian Mihai Dragoș, Codruța Mare,
Simona Laura Dragoș and Alexandra Pintea
Book Review: The Statistical Monograph of the Bucharest University
of Economic Studies. 100 Generations of Graduates .................................................... 308
Mihai Korka
AE Immigrant Entrepreneurship in Romania: Drawing Best Practices From Middle Eastern Immigrant Entrepreneurs’ Experiences
260 Amfiteatru Economic
IMMIGRANT ENTREPRENEURSHIP IN ROMANIA:
DRAWING BEST PRACTICES FROM MIDDLE EASTERN IMMIGRANT
ENTREPRENEURS’ EXPERIENCES
Ozgur Ozmen1, Raluca Mariana Grosu2 and Mariana Dragusin3 1)Nevşehir HBV University, Nevşehir, Turkey
2)3)Bucharest University of Economic Studies, Bucharest, Romania
Please cite this article as:
Ozmen, O., Grosu, R.M. and Dragusin, M., 2021.
Immigrant Entrepreneurship in Romania: Drawing
Best Practices From Middle Eastern Immigrant
Entrepreneurs’ Experiences. Amfiteatru Economic,
23(56), pp. 260-275.
DOI: 10.24818/EA/2021/56/260
Article History
Received: 15 September 2020
Revised: 30 October 2020
Accepted: 26 November 2020
Abstract
Migrant entrepreneurship represents a topic of a high societal and academic significance. For
a host country, immigrant entrepreneurs’ endeavours are, in many cases, an under-utilized
lever for local and regional economic revival. Based on a complex field research carried out
between July 2017 and December 2018, the present paper approaches immigrant
entrepreneurship in Romania, with a particular focus on Middle Eastern immigrant
entrepreneurs. Our multi-method qualitative field research – consisting of semi-structured
interviews, observations and informal discussions – envisaged 97 immigrant businesses,
targeting the analysis of the phenomenon from different perspectives: economic, social,
cultural, political, and institutional. Aiming to draw best practices from the investigated
Middle Eastern immigrant entrepreneurs’ experiences, the research results outlined in this
paper emphasize both descriptive and practical features. Migration-related particularities,
entrepreneurs’ profile, business obstacles, business practices, or perspectives on and
comparisons of the Romanian business environment with the one in the origin country
represent just few examples of aspects thoroughly approached in the paper.
Keywords: migration, entrepreneurship, immigrant entrepreneurship, immigrant
entrepreneur, Romania, Middle Eastern immigrants
JEL Classification: F22, J15, M10, M20, O15
Corresponding author, Raluca Mariana Grosu – e-mail: [email protected]
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 261
Introductory remarks with a focus on recent literature review
In an extremely interconnected world, immigrant entrepreneurship represents a topic of a
high societal and academic significance. Mainly referring to immigrants starting and
developing businesses in the host country (Basu, 2006), immigrant entrepreneurship is the
focus of an increasing number of scientific studies. Their outcomes have both important
theoretical and practical dimensions, with a real capacity to favour improved decisions
impact, at many levels.
Commonly approached also as ethnic and/or minority entrepreneurship, immigrant
entrepreneurship as a field of study is supported by a wide, comprehensive and growing
scientific literature. Strong economies in developed countries – like the United States of
America, Canada, Australia or the United Kingdom, the Netherlands, or Germany, in Europe
– that are also hosts to vast immigrant communities, are being investigated by immigrant
entrepreneurship scholars. Considering recent works in the area, distinct research paths can
be identified, such as: immigrants’ entrepreneurial intentions; profile of the immigrant
entrepreneur, their needs, motivations and potential; gender-related aspects; comparisons
with native/local entrepreneurs both in terms of individual characteristics and business
profile; importance of social networks in business development and performance;
entrepreneurial orientation and business performance; policy implications and
recommendations; impact factors on the development of immigrant entrepreneurship; etc.
(Kitching, Smallbone and Athayde, 2009; Kloosterman, 2010; Ilhan-Nas, Sahin and Cilingir,
2011; Baycan, Sahin and Nijkamp, 2012; Neville et al., 2014; Dinu, Grosu and Saseanu,
2015; Kerr and Kerr, 2016; Vinogradov and Jorgensen, 2017; Bolzani and Boari, 2018;
Rahman, 2018; De Luca and Ambrosini, 2019; Jones, Ram and Villares-Varela, 2019).
Studies on immigrant entrepreneurship in less developed countries, are not so prominent in
the scientific literature. For example, considering the case of Romania, the country
investigated in this paper, research in the area of immigrant entrepreneurship is still in an
incipient stage. There are few existing studies in the literature mainly focused on: descriptive
issues of the entrepreneurial process of immigrant entrepreneurs; industry-related aspects;
religion-related aspects; women immigrant entrepreneurship; etc. (Constantin, Goschin and
Dragusin, 2008; Grosu and Saseanu, 2014; Grosu, 2015; Sharbek and Grosu, 2018).
Romania is highly acknowledged – by both scholars and practitioners – as an important
provider of immigrants, especially for the European Union (EU) members, with a well-
represented diaspora (Grosu and Constantin, 2013; Grosu and Dinu, 2016; Davidescu et al.,
2017). Yet, this also represents an attractive destination country for immigrant population,
especially for non-EU citizens, out of which many take the entrepreneurial path in the
Romanian business environment. The highest number of newcomers in Romania are from
the Republic of Moldavia. Other significant non-EU foreign communities are from China,
Israel, or the Middle East. Arrivals of non-native Romanians from the EU countries are
mainly registered from Austria, France, Germany, Hungary and Italy. In terms of
entrepreneurship, Middle Eastern immigrant entrepreneurs, like Turkish, Jordanians, Syrians
ones etc., are key players on the Romanian market. It is estimated that, especially Turks had
a “significant contribution to the success of transition to a market economy, as well as to
economic recovery, in post-1989 Romania” (Constantin, Goschin and Dragusin, 2008, p. 51).
Even if the existing body of works on immigrant entrepreneurship is very vast and provides
various outcomes/ approaches on diverse regions and countries from all over the world,
research on this topic in Romania is still in an incipient stage. Yet, this emerging field of
AE Immigrant Entrepreneurship in Romania: Drawing Best Practices From Middle Eastern Immigrant Entrepreneurs’ Experiences
262 Amfiteatru Economic
study has great potential of development, offering scholars many research directions, worth
exploring. Considering the previously framework, through our paper we aim at bringing a
contribution, from theoretical perspective, to the development of the scarce scientific
literature regarding the topic of immigrant entrepreneurship in Romania, in general, and that
of Middle Eastern originated one, in particular. From a practical perspective, our work, has
the potential to contribute to: shaping a clearer picture of the nature and content of the
entrepreneurial processes initiated and carried out by immigrants in Romania; highlighting
their role and raising awareness of the underutilized economic leverage they currently
represent; promoting Middle Eastern immigrant entrepreneurs as inspiring role models,
worth following, especially by other members of their community; highlighting Middle
Eastern immigrant entrepreneurs’ best practices in Romania, mainly to foster the
entrepreneurial intentions of other immigrant communities.
In this sense, the present paper is structured into four main parts, introduction and conclusions
included. ‘Research methodology’ is the next part of the paper, where we put forward main
methodological aspects related to our research, closely followed by the section dedicated to
this study’s results.
1. Research methodology
This scientific work is developed based on a part of the outcomes of a multi-method
qualitative field research carried out between July 2017 - December 2018, with the aim to
provide a more complex and comprehensive image on the Middle Eastern immigrant
entrepreneurship phenomenon in Romania. Our research targeted the analysis of this
phenomenon considering relevant social, economic, political and institutional contexts,
following the “mixed embeddedness” approach (Kloosterman, van der Leun and Rath, 1999).
Our investigated sample was composed of 97 businesses owned by Middle Eastern
immigrant entrepreneurs in Romania, out of which: three big companies and 94 micro, small
and medium sized enterprises. All of them were active in the capital-city region, respectively
Bucharest-Ilfov. This is not a surprising fact as this is the most economically developed
region in Romania, being ‘host’ not only to most of the companies at national level, but also
of the immigrants, implicitly to immigrant business owners.
The lack of a clear evidence on immigrant entrepreneurs – defined as immigrants who initiate
and develop a business in their host country (Basu, 2006) – in Romania, made impossible the
access to a database with relevant information. The General Inspectorate for Immigration in
Romania registers immigrants according to their migration purpose, but this does not further
monitor if an immigrant with a clear business purpose did really start a business (Grosu,
2015). Furthermore, the national institution in charge with registering businesses,
respectively the National Trade Register Office, only considers the foreign participation to
capital in an enterprise, as a foreign-related aspect. This might refer to “enterprises started
and developed – entirely, or partly – through the participation of private individuals or
corporate entities with their stable residence or their headquarters outside Romania” (Oficiul
National al Registrului Comertului, 2019). This approach reflects concepts related more to
international entrepreneurship (e.g.: foreign investors), than to immigrant entrepreneurship.
In such a framework, amplified by the high reluctance of immigrant entrepreneurs to take
part in research, we approached the ‘snowball’ and the ‘networking’ sampling techniques to
construct our sample. This was a very challenging process, with many obstacles encountered
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 263
along the way. Even if one of the authors is known by the society of the mosques, some of
the immigrant entrepreneurs were very suspicious concerning our study. They were
convinced that our research was organized by the Romanian Secret Services, targeting the
tagging of immigrant entrepreneurs in Romania. Thus, many of those approached during the
sampling process did not want to be involved in generating the needed information, mainly
claiming reasons like: lack of time to answer our questions; limited Romanian language
knowledge; urgent need to arrive somewhere else, etc.
The main methods used to collect data from the targeted sample, were: face-to-face semi-
structured interviews, observations and informal discussions with the customers and/or
employees of the investigated businesses. (Ozmen and Grosu, 2020)
The face-to-face semi-structured interviews carried-out with the immigrant entrepreneurs
were mainly formal – being based on an interview guide – and lasted for around 40 minutes,
each. Developed following the research’s aim and objectives, the interview guide was first
debated with scholars of entrepreneurship and pilot-tested on 3 immigrant entrepreneurs,
its improved version being further applied in our research. Interviews were held in
Romanian, but, also in Turkish or Arabic languages. In the latter case, a translator took
part to the interview, too. Thus, the number of persons involved in the interview varied
from two to four, depending on the case. Usually, the interviews took place at the owner’s
enterprise headquarter but there were also situations when several interviews were carried
out in Turkish and Arabic mosques in Bucharest. Normally, interviews were not disturbed
by external factors. However, in some cases, interruptions translated in: clients entering
the place (store, restaurant, warehouse, etc.), employees entering the immigrant
entrepreneur’s office, phone calls, friends coming to approach the interviewee in the
mosque, etc. (Ozmen and Grosu, 2020)
With regard to the observation method, our role was that of complete observers. In general,
such actions were performed before having the interviews with the entrepreneurs and before
developing the informal discussions with the employees and/or customers. We mainly
targeted customers and employees to discover pertinent owner/ business related aspects. The
observation regarding the first category aimed at the buyers’ behaviour within the grocery
stores, other retail stores, restaurants, etc., depending on the specific object of activity. We
also observed the employees’ conduct, along with the entire activity of the analysed business.
The informal discussions lasted around five minutes with each individual. However, it is
worth mentioning that we did not use the methods of ‘observations’ and/or ‘informal
discussions’ in all the investigated businesses. (Ozmen and Grosu, 2020)
All our field research translated into many notes taken during the development of the above-
mentioned methods. However, information about how the whole data collection process
went, about our own perceptions, feelings and thoughts, regarding non-verbal
communication of investigated people, in each of the targeted entity were also recorded. All
these notes generated invaluable information, part of it being further outlined in the paper.
2. A comprehensive image on the Middle Eastern immigrant entrepreneurship
phenomenon in Romania - main research results
The investigated businesses were owned by Middle Eastern male entrepreneurs that
immigrated in Romania between 1978 and 2015 (with the first 11 years, as part of the
communist regime), from countries like: Turkey, Syria, Jordan, Iraq, Kuwait, and Lebanon.
AE Immigrant Entrepreneurship in Romania: Drawing Best Practices From Middle Eastern Immigrant Entrepreneurs’ Experiences
264 Amfiteatru Economic
Their migration reasons were varied, ranging from studying, the desire to find a job or to
become entrepreneur, to skipping the army enrolment, family reunification, tourism,
avoiding the war and/or other political issues faced in the origin country, etc.
“My friends were here, my brother was here, my uncle was here … So, I decided to come
too.” (Turkish entrepreneur)
“Romania is a safe and beautiful country and I first came here as a tourist. I liked it so
much as I did not want to return back home.” (Turkish entrepreneur)
“The firm I’ve been working for in Turkey had a subsidiary in Romania and this is how
I got here. After a while, the company was closed but I liked so much to be in Romania,
that I did not want to go back to Turkey. I have searched for work in different places,
but the result was not positive and, at the end, because I did not have any other choice,
I became entrepreneur. Looking back now, I do not regret anything.” (Turkish
entrepreneur)
“My brother was in Romania. I decided to come here to offer a better future to my
children.” (Arab entrepreneur)
The time span the investigated immigrant entrepreneurs started a business in Romania, seems
to be correlated too, in most of the cases, with their main reasons for migration. Most of the
inquired Middle Eastern immigrant entrepreneurs, especially Turks, started their businesses
in the same year when they have immigrated to Romania. However, there were incomers
who also followed the entrepreneurial path after one year of immigration, or even up to 12
years. These are situations encountered mainly amid Arab immigrant entrepreneurs.
Compared to Turks, that, generally, migrated to Romania with a strong business reason – no
matter the period – Arabs mainly came to Romania, especially during the communist regime,
for studying purposes. The most attractive industries for the immigrant entrepreneurs in our
sample for business start-up and scale-up proved to be: commerce, production, construction,
real estate, and HORECA. Both forms of opportunity and necessity-driven entrepreneurship
were specified by the investigated Middle Eastern immigrants, the main mentioned “push”
factor being the impossibility to find a job in the host country.
“The main reason I became entrepreneur was the fact that I did not find a job.” (Arab
entrepreneur)
With a dominant high level of education declared (university, masters and, in some cases,
doctorate), the majority of responding entrepreneurs stated that education is a crucial provider
of basic economic knowledge, paramount to their competences’ and business skills’
development. It is interesting to note that they’ve also appreciated the opportunity to learn
foreign languages, especially English, considered a “must” to running their businesses.
Furthermore, the respondents perceived that one of the main positive outcome of following
university studies was the chance to expanding solid social networks. However, we’ve
registered opposite opinions too, according to which entrepreneurship is not directly
connected to education. Such assertions were mainly supported by interviewees who dropped
out of school, or managed to pass only some of the primary education stages. From their
point of view, one cannot learn entrepreneurship in school. For them, “entrepreneurship runs
through their veins”. It is something “they were just born with”. They are consistent with the
idea that entrepreneurship is more a native skill, which can be developed throughout life,
while performing various activities and being around worth following role models. In many
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 265
cases, close relatives, mainly males have inspired respondents’ decisions to take the
entrepreneurial path. Often, fathers were highly acknowledged as successful role models.
Other representative figures with significant impact on their career, were their grandfathers,
uncles and/or their brothers. In general, the masculine figures in their families seems to have
put a strong mark on their evolution as entrepreneurs. Relevant samples of such previously
presented opinions are provided below.
“I have followed university studies. […] I have learned English. This helps me a lot
while doing business.” (Turkish entrepreneur)
“The things learned during the university helped me develop my business’ vision.”
(Turkish entrepreneur)
“I have a university degree. […] I consider that this didn’t help me in my entrepreneurial
career. All I know is from my father. My father taught me business. University only
helped me to create additional social networks.” (Turkish entrepreneur)
“My father determined me to become an entrepreneur. His influence was higher than
the one education had on me.” (Turkish entrepreneur)
“I always wanted to be an entrepreneur. This is my character.” (Turkish entrepreneur)
“This was my dream since I was a child. […] And I did it for being free.” (Turkish
entrepreneur)
“The university helped me to learn Romanian and about the Romanian culture,
traditions, etc. It was my main achievement to become an entrepreneur in this country.”
(Arab entrepreneur)
“I have only graduated high school. This did not help me at all. My father was the only
one who inspired me in doing business. Ever since I was 7, I started to work with him in
his business. I have learnt from him how to do business. This is all I can do ... business.
I don’t know anything else.” (Arab entrepreneur)
“Entrepreneurship is in my DNA, in my blood. It comes from my family.” (Arab
entrepreneur)
The majority of interviewees were not at their first entrepreneurial attempt. Other businesses
– many of them in the same branch – were previously developed by the investigated Middle
Eastern immigrant entrepreneurs. Furthermore, some of the respondents, before becoming
entrepreneurs, were employees – in many of the cases, in the same area in which they’ve
started their companies, either in Romania, or in their origin country. The way in which they
talked about their businesses, their previous entrepreneurial endeavours and/or work
experience, revealed strong dedication, passion and willingness to acquire knowledge, skills,
experience and expertise in the area in which they run the business.
“Before starting my businesses, I knew everything about this market; I was hired in
this branch before.” (Turkish entrepreneur)
Generally, Turks have a business partner, of the same nationality, but there were also cases
of Turkish immigrant entrepreneurs with Romanian shareholders. On the other hand, Arabs,
usually, started their businesses by themselves. However, there were examples of multiple
shareholding in the analysed Arab immigrant businesses, too. Somehow contradictory to
traditional cultural beliefs, the partner is, in multiple cases, the entrepreneur’s wife, often of
AE Immigrant Entrepreneurship in Romania: Drawing Best Practices From Middle Eastern Immigrant Entrepreneurs’ Experiences
266 Amfiteatru Economic
Romanian nationality. Such a choice might be probably based on her Romanian language
knowledge, and on the fact that she might be more accustomed with the host country’s
culture, fiscal system, and/or economic environment. As stated during the interviews, the
lack of knowledge about many of the mentioned aspects represented important obstacles, for
many immigrant entrepreneurs, Turks included. But, in most of the cases, these barriers were
overcome with the help of legal and/or fiscal advisors.
“When I was in the process of starting my business, I did not know how to properly
manage this, especially from a bureaucratic point of view. Therefore, I have hired a
legal advisor.” (Turkish entrepreneur)
“I have faced many obstacles because I did not know Romanian. I wasn’t able to
understand anything. But it all ended when I met my wife. She was my lifeline! She helped
me a lot with my business and she became my business partner, as I wanted to show my
gratitude to her.” (Arab entrepreneur)
Other common challenges highlighted by the interviewed Middle Eastern immigrant
entrepreneurs during their entrepreneurial career in Romania referred to: poor access to
financial resources, human resource-related issues, bureaucracy, ambiguous institutional and
legislative frameworks, corruption, difficulties in obtaining visas, and, even, discrimination.
“There is a lot of bureaucracy in Romania. One has to deal with different mentalities,
and there is a low quality in many employees. Furthermore, when one starts a business,
it is difficult to get any support and to have access to proper financing”. (Turkish
entrepreneur)
“I had employees that were stealing from my business, were rude, and not very serious.”
(Turkish entrepreneur)
“I sometimes feel that we, as Arab entrepreneurs, do not benefit of similar ‘rights’ as
Romanians do. If you are in a place (for example, an office) where there are also
Romanians and you need something, Romanians have priority.” (Arab entrepreneur)
“[…] I consider that there are too many controls performed by the Romanian authorities
regarding my business.” (Arab entrepreneur)
“I had a lot of problems before finding the right, trustworthy employees.” (Arab
entrepreneur)
According to collected answers, no matter the structure of the ownership, all the investigated
Middle Eastern immigrant entrepreneurs seem to have generated new jobs, mainly occupied
by Romanian employees. Only few of them declared having employees of other nationalities,
like: Turks and Arabs, but also Bulgarians, Hungarians, Macedonian, or Moldavians.
Furthermore, somehow, totally in contradiction with other scientific studies, most of our
respondents did not place their businesses within their community, nor in communities with
a high concentration of immigrants. Their choice was rather to set them in neighbourhoods
that are not recognized for their immigrant character, but inhabited by Romanians. This
translates into real economic contributions to those areas. So, the offer of products and/ or
services of such an entrepreneur is usually not aiming first his own community of immigrants
but the potential segments of Romanian customers. These are usually informed decisions,
often based on professional feasibility studies and market research, providing relevant
conclusions about the economic value of the business. However, as stated by those who
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 267
started their businesses immediately after the fall of the communist regime, no studies were
carried out back then. The Romanian market was “so free of competition and so full of
customers willing to buy anything”, that the development of such studies was not justified.
Even more, almost all business started in that period were success granted.
“My business is addressing Romanians. I came here for them. They represent my market
and I have previously investigated their needs and expectations for my products.”
(Turkish entrepreneur)
“I address to everyone interested in my services. I do not have any target market,
considering my clients’ diverse nationality. From this perspective, I can say I address to
mass market.” (Turkish entrepreneur)
“I target both Romanians and Turks.” (Turkish entrepreneur)
“Before starting my business, I knew the prices and the quality of the products my
competitors sold.” (Turkish entrepreneur)
“I have benefited from a feasibility study before starting the business.” (Turkish
entrepreneur)
“In my branch (bakery) there are a lot of Turkish entrepreneurs. I knew about them and
about their background in the bakery industry in Turkey.” (Turkish entrepreneur)
“I target both Romanians and Arabs.” (Arab entrepreneur)
“Before starting my business, I knew a lot about my market. […] And I still know. I’m
always researching about my competitors.” (Arab entrepreneur)
“Before starting my business, I have studied the market. I have identified Romanians’
needs and I have developed a business to satisfy their needs. I brought products for
satisfying Romanians’ needs.” (Arab entrepreneur)
“When I started my business in 1991, I had exclusivity in my branch! What a great time!
[…] No competition, many customers willing to buy all my products!” (Arab
entrepreneur)
Another research finding from our sample is that immigrant entrepreneurs and the ethnic
communities they are members of are adding to the cultural diversity through their specific
customs and traditions, thus positively influencing the region’s socio-cultural richness. They
also often propose new products/services, many related to their national specificity, adding
to a more diversified supply on the market. Almost all the investigated entrepreneurs stated
that they developed their businesses in Romania based on the novelty principle. This
translated not only in introducing new products/services but also in providing innovative
production and selling technique and new approaches in business model design.
“I have a very well-developed logistic system and I offer to my clients integrated
solutions” (Turkish entrepreneur)
“I have introduced new, trendy models in clothing.” (Turkish entrepreneur)
“We always develop new products, of high quality, and our prices are very affordable.”
(Arab entrepreneur)
AE Immigrant Entrepreneurship in Romania: Drawing Best Practices From Middle Eastern Immigrant Entrepreneurs’ Experiences
268 Amfiteatru Economic
In order to strengthen their effectiveness, some respondents admitted to having searched for
additional support, other than the regularly available one for all Romanian entrepreneurs.
Aware of potential ethnic-related hindrances some immigrant entrepreneurs have joined
specific business associations. This is mainly the case of the Turkish entrepreneurs who are
part of TIAD (Turkish Businessmen Association in Romania), the most representative
business association of foreign entrepreneurs in Romania. Forms of support, especially moral
and financial, was also received from families, friends and/or business partners. Furthermore,
correlated somehow with their religion and culture, very often, Divinity was mentioned as
the main supporter by many respondents. In terms of financial aid, most of the investigated
Middle Eastern immigrant entrepreneurs started their business mainly on their own savings,
raised especially from their origin countries.
The immigrant interviewees’ perception on the ease of living and doing business in Romania
is divided, probably mainly due to national culture’s specificity, language and writing
included. A number of respondents in our sample signalled some feelings of frustration. For
example, some Syrian immigrant entrepreneurs expressed an intention to return to their origin
country. Though they’ve admitted that this would be impossible, especially because of the
threatening war. Even if Syria was perceived by most of the investigated Syrian immigrant
entrepreneurs as having a friendlier business environment compared to Romania, with better
fiscal policy, lower taxation schemes and easier accounting policy, the on-going war in their
country still keeps them in Romania.
As expected, in the case of Arab immigrant entrepreneurs’ perception, the Romanian
business environment has significant and often problematic differences in comparison with
their native country. However, as revealed by many, those having a Romanian wife can easier
overcome some of the fiscal and regulatory issues, by assigning her an important role in their
business.
On the contrary, the vast majority of Turkish respondents unveiled a more positive/ friendlier
attitude, stating that there are no major distinctions between the two business environments,
Turkish and Romanian. They had no restrains in emphasising many similarities, for example
accounting rules, which makes easier for Turkish entrepreneurs to run businesses in
Romania, compared to Arabs. Furthermore, the historical relationships between Romanians
and Turks are likely to explain a more flexible, adaptable, and more open displayed attitude
of inquired Turkish entrepreneurs towards the host country’s culture. However, in some
cases, heavier fiscal and regulatory constraints and sharper bureaucracy were reported by
Turkish interviewees. Furthermore, official entities’ control rounds in Romania were
assessed as more complex and frequent compared to the ones usually performed in Turkey.
Just like Romanian entrepreneurs themselves, many of our immigrant respondents’
complains are related to the legal framework. Despite its thick structure, they all blame its
high degree of interpretability, its rapid and frequent changes in both content and form, thus
generating a lot of confusion and, in some situations even corruption.
Several “pull” factors have been highlighted by the immigrant respondents. For instance,
Romania is considered to having a market with a lower competition compared to the one in
the interviewees’ countries of origin, to provide clients that are easier to satisfy and
employees that can be paid at lower salaries, however, not quite trustworthy. Furthermore,
access to capital is less demanding in Romania, and the required start-up capital is much
lower compared to the one needed in Turkey, for example, as expressed by the investigated
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 269
Middle Eastern immigrant entrepreneurs. Other host country’s strengths, distinctly pointed
out by our respondents refer to: Romania’s excellent geographical position, its status of an
EU and NATO member, its safety and stability, and its rich and inclusive culture.
“The Romanian legal framework has low stability. I consider that in Turkey it is easier
to start a business. I know the laws. In Romania, bureaucracy is very high.” (Turkish
entrepreneur)
“There are many controls in Romania and, most of them, are very intense. In Turkey,
the competition is higher. In Romania, the profit is higher.” (Turkish entrepreneur)
“After Romania’s accession to the EU, things have improved a lot. […] The business
environment in Romania is safer and the competition is not very high, especially when
compared to Turkey.” (Turkish entrepreneur)
“Compared to Turkey, Romania is more peaceful, the stress level is lower, and my time
management is better. On the other hand, the business environment in Turkey is more
developed.” (Turkish entrepreneur)
“If you know the language and if you understand the bureaucracy, you can do great
business in Romania. It is better than in Turkey; the profit is higher.” (Turkish
entrepreneur)
“Syria does not have a free market. Romania offers a free market. Romania is better for
business.” (Arab entrepreneur)
“In Syria, taxes are lower, but the bureaucracy and the people are the same.” (Arab
entrepreneur)
“In Syria, everything is better, but the political conditions are bad. When the war ends,
I will go back to Syria to develop a business there. I want to go back home, but the war
is a big problem.” (Arab entrepreneur)
“In Syria, trust is very important. You can do business just based on trust. People are
very correct in Syria. In Romania, people are not always trustworthy, and it is still a lot
of corruption.” (Arab entrepreneur)
“In Lebanon the controls are lighter, and the law is more permissive.” (Arab
entrepreneur)
“In Egypt, the regulatory frame is more stable and there are fewer taxes.” (Arab
entrepreneur)
“In Egypt, you receive a lot of subsidies, especially for agriculture. As producer, is better
to develop businesses in Egypt, especially in the agriculture area, than in Romania
where agriculture is not so valued.” (Arab entrepreneur)
“In Romania, the market is bigger than in Jordan, but in Jordan controls and fines are
lighter. In Romania, whenever you get a control, for sure, you will get a fine.” (Arab
entrepreneur)
“Iraq has a greater market compared to Romania, but in Iraq things are not going very
well.” (Arab entrepreneur)
“In Kuwait, the market is smaller, and the competition is higher.” (Arab entrepreneur)
AE Immigrant Entrepreneurship in Romania: Drawing Best Practices From Middle Eastern Immigrant Entrepreneurs’ Experiences
270 Amfiteatru Economic
Based on their entrepreneurial experience and considering all the previously expressed
opinions, a series of measures were suggested by the investigated Middle Eastern immigrant
entrepreneurs in Romania, able to improve their entrepreneurial endeavours. Most of them
are valid for all types of entrepreneurs active on the Romanian business environment. Among
them, the most frequently mentioned refer to: reducing bureaucracy, lowering taxes,
decreasing credits’ interest rates to ease access to finance, stabilising the legal framework,
especially by complying to the EU’s recommendations.
“Controls accomplished by different authorities should be stricter and fewer. […] Some
authorities’ attitude, sometimes, makes you became corruptible, when bribery is
encouraged, and, if you become corrupt you are afterwards punished in accordance with
the law. […] These practices should be stopped. Bribery is not allowed in a solid
entrepreneurial system.” (Turkish entrepreneur)
“Bureaucracy needs to be reduced.” (Turkish entrepreneur)
“Credits should have lower interests.” (Turkish entrepreneur)
“Taxes should be lowered.” (Turkish entrepreneur)
“There should be developed a uniform system for everyone, in accordance with the EU’s
requirements, and not to change it very often. Romania should better protect its
entrepreneurial environment. All the entrepreneurs must be treated in an equal
manner.” (Turkish entrepreneur)
“The process of getting a credit should be easier, employees should have insurances (the
process for getting them for the employees should be easier), and the certifications
should be easier to obtain.” (Turkish entrepreneur)
“Romania should lower corruption, improve justice, and properly apply the EU’s
recommendations.” (Arab entrepreneur)
“Laws should be better, more encouraging for entrepreneurs and for investors.” (Arab
entrepreneur)
“In their first years of business, immigrant entrepreneurs should be supported by
Romanian authorities, especially from juridical point of view.” (Arab entrepreneur)
Concluding remarks: outlining best practices specific to Middle Eastern immigrant
entrepreneurs
Based on various results of a multi-method qualitative field research, this paper aimed at
contributing to a more complex and comprehensive image on the Middle Eastern immigrant
entrepreneurship phenomenon in Romania. In this final section, the main conclusions of our
research are also crystallized into a best practices guide.
Middle Eastern immigrant entrepreneurs are significant players on the Romanian market,
contributing to the total added value in the country’s economy. Their efforts, business
initiatives are worth to be promoted as strong role models for other immigrants and their
communities. Both forms of opportunity and necessity-driven entrepreneurship have been
identified among the investigated Middle Eastern immigrants. The most frequent mentioned
“push” factor was the impossibility to find a job in the host country.
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 271
The investigated businesses were owned by Middle Eastern male entrepreneurs that
immigrated in Romania between 1978 and 2015, from Turkey, Syria, Jordan, Iraq, Kuwait,
or Lebanon. Romania’s geographical position, its EU and NATO member status, its safety
and stability, along with its diverse and inclusive culture represented strong arguments in
choosing it as a destination country. Furthermore, from a business perspective, when
compared to their origin countries, Romania is considered to: have a market with a lower
competition; provide clients that are relatively easier to satisfy and employees that can be
paid at lower salaries; offer easier access to capital; require a lower start-up capital. The
reasons for migration were varied, ranging from studying, the desire to find a job or to
become entrepreneur, to family reunification, tourism, etc. However, especially when
compared to Arabs, Turkish immigrant entrepreneurs had clearer business-related
immigration reasons in Romania. The main industries attractive for the investigated
immigrant entrepreneurs to start and/or develop their businesses were: commerce,
production, construction, real estate, and HORECA. In terms of shareholding, usually, Turks
have a business partner – in most of the cases of Turkish nationality. However, Romanian
shareholders were also encountered in businesses ran by Turkish immigrant entrepreneurs.
On the other hand, in general, Arabs, started their business mostly by themselves.
Almost all the investigated entrepreneurs were not at their first entrepreneurial attempt, and
many of them were also having previous experience as employees, usually in the same
branch. Thus, the experience and expertise in the same field of activity, was perceived as an
advantage by many of our respondents.
With a high level of education, most of the investigated immigrant entrepreneurs considered
education an important provider not only of basic economic and foreign languages
knowledge but also a provider of an appropriate context for business skills development.
Knowing a foreign language is considered extremely important when doing business. Other
key aspect strongly considered for the entrepreneurial endeavour was the social networks.
This was a focus for many inquired immigrant entrepreneurs, even during their university,
master and/or doctorate studies. However, there were also entrepreneurs who dropped out of
school, or who have low level of education, primarily among those who strongly believe that
entrepreneurship success is not education related. Role models are perceived as very
important in entrepreneurship processes by the interviewees. Mostly masculine figures in
the family – fathers, grandfathers, uncles and/or brothers – have been indicated as powerful
influencers on the respondents’ decisions to take the entrepreneurial path.
The investigated immigrant entrepreneurs who started their business immediately after the
fall of the communist regime – in a market free of competition, with an enormous unsatisfied
demand for products and services – guided themselves only by their strong native
entrepreneurial spirit and by empirical basis. Such an approach was less valid for more recent
immigrant entrepreneurship practices. Professional feasibility studies and market research
were previously designed to support improved business decisions.
Through the activities carried out, the investigated Middle Eastern immigrant entrepreneurs
had a positive contribution to the development of the Bucharest-Ilfov region, at various
levels: economic, social and cultural. Relevant examples in this regard include: new jobs
creation – mainly for Romanians, but also for Turks, Arabs, Bulgarians, Hungarians,
Macedonians and Moldovans, paying taxes, introducing new, innovative products/services
on the Romanian market, the implementation of new managerial styles, enrichment of the
general fund of customs and traditions with those specific to the culture of the country of
AE Immigrant Entrepreneurship in Romania: Drawing Best Practices From Middle Eastern Immigrant Entrepreneurs’ Experiences
272 Amfiteatru Economic
origin, etc. Yet, various obstacles were perceived by the investigated immigrant
entrepreneurs during their entrepreneurial career: language and culture barrier, poor access
to financial resources, human resources-related issues, bureaucracy, difficult fiscal system,
unclear legislative framework, corruption, difficulties in obtaining visas, and, even,
discrimination. The method invoked as being frequently used to overcome such obstacles
was to turn to legal/tax advisers, in the case of Turkish respondents, respectively to co-opt as
a business partner the Romanian wife, in the case of Arabs. Families, friends and/or business
partners were mentioned by the majority of those interviewed as representing major moral
and financial support. In addition, many respondent entrepreneurs suggested specific
measures that could help create a more immigrant-friendly business environment in
Romania: alleviating bureaucracy, lowering taxes, decreasing credits’ interest rates to ease
access to finance, stabilising the regulatory framework, especially by complying to the EU’s
recommendations.
The results of this study are converging to the conclusion, similar to others in the field, that
entrepreneurship is a viable career option, regardless of the motivation or timing of the
person’s decision to migrate. Exploiting the various business opportunities identified in the
host country could result in the creation of profitable businesses by immigrants.
In compliance with the paper regarding “Evaluation and Analysis of Good Practices in
promoting and supporting Migrant Entrepreneurship”, developed at the EU level (European
Commission, 2016) and with the main results of our research, we designed a guide
summarizing best practices, drawn mainly from the experiences of Middle Eastern immigrant
entrepreneurs included in our research sample and that are not resource intensive. These are
circumscribed to the following seven, out of the ten “dimensions in action”, specified by the
guide (European Commission, 2016): “networking”; “legal advice”; “individual business
support”; “group business training”; “mentoring”; “access to finance”; “language and
cultural sensitivity”. The guide should constitute a benchmarking tool able to contribute to
fostering entrepreneurship amid immigrant communities in Romania further on.
Using/extending opportunities to network with other immigrant entrepreneurs and/or
Romanian entrepreneurs, including potential providers and customers; this would favour
access to new information and knowledge, stimulating business set-up and scale-up efforts.
Accessing available schemes/formal entities responsible for the enhancement of
entrepreneurship in general, and that of immigrant entrepreneurship in particular.
Carefully periodically checking, the institutional and legal frameworks in order to
easily comply with the host’s country regulations; request for professional support if
necessary.
Exploring the help that one or more family members or non-members could provide,
based on trust and complementarity, through their experience and knowledge, in the
entrepreneur’s efforts to overcome the inherent administrative and socio-cultural challenges.
Encourage group training to share and exchange acquired experiences, especially
regarding host country related aspects; try to preserve/pass traditional ethnic crafts’ know-
how to support/build your distinct brand.
Consider employing both ethnic and native persons in your area.
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 273
Take into consideration strong role models of former and current successful
entrepreneurs, as inspiring and powerful factors influencing entrepreneurial decisions in
immigrant communities.
Consider a mentoring partnership, both as mentee seeking for complementary support
and/or mentor sharing with ethnic community members accumulated experience as
established entrepreneur within the Romanian business environment.
Consider multiple business financing sources, including alternative ones (like
crowdfunding, Business Angels etc.) and other mainstream support schemes for
entrepreneurs, encompassing guarantee funds too.
Find viable ways to reduce language barriers in the host country, by taking native
language informal and formal courses, especially regarding business related terminology.
This narrow ethnic group approach should be integrated into a more holistic one at local,
regional or national levels, to properly cover all the ten “dimensions in action” – “visibility”,
“facility provision” and “impact” included – consistent with the guide book’s
recommendations (European Commission, 2016). It would encourage the design of
specifically tailored programs providing multi-dimensional support, like other successful
ones in Europe, capable of generating positive spill overs, inside and outside the immigrant
communities in Romania.
References
Basu, A., 2006. Ethnic minority entrepreneurship. In: M. Casson, B. Yeung, A. Basu and N.
Wadeson eds., 2006. The Oxford Handbook of Entrepreneurship. New York: Oxford
University Press, pp.580-600.
Baycan, T., Sahin, M. and Nijkamp, P., 2012. The urban growth potential of second-
generation migrant entrepreneurs: A sectoral study on Amsterdam. International Business
Review, [e-journal] 21(6), pp.971-986. https://doi.org/10.1016/j.ibusrev.2011.11.005.
Bolzani, D. and Boari, C., 2018. Evaluations of export feasibility by immigrant and non-
immigrant entrepreneurs in new technology-based firms. Journal of International
Entrepreneurship , [e-journal] 16, pp.176-209. DOI: 10.1007/s10843-017-0217-0.
Constantin D.L., Goschin Z. and Drăguşin M., 2008, Ethnic entrepreneurship as an
integrating factor in civil society and a gate to religious tolerance: a spotlight on Turkish
entrepreneurs in Romania. Journal for the Study of Religions and Ideologies, 7(20),
pp.49-79.
Davidescu, A.A.M., Strat, V.A., Grosu, R.M. and Zgura, I.D., 2017. Determinants of
Romanians' Migration within the European Union: Static and Dynamic Panel Gravity
Approaches. Amfiteatru Economic, 19(46), pp.621-639.
De Luca, D. and Ambrosini, M., 2019. Female immigrant entrepreneurs: more than a family
strategy. International Migration, [e-journal] 57(5), pp.201-215. https://doi.org/10.1111/
imig.12564.
Dinu, V., Grosu, R.M. and Saseanu, A.S., 2015. Romanian Immigrant Entrepreneurship:
Utopia or Reality? An Overview of Entrepreneurial Manifestations of Romanian
Immigrants in Andalusia, Spain. Transformations in Business & Economics, 14(1(34)),
pp.48-64.
AE Immigrant Entrepreneurship in Romania: Drawing Best Practices From Middle Eastern Immigrant Entrepreneurs’ Experiences
274 Amfiteatru Economic
European Commission, 2016. Evaluation and Analysis of Good Practices in promoting and
supporting Migrant Entrepreneurship. [online] Available at: <https://ec.europa.eu/
easme/sites/easme-
site/files/documents/guide_book_promoting_and_supporting_migrant_entrepreneurship.
pdf> [Accessed 2 June 2020].
Grosu, R.M. and Dinu, V., 2016. The migration process of Romanians to Andalusia, Spain.
Focus on socio-economic implications. E & M Ekonomie a Management, [e-journal]
19(2), pp.21-36. http://dx.doi.org/10.15240/tul/001/2016-2-002.
Grosu, R., 2015. Dynamics of immigrant entrepreneurship in Romania. Economy of Region,
2, pp.172-182.
Grosu, R.M. and Saseanu, A.S., 2014. Immigrant entrepreneurship – a challenge to
commodity science in the age of globalization. In: C. Andrzej and S. Jerzy eds., 2014.
Commodity Science in Research and Practice - Achievements and challenges of
commodity science in the age of globalization. Krakow: Polish Society of Commodity
Science, pp.119-130.
Grosu, R.M. and Constantin, D.L., 2013. The International Migration in the EU. A
Descriptive Analysis Focused on Romania. Acta Universitatis Danubius – Oeconomica,
9(4), pp.:306-319.
Kerr, S.P. and Kerr, W.R., 2016. Immigrant Entrepreneurship. Harvard Business Scholol.
Working Paper 17-011. (NBER Working Paper Series, No. 22385, July 2016.) [online]
Available at: <https://www.hbs.edu/faculty/Pages/item.aspx?num=51304> [Accessed 22
September 2019].
Kitching, J., Smallbone, D. and Athayde, R., 2009. Ethnic Diasporas and Business
Competitiveness: Minority-Owned Enterprises in London. Journal of Ethnic and
Migration Studies, [e-journal] 35(4), pp.689-705. https://doi.org/10.1080/
13691830902765368.
Kloosterman, R.C., 2010. Matching opportunities with resources: A framework for analyzing
(migrant) entrepreneurship from a mixed embeddedness perspective. Entrepreneurship
and Regional Development, [e-journal] 22(1), pp.25-45. https://doi.org/
10.1080/08985620903220488.
Kloosterman, R., Van der Leun, J. and Rath, J., 1999. Mixed embeddedness: (in)formal
economic activities and immigrant businesses in the Netherlands. International Journal
of Urban and Regional Research, 23(2), pp.253-267.
Ilhan-Nas, T., Sahin, K. and Cilingir, Z., 2011. International ethnic entrepreneurship:
antecedents, outcomes and environmental context. International Business Review,
[e-journal] 20(6), pp.614-626. 10.1016/j.ibusrev.2011.02.011.
Jones, T., Ram, M. and Villares-Varela, M., 2019. Diversity, economic development and new
migrant entrepreneurs. Urban Studies, [e-journal] 56(5), pp.960-976. https://doi.org/
10.1177/0042098018765382.
Neville, F., Orser, B., Riding, A. and Jung, O., 2014. Do young firms owned by recent
immigrants outperform other young firms?. Journal of Business Venturing, [e-journal]
29, pp.55-71. DOI: 10.1016/j.jbusvent.2012.10.005.
Oficiul National al Registrului Comertului, 2009. Societati cu participare straina la capital.
[online] Available at: <https://www.onrc.ro/index.php/ro/statistici?id=254> [Accessed
19 November 2019].
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 275
Ozmen, O. and Grosu, R.M., 2020. Business in a Foreign Country: A Contextual Analysis of
Immigrant Entrepreneurship and Their SMEs. In: Thrassou, A., Vrontis, D., Weber, Y.,
Shams, S.M.R. and Tsoukatos, E., 2020. The Changing Role of SMEs in Global Business.
Vol. II: Contextual Evolution Across Markets, Disciplines and Sectors. Book
Series: Palgrave Studies in Cross-disciplinary Business Research, In Association with
EuroMed Academy of Business. Palgrave Macmillan, pp.39-60. DOI 10.1007/978-3-030-
45835-5.
Rahman, M.M., 2018. Development of Bangladeshi immigrant entrepreneurship in Canada.
Asian and Pacific Migration Journal, [e-journal] 27(4), pp.404-430. https://doi.org/
10.1177/0117196818810096.
Sharbek, N. and Grosu, R.M., 2018. Entrepreneurship: nurturing self-esteem in a new
generation of immigrant Arab women in Romania. Romanian Journal of Regional
Science, 12(2), pp.70-91.
Vinogradov, E. and Jorgensen, E.J.B., 2017. Differences in international opportunity
identification between native and immigrant entrepreneurs. Journal of International
Entrepreneurship, [e-journal] 15(2), pp.207-228. DOI: 10.1007/s10843-016-0197-5.
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 293
As it could be seen in the literature there is empirical evidence that factors such as income, Gini Index, financial development, economic growth, urbanization, financial literacy, human development, and others have an impact on the insurance market.
At the microeconomic level, we identify many articles that study private health insurance in a certain country based on the survey data. In this regard, Bolhaar, Lindeboom and van der Klaauw (2012) show that in the Irish health care system, the asymmetric information is vital to the acceptance of supplementary private health insurance. Nguyen and Knowles (2010) show that the demand for school-age children and adolescent/student health insurance in Vietnam increases significantly with the expected benefits of insurance as measured by proximity to and quality of a tertiary hospital. The findings of Yang (2018) suggest that a key factor that impacts people's enrolment in the voluntary health insurance scheme in rural China is the perception of the quality of care. More recently, Innocenti, et al. (2019) examine if past negative health experiences are positively associated with intentions to purchase insurance for mitigating the risks of income losses due to illnesses and disabilities. Pendzialek, Simic and Stock (2016) review the empirical studies on price elasticity of demand for health insurance.
In a recent review of literature, Śliwiński and Borkowska (2020) divided the determinants of the demand for private voluntary health insurance into the following categories: income and time value, education, age, gender and family size, health status, level of risk aversion and others. However, it is expected that the key factors of health insurance demand are income or health status. At macroeconomic level, it has already been proven (Enz, 2000; Beck and Webb, 2003) the significant and positive influence of income on the insurance demand. At the national level, Propper (1993) in UK, Christiansen, et al. (2002) in Denmark; Barrett and Conlon (2003) in Australia; Machnes (2006) in Israel; Finn and Harmon (2006) in Ireland found the same results. Recently, Tavares (2020) using collected data from the Survey of Health, Aging and Retirement in Europe for 21 countries shows a direct relationship between income and health insurance. The study concludes that younger European men, with higher income, well-educated, married, employed, are more likely to buy a health policy.
Inherent several individual characteristics and socio-economic variables are found to be correlated with the private health insurance demand. In an in-depth analysis of this subject, Kiil (2012) identified 39 papers and concludes that the probability of buying health insurance increases with income and education level. The insurance literature identifies both at the microeconomic level and at the macroeconomic level (Li, et al., 2007; Kjosevski, 2012), a series of common determinants, but also particularities (especially at individual level).
We believe that we all agree with this statement “While people are unable to save their health, they are able to invest in their health” (Lieberthal, 2016, pp. 35). It remains to identify what determines Europeans to purchase private health insurance in a comprehensive study, starting from the assumption that we cannot buy our health, but we can improve the conditions in which we will be treated, investing in a private health policy.
1.2. Hypotheses
H1: Health insurance density is positively influenced by the incomes of the inhabitants.
As in the case of any product or service, we expect a direct correlation with incomes. Additionally, especially in the case of low purchasing powers, insurance products are considered as luxury products. That is why we actually expect a non-linear relationship, maybe exponential.
AE Socio-Economic and Macro-Financial Determinants and Spatial Effects on European Private Health Insurance Markets
294 Amfiteatru Economic
H2: Health insurance density is positively influenced by the level of financial development
of the country.
Even in the presence of incomes (as control factor), financial development may play a
significant role. Despite the rapid economic growth, financial instruments in the former
communist countries did not get near the developed European economies. This effect may be
due to two major causes. Insurance premiums paid by the insured are invested by the
insurance companies on the stock market. If the latter is less significant in a national
economy, it may transform in a barrier for the development of the insurance sector, in general,
and of the health one in particular. The second cause is determined by the financial education
of the citizens. Low levels of this cast down the insurance market development.
H3a: An East-West clusterization direction is to be found for the health insurance sector.
H3b: There are significant diffusion and contagion effects.
The former communist countries of the Eastern Europe have had, for a very long time, state-
controlled health systems. Consequently, their citizens were not educated towards a private
sector that would require private health insurance contracts. That is why we expect to see a clear
clusterization of the sample on the East-West direction, with Eastern countries having much
lower health insurance densities. Additionally, as stated in the previous working hypotheses, and
following the works of Mare, et al. (2016, 2019a, 2019b) we expect significant contagion or
diffusion processes on the health insurance market, conditioned by both the level of human and
financial development and the income (internal conditions specific to each country). The
contagion and diffusion processes should appear due to the high interdependencies existing
among the European countries, that make information travel much faster.
2. Research methodology
2.1. Methodology
In order to evaluate the three working hypotheses, we have employed, on one hand, the
classical OLS regression and, on the other, spatial econometrics tools. Due to the high level
of heterogeneity and to treat heteroskedasticity, the OLS regression was constructed in the
robust form, with the White correction method applied (White, 1980). The robust estimator
obtained (White’s estimator or heteroskedasticity-consistent estimator – HCE) is given by
the following formula:
122
1
1 )'()ˆ,...,ˆ(')'()ˆ( XXXdiagXXXv nOLSHCE (1)
Hypotheses H1 and H2 were tested through this method.
Spatial analysis methods were applied in the last part of the research, to evaluate H3. If
significant clusterization processes occur, either on the North-South or the East-West
directions, the coefficients of latitude and longitude should be significant in an OLS spatial
regression. The sign of the coefficients shows the direction of the increment in the variable
(in this case, the health insurance density). The latitude and longitude variables are, actually,
the values of the centre means corresponding to the centroids of each spatial unit assessed.
But sometimes, the intensity of a phenomenon is much better explained if compared to
another. That is why, in order to assess the contagion and diffusion processes stated in H3b,
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 295
we have constructed the rate maps of the LN_DENS over HDI and FIN_DEV. The spatial
factor was introduced through the spatial weights matrix used to spatially smoothen the rate
maps. This procedure is similar to the robust approach in the classical OLS, being meant to
treat the high heterogeneity characteristic to spatial data.
The existence of contagion and diffusion is confirmed if clear spatial arrangements are
emphasized by the spatially smoothed rate maps. Another method is to test if there are spatial
interactions that should be included in a regression model. The starting point is, again the
OLS model, but with the neighboring scheme given by the spatial weights matrix attached.
If the spatial diagnostic post-estimation tests (Moran’s I for errors and Lagrange Multiplier
– LM tests) reject the null hypothesis that the best fitting model is the OLS and they
emphasize the need to re-specify the regression with spatial components (spatial
autoregressive or spatial moving average), then contagion and diffusion exist. The opposite
situation is valid when the OLS regression is accepted.
2.2. Data and variables
Data refers to 30 European countries for which information related to health insurance is
available, aggregated at national level (Appendix no. 1). Variables to be used in the analysis
for the hypotheses testing are synthesized in Table no. 1. They are coded, with explanations
and the data source.
Table no. 1. Variables used in the analysis and sources of data
Dependent variables
HLTH_INS_
DENS
Health Insurance Density (total premiums per inhabitant - domestic
market) is calculated as the ratio of total insurance premiums (in
Euros) to total population. Due to the large positive asymmetry and the
nonlinear correlation with the explanatory variables, in regressions it
will be used in its logarithmic form.
Data from 2018. Source: Insurance Europe
LN_DENS The natural logarithm of the variable HLTH_INS_DENS.
Explanatory variables
GDP_CAP GDP per capita (103 US $) is gross domestic product divided by
midyear population.
Data from 2018. Source: World Bank.
GDP_PPP GDP by Purchasing Power Parity indicator provides per capita values
for gross domestic product (GDP) expressed in 103 current
international dollars converted by purchasing power parity (PPP)
conversion factor.
Data from 2018. Source: World Bank.
GINI_IND Gini Index measures the extent to which the distribution of income
among individuals or households within an economy deviates from a
perfectly equal distribution.
Data from 2018. Source: World Bank estimates.
URBAN Urban population (% of total) refers to people living in urban areas as
defined by national statistical offices.
Data from 2018. Source: World Bank.
AE Socio-Economic and Macro-Financial Determinants and Spatial Effects on European Private Health Insurance Markets
296 Amfiteatru Economic
LIFE_EXP Life Expectancy at birth indicates the number of years a newborn
infant would live if prevailing patterns of mortality at the time of its
birth were to stay the same throughout its life. Data from 2018. Source:
World Bank.
HDI The Human Development Index (HDI) is a summary measure of
average achievement in key dimensions of human development: a long
and healthy life, being knowledgeable and have a decent standard of
living. The HDI is the geometric mean of normalized indices for each
of the three dimensions. Data from 2018. Source: United Nations
Development Programme – Human Development Reports.
MK_CAP Market capitalization of listed domestic companies (% of GDP) is the
share price times the number of shares outstanding (including their
several classes) for listed domestic companies. Due to the large
fluctuations of prices on the capital market and the lack of data for a
certain year or country in the sample, the maximum value from the
time interval 2014 - 2018 was considered. Source: World Bank.
PRV_CRD Domestic credit to private sector (% of GDP) refers to financial
resources provided to the private sector by financial corporations, such
as through loans, purchases of nonequity securities, and trade credits
and other accounts receivable, that establish a claim for repayment.
Data from 2018. Source: World Bank.
FIN_LIT Financial Literacy Index survey probes four basic financial concepts:
risk diversification, inflation, numeracy, and compound interest. Based
on interviews with more than 150,000 adults across 148 countries, the
survey gives researchers, policy makers, and practitioners a unique,
first-of-its kind, and in-depth look at financial literacy across the globe.
Data from 2018. Source: Standard & Poor’s.
FIN_DEV Financial Development is a composite index based on the variables
MK_CAP, PRV_CRD and FIN_LIT. Calibrated for possible values
from 0 to 100. High values of the index are due to high values of the
credit market, the capital market and the good financial education of
the population.
Source: authors’ calculation.
Source: World Bank, Insurance Europe, Standard & Poor’s, United Nations Development
Programme and authors’ calculation.
2.3. Construction of the FIN_DEV composite index
FIN_DEV is defined based on the MK_CAP, PRV_CRD and FIN_LIT variables. These have
different measurement units and scales. In order to aggregate them, we have, first,
standardized them. Let:
I1 = MK_CAP (2)
I2 = PRV_CRD (3)
I3 = FIN_LIT (4)
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 297
Using the index i = 1,…,N for the countries in the sample and Ij , j = 1,2,3 the three indicators.
A normalized value, centered and reduced, can be computed for each indicator and country,
through the classical standardization procedure (subtracting the average and dividing by the
standard deviation).
jI
jjijiji
IInormIX
)(
(5)
the normalized value for indicator Ij for the i country.
In order to obtain values from 0 to 100 for each indicator, we use the normal distribution
(Gauss) upon the standardized value of each indicator.
321
2/
3
2/
2
2/
1
32
22
!2
2
1100
2
1100
2
1100
_ccc
dtecdtecdtec
DEVFIN
iii X
t
X
t
X
t
i
(6)
Where c1, c2 and c3 represent shares (%) of the I1, I2 and I3 indicators that can be changes
under the restriction: c1 + c2 + c3 = 1 (100%). In our application, we have maintained equal
shares (c1 = c2 = c3) for MK_CAP, PRV_CRD and FIN_LIT.
To verify the robustness of the aggregation of a composite index based on its component
items, we use Cronbach’s alpha (tau-equivalent reliability).
2
1
2
11 X
k
i i
Tk
k
Xi denotes the observed score for item i, while X=(X1+X2+…Xi+…+Xk) is the composite
index (variable) index formed by the k items X1, X2,…Xi,…Xk. σi denotes the standard deviation
of Xi and σX denotes the standard deviation of X.
3. Results and discussion
The distribution of the health insurance density values for the countries in the sample points
out a very high variation, from 0.86 €/habitant in Hungary to 2719 €/habitant in the
Netherlands.
Figure 1 shows the country distribution of the HLTH_INS_DENS variable, both on the linear
and logarithmic scales. We do this in order to be able to also observe the differences existing
between countries with low values.
Evaluating the descriptive statistics of the variables we observe that the relative variation of
the health insurance density (2.213) is much higher than that of the income proxy variables
(GDP_CAP and GDP_PPP) (Table 2). This suggests that the development of the health
insurance sector is not strictly linked to the population’s purchasing power, but also to other
factors (like, for example, the financial development of a country).
AE Socio-Economic and Macro-Financial Determinants and Spatial Effects on European Private Health Insurance Markets
298 Amfiteatru Economic
Figure no. 1a. Linear scale Figure no. 1b. Logarithmic scale
Figure no. 1. Health insurance density distribution for the countries in the sample
Source: authors’ calculation in Excel, using data from Insurance Europe (2018)
Table no. 2. Descriptive statistics
Min Max Mean St Dev Coef Var
HLTH_INS_DENS 0.865 2719 235.8 521.8 2.213
GDP_CAP 9.370 116.65 38.833 25.74 0.663
GDP_PPP 22.60 116.79 46.490 19.47 0.419
GINI_IND 24.2 41.4 31.5 4.19 0.133
URBAN 53.7 98 74.5 11.6 0.156
0.9
3.7
5.6
9.0
9.0
12.7
12.8
17.1
20.9
21.7
32.0
36.2
45.7
76.0
80.2
108.0
114.1
120.6
124.1
145.3
152.9
162.0
179.2
190.0
251.5
264.9
480.8
532.5
1145.7
2718.7
0 1000 2000 3000
Hungary
Romania
Poland
Estonia
Bulgaria
Czech R.
Turkey
Croatia
Latvia
Greece
Malta
Slovakia
Italy
Portugal
Finland
U Kingdom
Denmark
Norway
Sweden
Luxemburg
Belgium
Cyprus
Spain
France
Austria
Slovenia
Germany
Ireland
Switzerland
Netherlands
0.9
3.7
5.6
9.0
9.0
12.7
12.8
17.1
20.9
21.7
32.0
36.2
45.7
76.0
80.2
108.0
114.1
120.6
124.1
145.3
152.9
162.0
179.2
190.0
251.5
264.9
480.8
532.5
1145.7
2718.7
0 1 10 100 1000 10000
Hungary
Romania
Poland
Estonia
Bulgaria
Czech R.
Turkey
Croatia
Latvia
Greece
Malta
Slovakia
Italy
Portugal
Finland
U Kingdom
Denmark
Norway
Sweden
Luxemburg
Belgium
Cyprus
Spain
France
Austria
Slovenia
Germany
Ireland
Switzerland
Netherlands
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 299
Min Max Mean St Dev Coef Var
LIFE_EXP 74.8 83.8 80.3 2.69 0.034
HDI 0.806 0.954 0.890 0.042 0.047
MK_CAP 8.3 160.9 58.1 38.0 0.655
PRV_CRD 25.7 139.6 78.5 28.5 0.363
FIN_LIT 22 71 50.3 13.606 0.270
FIN_DEV 4.9 89.6 49.2 23.9 0.486
Source: authors’ calculation in Excel
Additionally, one can see that the distribution of the health insurance density has a significant
positive skewness (Figure no. 2). Consequently, we will use the log of the variable in the
regression models, LN_DENS. In this way, the shape of the distribution gets closer to the
normal one (Figure no. 3).
Moreover, skewness reduction also relaxes the heteroskedasticity problem. After taking the
log, the distribution is quasi-symmetrical and gets closer to a normal one (Figure no. 3).
Figure no. 2. Asymmetrical distribution of
the HLTH_INS_DENS variable
Figure no. 3. Distribution of the LN_DENS
variable
In the single factor regressions (Table no. 3, OLS 1-8), almost all variables are significant
(with the exception of the GINI_IND) and with the expected sign.
But in the multiple regressions (Table no. 4, OLS 9-10) some of the coefficients lose their
significance. This is due to the strong correlations existing between the exogenous variables
(Table no. 5).
In order to overcome this inconvenient and be able to demonstrate the influence of both the
purchasing power and the financial development upon the health insurance density, we
employ composite indexes (variables). For a country’s standard of living there is such an
index – HDI – the Human Development Index.
Among other factors, this includes the GDP/cap and the Life Expectancy – which are also
considered in our study. But for the financial development of a country we had to construct
an aggregated index, FIN_DEV, which contains the variables related to the credit and stock
markets and to the financial education of population. The index is representative for its
0
.002
.004
.006
.008
De
nsity
0 1000 2000 3000HLTH_INS_DENS
0.1
.2.3
.4
De
nsity
-2 0 2 4 6 8LN_DENS
AE Socio-Economic and Macro-Financial Determinants and Spatial Effects on European Private Health Insurance Markets
300 Amfiteatru Economic
components (Cronbach alpha = 0.761). The technical aspects related to the construction of
FIN_DEV are to be found in the Methodology. The possible theoretical values of the indicator
range from 0 to 100, with high values standing for a good financial development of the
country. In our sample of 30 European countries, the lowest value was obtained for Romania
(4.9), while the highest for the Netherlands (89.6). The country distribution of this indicator
is to be found in Appendix no. 2.
Table no. 3. Results of the simple OLS regressions: dependent variable LN_DENS,
coefficients and t-stat
OLS
(1)
OLS
(2)
OLS
(3)
OLS
(4)
OLS
(5)
OLS
(6)
OLS
(7)
OLS
(8)
GDP_CAP ***0.05
(4.65)
. . . . . . .
GDP_PPP . ***0.06
(4.02)
. . . . . .
GINI_IND . . 0.02
(1.25)
. . . . .
URBAN . . . *0.04
(1.72)
. . . .
LIFE_EXP . . . . ***0.49
(6.10)
. . .
MK_CAP . . . . . ***0.03
(5.47)
. .
PRV_CRD . . . . . . ***0.04
(4.16)
.
FIN_LIT . . . . . . . **0.06
(2.69)
constant ***2.36
(5.21)
**1.56
(2.27)
**7.00
(2.82)
0.90
(0.47)
***-35.9
(-5.47)
***2.18
(5.15)
0.98
(1.31)
1.15
(1.01)
R2 0.436 0.366 0.046 0.095 0.571 0.516 0.415 0.206
Note: ***,**, *: significant at 1%,5% and 10% level.
Source: authors’ calculation in STATA.
The two composite indexes, HDI and FIN_DEV are both statistically significant, regardless
if they are used as single factors or together, in the same regression (Table no. 4, OLS 11-
13). These results validate hypotheses H1 and H2.
Table no. 4. Results of the multifactorial OLS regressions and the composite indexes
regressions: dependent variable LN_DENS, coefficients and t-stat
OLS (9) OLS (10) OLS (11) OLS (12) OLS (13)
GDP_CAP 0.009
(0.73)
. . . .
GDP_PPP . 0.013
(0.93)
. . .
GINI_IND -0.002
(-0.03)
-0.003
(-0.05)
. . .
URBAN *-0.039
(-1.88)
*-0.040
(-1.93)
. .
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 301
OLS (9) OLS (10) OLS (11) OLS (12) OLS (13)
LIFE_EXP **0.270
(2.25)
**0.263
(2.24)
. .
MK_CAP **0.0178
(2.22)
**0.018
(2.25)
. . .
PRV_CRD 0.008
(0.81)
0.009
(0.90)
. . .
FIN_LIT 0.020
(0.85)
. 0.021
(0.94)
. . .
HDI . . ***32.44
(6.15)
. **19.67
(2.51)
FIN_DEV . . . ***0.055
(5.84)
**0.029
(2.12)
constant *-17.60
(-1.76)
*-17.36
(-1.80)
***-24.73
(-5.26)
***1.425
(2.78)
**-14.78
(-2.29)
R2 0.739 0.743 0.575 0.550 0.635
Note: ***,**, *: significant at 1% ,5% and 10% level.
Source: authors’ calculation in STATA.
Table no. 5. The Pearson correlation coefficients between variables
(1) (2) (3) (4) (5) (6) (7) (8) (9)
LN_DENS (1) 1.000
GDP_CAP (2) 0.660 1.000
GDP_PPP (3) 0.605 0.971 1.000
GINI_IND (4) -0.215 -0.159 -0.121 1.000
URBAN (5) 0.309 0.448 0.400 -0.017 1.000
LIFE_EXP (6) 0.756 0.656 0.594 -0.254 0.464 1.000
MK_CAP (7) 0.718 0.607 0.531 -0.024 0.530 0.623 1.000
PRV_CRD (8) 0.644 0.486 0.389 0.001 0.451 0.655 0.682 1.000
FIN_LIT (9) 0.454 0.591 0.500 -0.525 0.508 0.421 0.394 0.296 1.000
Source: authors’ calculation in STATA.
The present sample consists in European countries, that/which are linked through significant
socio-economic relationships. This intense interdependence must be taken into account, as
clusterization or contagion effects may appear. In order to treat this, we have included space
as an analysis dimension. By doing this, we have also accounted for the neighbouring effect
in the analyzed sample. Results obtained through the classical procedures (Tables no. 3 and
4), show the positive relationship existing between the level of development and the health
insurance market.
AE Socio-Economic and Macro-Financial Determinants and Spatial Effects on European Private Health Insurance Markets
302 Amfiteatru Economic
Figure no. 4. Quartile map LN_DENS over FIN_DEV
Source: authors’ calculation in GeoDa.
Figure no. 5. Quartile map LN_DENS over HDI
Source: authors’ calculation in GeoDa.
As the former is characterized by a significant East-West clusterization in Europe, we also
expect this to be valid for the development of the health insurance sector. Former communist
countries, in which the health sector is strongly sustained by the state, have a lower
development level of this type of insurance products. To test this, we have employed the
simple spatial regression with latitude and longitude as independents. Only the coefficient of
longitude turned out to be significant (p-value = 0.002 < 0.05) and negative, proving that as
we go from West to East, the health insurance density lowers (Table no. 6). For the
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 303
assessment of the contagion and diffusion processes, we have first constructed the rate maps
of LN_DENS over FIN_DEV and HDI. But due to the high heterogeneity level, these maps
were constructed in the spatially smoothed form (Figures no. 4 and 5).
According to Mare et al. (2019a, 2019b), if a significant diffusion and contagion process is
to be found, a clear grouping direction should be present on these maps. None of the two
figures have this feature. The lack of a diffusion and contagion process is finally confirmed
by the regression analysis, as the classical OLS model is not rejected by any of the spatial
diagnostics tests (Table no. 6 – all spatial diagnostic tests have probabilities >> 0.05).
Consequently, we partially accept H3 – there is a significant clusterization based on longitude
(H3a), but no significant contagion and diffusion processes (H3b rejected).
Table no. 6. Spatial regression results; dependent variable LN_DENS;
coefficients and t-stat
OLS (14) OLS (15) OLS (16) OLS (17)
Latitude 0.1e-6 (0.2e-6) - - -
Longitude ***-0.8e-5 (0.2e-5) - - -
HDI - ***32.436 (5.27) - **19.669 (7.82)
FIN_DEV - - ***0.055 (0.009) **0.029 (0.014)
constant ***4.269 (1.31) ***-24.73 (4.7) ***1.43 (0.51) **-14.77 (6.46)
R2 0.313 0.575 0.55 0.635
Spatial diagnostic test (probs)
Moran’s I error - 0.107 0.134 0.162
LM lag - 0.223 0.101 0.122
Robust LM lag - 0.689 0.279 0.298
LM error - 0.187 0.217 0.247
Robust LM er. - 0.519 0.991 0.876
LM SARMA 0.387 0.261 0.298
Note: ***,**, * : significant at 1% ,5% and 10% level.
Source: authors’ calculation in GeoDa.
Conclusion
For a sample of 30 European countries, we have assessed the macroeconomic factors that
determine the development of the private health insurance sector. As expected, there is a high
heterogeneity in terms of health insurance density in the analyzed sample. These high
discrepancies are due to the past of each society and the level of socio-economic
development. And there is an important link between the two. The Western group of countries
was democratic after the 2nd World War, thing that led to a certain type of development and
social education. It is very well known that there is a significant clusterization process in
Europe, with Western states having a much higher purchasing power and standard of living.
Our first set of results clearly shows a positive relationship between the GDP (measured in
any form) and the health insurance density. This can be explained by the fact that the private
health insurance sector was able to develop in countries where the economic environment
allowed for private intervention in the economy, that brought, especially in the health sector,
a much higher quality than the public one. Additionally, this cluster of countries also has the
most developed financial markets and a high level of financial literacy of the citizens. This,
AE Socio-Economic and Macro-Financial Determinants and Spatial Effects on European Private Health Insurance Markets
304 Amfiteatru Economic
again, brings into light another important specificity – the health insurance contract, in
respect to other such products, is seen as a luxury product, so a high standard of living is
necessary for a person to buy such a contract. On the contrary, the Eastern, ex-communist
states, have a long history of publicly sustained sectors, with a very important gap in the
economic and financial education of their inhabitants. In these socialist nations, the inhabitant
was used to be provided with all the social security issues by the state.
A second very important result is the demonstration that, in Europe, both the level of human
development and the financial one, significantly and positively influence the private health
insurance sector. These results emphasize the transmission channels on which actors in this
domain should interfere in order to increase the demand for such products. A last group of
results may be divided into two. On one hand, the spatial analysis conducted has clearly
shown that the positive relationship between the level of development and the health
insurance sector has materialized in a similar clusterization direction in Europe based on the
latter – East-West, with Eastern countries having much lower levels. On the other, an
important assumption was not validated. Previous studies in the field of insurance have
demonstrated contagion and diffusion processes existing for different types of insurance
products. As there is a very high interaction in Europe, due to the European Union and all the
socio-economic agreements, we have also expected such contagion and diffusion to be
present for the private health insurance market. This, because information travels fast, there
is very high mobility of people, goods and capital, etc. On the contrary, the present analysis
has emphasized the lack of spatial contagion and diffusion processes. This is a very important
result, as this means that the private health insurance sector, in comparison with other types
of insurance, is more influenced by the internal, domestic conditions specific to each country.
And this impact is that high, that it cancels the transmission channels and the information
coming from outside a certain nation.
An important implication arises from the analysis – actors interested in developing the private
health insurance market should address the internal specificities of each national market and
create products according to the country’s financial and social development.
Funding
This work was supported by a grant of the Romanian Ministry of Education and Research,
CNCS - UEFISCDI, project number PN-III-P1-1.1-TE-2019-0554, within PNCDI III.
References
Balcilar, M., Gupta, R., Lee, C.C. and Olasehinde-Williams, G., 2020. Insurance-growth
nexus in Africa. The Geneva Papers on Risk and Insurance-Issues and Practice, 45(2),
pp.335-360. DOI: 10.1057/s41288-019-00145-7.
Beck, T. and Webb, I., 2003. Economic, demographic, and institutional determinants of life
insurance consumption across countries. The World Bank Economic Review, 17(1),
pp.51-88.
Beckfield, J., Olafsdottir, S. and Sosnaud, B., 2013. Healthcare systems in comparative
perspective: classification, convergence, institutions, inequalities, and five missed turns.
Annual review of sociology, 39(1), pp.127-146.
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 305
Boes, S. and Gerfin, M., 2016. Does full insurance increase the demand for health care?.
Health economics, 25(11), pp.1483-1496.
Bolhaar, J., Lindeboom, M. and Van Der Klaauw, B., 2012. A dynamic analysis of the demand
for health insurance and health care. European Economic Review, 56(4), pp.669-690.
Browne, M.J. and Kim, K., 1993. An international analysis of life insurance demand. Journal
of Risk and Insurance, 60(4), pp.616-634.
Cantarero-Prieto, D., Pascual-Sáez, M. and Gonzalez-Prieto, N., 2017. Effect of having
private health insurance on the use of health care services: the case of Spain. BMC health
services research, 17(1), pp.716.
Chang, C.H. and Lee, C. C., 2012. Non-linearity between life insurance and economic
development: A revisited approach. The Geneva Risk and Insurance Review, 37(2),
pp. 223-257.
Christiansen, T., Lauridsen, J. and Kamper-Jørgensen, F., 2002. Demand for private health
insurance and demand for health care by privately and non-privately insured in
Denmark. Syddansk Universitet.
Chui, A.C. and Kwok, C.C., 2009. Cultural practices and life insurance consumption: An
international analysis using GLOBE scores. Journal of Multinational Financial
Management, 19(4), pp. 273-290.
Curak, M., Dzaja, I. and Pepur, S., 2013. The effect of social and demographic factors on life
insurance demand in Croatia. International Journal of Business and Social Science, 4(9),
pp. 65-72.
Dragos, S.L., 2014. Life and non-life insurance demand: the different effects of influence
factors in emerging countries from Europe and Asia. Economic research-Ekonomska
istraživanja, 27(1), pp. 169-180.
Elango, B. and Jones, J., 2011. Drivers of insurance demand in emerging markets. Journal
of Service Science Research, 3(2), pp. 185-204.
Enz, R., 2000. The S-curve relation between per-capita income and insurance
penetration. The Geneva Papers on Risk and Insurance-Issues and Practice, 25(3),
pp. 396-406.
Finn, C. and Harmon, C.P., 2006. A dynamic model of demand for private health insurance
in Ireland. Discussion Papers Series IZA DP No. 2472, pp. 1-38.
Gaganis, C., Hasan, I. and Pasiouras, F., 2020. Cross-country evidence on the relationship
between regulations and the development of the life insurance sector. Economic
Modelling, 89(July), pp. 256-272.
Hwang, T. and Gao, S., 2003. The determinants of the demand for life insurance in an emerging
economy – The case of China. Journal of Managerial Finance, 29(5/6), pp. 82-96.
Innocenti, S., Clark, G. L., McGill, S. and Cuñado, J., 2019. The effect of past health events
on intentions to purchase insurance: evidence from 11 countries. Journal of Economic
Psychology, 74(Oct), pp. 1-21.
Insurance Europe, 2015. Statistics: European insurance industry database. Density (total
premiums per inhabitant): domestic market. [pdf] Insurance Europe.
<https://www.insuranceeurope.eu/sites/default/files/attachments/European%20Insurance
%20-%20Key%20Facts%20-%20August%202015.pdf> [Accessed 20 September 2020].
Insurance Europe, 2018. European Insurance in Figures, 2018 data. [pdf] Insurance Europe.
Available at: <https://www.insuranceeurope.eu/sites/default/files/attachments/European
%20Insurance%20in%20Figures%20-%202018%20data.pdf> [Accessed 20 September
2020].
AE Socio-Economic and Macro-Financial Determinants and Spatial Effects on European Private Health Insurance Markets
306 Amfiteatru Economic
Kiil, A., 2012. What characterizes the privately insured in universal health care systems?
A review of the empirical evidence. Health Policy, 106(1), pp. 60-75.
Kjosevski, J., 2012. The determinants of life insurance demand in central and southeastern
Europe. International Journal of Economics and Finance, 4(3), pp. 237-247.
Li, D., Moshirian, F., Nguyen, P. and Wee, T., 2007. The demand for life insurance in OECD
countries. Journal of Risk and Insurance, 74(3), pp. 637-652.
Lieberthal, R.D., 2016. What Is Health Insurance (Good) For?: An Examination of Who Gets
It, Who Pays for It, and How to Improve It. Springer International Publishing.
Lin, C., Hsiao, Y.J. and Yeh, C.Y., 2017. Financial literacy, financial advisors, and
information sources on demand for life insurance. Pacific-Basin Finance Journal,
43(June), pp. 218-237.
Liu, T.C. and Chen, C.S., 2002. An analysis of private health insurance purchasing decisions
with national health insurance in Taiwan. Social science and medicine, 55(5), pp. 755-774.
Machnes, Y., 2006. The demand for private health care under national health insurance. The
European Journal of Health Economics, 7(4), pp. 265-269.
Mare, C., Dragos, S.L., Dragota, I.M., Muresan, G.M. and Urean, C.A., 2016. Spatial
convergence processes on the European Union’s life insurance market. Economic
Computation and Economic Cybernetics Studies and Research, 50(4), pp. 93-107.
Mare, C., Dragoș, S.L., Dragotă, I.M. and Dragoș, C.M., 2019a. Insurance Literacy and
Spatial Diffusion in the Life Insurance Market: A Subnational Approach in Romania.
Eastern European Economics, 57(5), pp. 375-396.
Mare, C., Dragoș, S.L. and Dragotă, I.M., 2019b. The impact of human development on the
Romanian life insurance market: A county spatial econometric analysis. Cogent Business
and Management, 6(1), pp. 1-15.
Nguyen, H. and Knowles, J., 2010. Demand for voluntary health insurance in developing
countries: the case of Vietnam’s school-age children and adolescent student health
insurance program. Social Science and Medicine, 71(12), pp. 2074-2082.
Olasehinde-Williams, G. and Balcilar, M., 2020. Examining the Effect of Globalization on
Insurance Activities in Large Emerging Market Economies. Research in International
Business and Finance, 53(1), pp. 1-15
Outreville, J.F., 1996. Life insurance markets in developing countries. Journal of risk and
insurance, 63(2), pp. 263-278.
Pendzialek, J.B., Simic, D. and Stock, S., 2016. Differences in price elasticities of demand
for health insurance: a systematic review. The European Journal of Health Economics,
17(1), pp. 5-21.
Pitacco, E., 2014. Health Insurance Products. Basic Actuarial Models. Springer International
Publishing.
Propper, C., 1993. Constrained choice sets in the UK demand for private medical insurance.
Journal of Public Economics, 51(3), pp. 287-307.
Śliwiński, A. and Borkowska, I., 2020. Private Voluntary Health Insurance: Market in Poland
and Determinants of Demand – Review of Literature. In: M. Janowicz-Lomott, K.
Łyskawa, P. Polychronidou and A. Karasavvoglou, eds. Economic and Financial
Challenges for Balkan and Eastern European Countries. [online] Cham: Springer
International Publishing. pp.177-192. Available at: <http://link.springer.com/10.1007/
978-3-030-39927-6_11> [Accessed 4 September 2020].
Standard & Poor’s, 2018. S&P Global Financial Literacy Survey: Financial Literacy Around
the World - GFLEC. [online] Standard & Poor’s. Available at: <https://gflec.org/wp-
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 307
content/uploads/2015/11/3313-Finlit_Report_FINAL-5.11.16.pdf?x93521> [Accessed
11 September 2020].
Tavares, A.I., 2020. Voluntary private health insurance demand determinants and risk
preferences: Evidence from SHARE. The International Journal of Health Planning and
Management, 35(3), pp. 685-703.
Trinh, T., Nguyen, X. and Sgro, P., 2016. Determinants of non-life insurance expenditure in
developed and developing countries: an empirical investigation. Applied
Economics, 48(58), pp. 5639-5653.
Truett, D.B. and Truett, L.J., 1990. The demand for life insurance in Mexico and the United
States: A comparative study. Journal of Risk and Insurance, 57(2), pp. 321-328.
White, H., 1980. A heteroskedasticity-consistent covariance matrix estimator and a direct test
for heteroskedasticity. Econometrica, 48(4), pp. 817-838.
World Bank, 2018. Indicators: Economy & Growth, Financial Sector, Gender, Health.
Washington, DC: World Bank.
Yang, M., 2018. Demand for social health insurance: evidence from the Chinese new rural
cooperative medical scheme. China Economic Review, 52, pp. 126-135.
Zerriaa, M. and Noubbigh, H., 2016. Determinants of life insurance demand in the MENA
region. The Geneva Papers on Risk and Insurance-Issues and Practice, 41(3), pp. 491-511.
Appendix no. 1. List of European countries in the sample
Austria (AT), Belgium (BE), Bulgaria (BG), Croatia (HR), Cyprus (CY), Czech Republic (CZ),
Denmark (DK), Estonia (EE), Finland (FI), France (FR), Germany (DE), Greece (GR), Hungary
(HU), Ireland (IE), Italy (IT), Latvia (LV), Luxemburg (LU), Malta (MT), Netherlands (NL),
Norway (NO), Poland (PL), Portugal (PT), Romania (RO), Slovakia (SK), Slovenia (SI), Spain
(ES), Sweden (SE), Switzerland (CH), Turkey (TR), United Kingdom (UK).
Appendix no. 2. Distribution of financial development (FIN_DEV)
Source: authors’ calculation in Excel.
4.913.5
24.825.325.726.7
28.528.8
34.334.534.635.53637.4
39.648.448.6
53.460.961.262.2
6470.3
75.776.577.4
84.184.4
88.989.6
0 20 40 60 80 100
…
Poland
Slovakia
Czech R
Estonia
Greece
Cyprus
Norway
France
Sweden
Netherlands
FIN_DEV
AE Book Review: The Statistical Monograph of the Bucharest University of Economic Studies. 100 Generations of Graduates
308 Amfiteatru Economic
Book Review
The Statistical Monograph of the Bucharest University
of Economic Studies. 100 generations of graduates
Author: Nicolae Istudor (coord.), Emilia Gogu, Dumitru Miron,
Alexandru Isaic-Maniu, Ion Vorovenci
Mihai Korka
Bucharest University of Economic Studies, Romania
Please cite this article as:
Korka M., 2021. Book Review: “The Statistical Monograph of the Bucharest University
of Economic Studies. 100 Generations of Graduates”. Author: Nicolae Istudor (coord.),
Emilia Gogu, Dumitru Miron, Alexandru Isaic-Maniu, Ion Vorovenci. Amfiteatru
Economic, 23(56), pp. 308-316.
DOI: 10.24818/EA/2021/56/308
With a history dating back more than a century, the Bucharest University of Economic
Studies (further on also referred to as ASE) is passing through a stage when its institutional
maturity and its valuable contribution to the Romanian higher education system are widely
acknowledged. On the other hand, the university is currently facing the challenges
characteristic of these turbulent times, when numerous factors are generating uncertainty
about both the present and the future. Within such a context, the concerns related to
understanding the specific trends that characterise the evolution of a university and to
designing sustainable development strategies are naturally intertwined with a genuine
interest in the history of the institution. Undoubtebly, the past academic experience and
events cannot be brought back to life and mechanically integrated into the current university
practices, but they can help us better assess the present and future opportunities and risks,
identify ways and means to surpass the challenges of the present and welcome the future
with a proactive attitude. An analysis of the institutional history can be an invaluable source
of wisdom when dealing with the challenges of the present and the uncertainties of the
future.
In an interview given in the spring of 2019, Ioan Aurel Pop, President of the Romanian
Academy, was saying that „Historical knowledge cannot help us guess the future, but it can
help us better manage the present and prepare the future... One may say that the past is less
valuable for the future than the present, but we should not forget that the present inevitably
becomes past and that, without noticing it, the future becomes present and then past. If we do
not value the past... then we do not value the life that is coming after us. [Apud Korka, M.
(2019) Un secol de design curricular în învățământul superior comercial din România. In
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 309
English: Centennial Curriculum Design in Romanian Commercial Higher Education, ASE
Printing House, Bucharest, p. 18].
Animated by the belief that there is a genuine need for well documented history, the rectors
of our university have encouraged historians and economists to research the past of the
university and to bring back to life events and personalities that marked the development of
our institution. The ASE Printing House has ensured the publication of these works at a
high graphic quality.
The Statistical Monograph of the Bucharest University of Economic Studies.
100 generations of graduates (further on referred to as the Statistical Monograph)
completes the series of works published in 2013 on the occasion of the university’s
centenary. Conceived as a chronicle in figures of the evolution of ASE, it proves to be a
dynamic portrait which documents the development of the institution from the moment it
was founded until the present day.
The monograph of a complex institution such as a university involves detailed research into
the multiple aspects of its operation. Such a work is usually accompanied by documents,
testimonials, photographs and other relevant types of information. The Statistical
Mongraph comprises all the above, together with pertinent statistics presented in the form
of complete series of data on the aspects that used to define the functioning of the
organisation over certain periods of time.
The first generations of professors of the Academy of Higher Commercial and Industrial
Studies (nowaday’s ASE) cherished genuine transparency and published, sometimes
annually, other times every 4-5 years, statistical reports regarding the institutional activity,
without hesitating to include the shortcomings alongside the achievements.
The tradition of publishing statistics on the activity of every higher education institution in
Romania was interrupted by the 1948 education reform, imposed by the political interests
of the new regime. Since that time, the publication of statistical data was restricted, leading
to a lower level of transparency, while allowing the political power to intervene at its own
will in the activity of the institutions which were part of the system. As an effect of this
reform inspired by dogmas and practices which had nothing to do with the traditions of the
Romanian education system, the Academy of Higher Commercial and Industrial Studies
was confronted with a change in its educational mission. Curricula had, thus, to be
reformulated, in order to include the political education of the future graduates, academic
staff were severelly purged, and the institutional structure became unstable. Moreover,
between 1948-1953, our institution was no longer allowed to organise doctoral studies, in
spite of its tradition and the good reputation it enjoyed within Romanian society.
Unfortunately, the concern for publishing statistical data regarding the activity of the
university has remained sporadic even in the 30 years which followed after the downfall of
the communist regime in December 1989.
It is within this context that one should understand the authors’ objective of presenting and
commenting upon a collection of annual statistical data on the functioning of our university
during the timespan of the first 100 generations of graduates (from the first enrolment in
1913 to the promotion which graduated in 2019). It was only due to the perseverance and
professionalism of a small group of experts in documentary and archival research that the
series of statistical data could be recovered and analysed, thus creating the premises for a
better understanding of our institutional development.
AE Book Review: The Statistical Monograph of the Bucharest University of Economic Studies. 100 Generations of Graduates
310 Amfiteatru Economic
It is often said that unwritten facts do not go down in history. In the case of our university,
the facts were systematically recorded, but more often than not they remained unknown to
members of the academic community and the public at large. This is why we are stating
that only the facts which are shared with the public can be remembered. The Statistical
Monograph of the Bucharest University of Economic Studies offers insights into the
history of our university and, thus, allows us to acknowledge its value.
The sources of documentation for this monograph were varied. Laws, decrees, government
decisions, minister’s orders, regulations and other documents with an impact on the
existence and functioning of the university were identified in the State Archives and the
Archives of the Bucharest University of Economic Studies. Besides these, there were
minutes of management meetings, documents prepared by the administrative bodies upon
request from the rector’s office, annual reports published periodically by the heads of the
institution, graduates’ testimonials, anniversary volumes edited by ASE or by some of the
faculties and departments. Extracting the information from all these sources and organising
it as series of data represented a laborious process, characterised by perseverance and
attention to detail. When the documentary sources included divergent data, the monograph
authors preferred to use comparative analysis, thus allowing the reader the freedom to
assess the value of the information.
The documentary material collected, processed and analysed by the authors is presented on
1170 pages, organised into 12 thematic chapters, grouped in two volumes, so as to
consistently present the multi-faceted transformations which took place during the
107 years covered by the monograph.
From a methodological perspective, the unitary approach is ensured by the fact that the
thematic chapters reflect the three distinctive stages in the history of ASE – the Bucharest
University of Economic Studies. These stages have different characteristics given by the
social and political environment in which the university operated, as well as by its publicly
assumed mission and by the way in which it was managed:
- The first period (1913-1948) was charaterised by the human, financial and material efforts
to build and consolidate the good reputation of the university. After experimenting for one
year with a single curriculum, the university leaders became interested in conceiving and
implementing a diversified educational offer, adapted to the requirements of an economy
and a state administration facing the effects of modernisation and internationalisation. It
was a time when outstanding intellectual personalities were attracted in order to form and
develop an elite teaching staff. This was the time when the Palace of the Academy of
Higher Commercial and Industrial Studies was erected and extended and when the labs
were equipped with specific teaching facilities.
- The second period (1948-1989) was characterised by the politicisation of the educational
content and function, the replacement of the traditional curriculum meant to train
economists with a variety of professional profiles with curricula that educated the
undergraduates in a political spirit and offered an increasingly vague professional
specialisation. The educational reform of 1948 was systematically followed by invasive
measures in the activity of higher education institutions translated into the loss of the
decision making autonomy, political control over the composition of the teaching staff and
over the teaching and research activity.
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 311
- The third period started after the downfall of Ceausescu’s regime in December 1989.
After a quick depoliticisation of the teaching activity, there followed years of concern
related to aligning the curriculum to the modern practices of similar higher education
institutions in the Euro-Atlantic area. In 2004-2005, the Romanian higher education in the
economic field took its first steps towards implementing the Bologna principles and
specific instruments. The Bucharest University of Economic Studies – co-founder of the
Romanian Association of Economic Faculties – played a decisive role in harmonising the
procedures for implementing the recommendations agreed at ministry level into the practice
of economic and business administration faculties with a view to creating the European
Higher Education Area.
The first volume of the Statistical Monograph starts with the presentation of documents,
historic data and statistical information regarding the founding, consolidation and
development of the higher education institution and
its campus throughout the three above-mentioned
periods of time. Besides showing how the material
infrastructure of the teaching and learning activities
developed, the book highlights aspects related to the
organisational structure of the unversity, the social
and medical support offered to the undergraduates
(student accommodation, canteens, medical centre),
as well as the editorial activities for teaching
purposes). Another aspect covered is the
dissemination of the results of original research
carried out by members of the teaching staff in
information bulletins and university annals. The
introductory chapter ends with the presentation of
the current ASE University Charter and with a
detailed quantitative analysis of the various facets in
their inter-relatedness which define our university
the way it is today.
The second chapter deals with the dynamics of the institutional organisation by faculties,
specialisations and departments correlated with the evolution of the academic offer and the
admission to university throughout the three historic stages. The wealth of statistical
information and organisational details regarding the education process is an invitation to
reflection addressed to anyone interested in the development strategies of higher education
institutions. Our university has found solutions in a variety of historic, social and economic
contexts, managing to survive during the first World War, to develop resilience during the
communist regime, to regain its identity and enhance the quality of its performance after the
December 1989 Revolution.
Restoring the lists including the teaching staff is the object of another chapter. The idea
started from a very special piece of writing – Report for the years 1913-1918. This was
published in 1918 and can be considered the first collection of statistical data of the
Academy of Higher Commercial and Industrial Studies, distributed to the public at large.
At the end of this document, there is a complete list with the teaching staff of the first five
academic years. Next to each teacher’s name there is the following information:
specialisation upon graduation; the field of the doctoral studies; the institution which
AE Book Review: The Statistical Monograph of the Bucharest University of Economic Studies. 100 Generations of Graduates
312 Amfiteatru Economic
awarded the PhD title; the document testifying the appointment at the Academy of Higher
Commercial and Industrial Studies. The General Ledger (Cartea Mare) put together by the
university’s Accounting Office between 1955 and 1958 was another source of inspiration
for listing and analysing all the full time staff of the Bucharest University of Economic
Studies over the last decades. The information is accompanied by short analyses of the
teaching staff dynamics, as well as by insightful graphs comparing ASE’s teaching staff
with that of other Romanian tertiary education institutions.
Another chapter tackles the dynamics of the managerial staff at university, faculty and
department levels, while also analysing the structure of the teaching staff by teaching
position and the ratio teaching staff-auxiliary and administrative staff. The synoptic tables
and statistical graphs annexed to this chapter make it easy to draw comparisons both among
the faculties and the various academic years. They highlight how the employment and
promotion principles and practices differed substantially during the three historic periods
included in this Statistical Monograph.
The second volume of the Statistical Monograph is dedicated to the students and graduates
who stand proof of the influence of the university on the national and international economic
and scientific environment. The statistical research in this volume follows the structure of the
first volume, organised around the three historic periods mentioned above. The statistical
tables and graphs are the result of perseverent work
for the reconstruction of the series of data. Due to the
diversity and simplicity of the information provided,
they can also be perceived as an invitation to further
in-depth analyses of the dynamics of the student
population at university level. According to the
authors of this volume, throughout its existence, our
university has contributed to shaping a significant
number of professional destinies, the cumulated
number of graduates who obtained a bachelor degree
reaching 237831, of which 8515 graduated between
1918-1948, 85099 between 1948-1989 and 144217
between 1990-2019. The number of master
programme graduates between 1995-2019 amounted
to 63026. The total number of doctoral study
graduates was 5046, structured as follows: 156
between 1913-1948, 917 during the 1953-1989
period and 3973 between 1990-2019. Besides these data regarding university education
during the 107 years of existence of our university, the volume offers information about the
tens of students who are enroling annually in postgraduate and postdoctoral advanced
research programmes organised by the Bucharest University of Economic Studies.
In the first two chapters of this volume offer an impressive collection of time series and
statistical distributions which are relevant for the multi-criteria analysis of the student
population observed as entities grouped around faculties and study programmes.
The third chapter is dedicated to presenting and commenting on the statistical data referring
to the students who attended various forms of postgraduate programmes. It is only natural
that starting with the 2005-2006 academic year, the focus is on the students enrolled in the
master programmes offered by the Bucharest University of Economic Studies. The statistics
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 313
presented highlight the fact that after an exuberant start, with an excessive number of
master programmes, the faculties reached a maturity stage, promoting those programmes
which had proven competitive on an educational market in full expansion. The institutional
reputation, the quality of the teaching-learning process and novelty of the programmes were
the main factors which determined the streamlining of the offer for postgraduate studies.
The legal framework of doctoral studies, the regulation for organising them and the statistics of
doctoral students, doctoral graduates and doctoral coordinators are the subject of the fourth
chapter. For the first time in the history of information made available by our university to the
public at large, the fifth chapter presents statistics by academic years, alongside complete lists
of the doctoral students who successfully defended their theses, starting with the first one in
1924 and ending with the list of doctoral titles in economic sciences awarded in 2019.
The next two chapters – the sixth and the seventh – contain unprecedented information
regarding the cultural, sports and social associative activities which have accompanied the
teaching-learning process throughout the whole existence of our university. Beyond the
statistical summaries and documentary evidence of the presence of these elements in the
life of the academic community, the reader can also find here information referring to the
institutional regulatory framework which has facilitated the associative actions with a
beneficial effect on campus life during all the historic periods analysed.
In addition to this, the eigth chapter of the Statistical Monograph offers an overview of the
scientific, political and economic personalities from Romania and abroad, whose
contributions to the development of economic sciences and the progress of economic higher
education were acknowledged at academic level between 1913 and 2019.
Through its content, The Statistical Monograph of the Bucharest University of Economic
Studies. 100 generations of graduates is sending a threefold message to its readers: it is a
homage brought to both the students and the teachers who have animated the university
theatre halls, study rooms and libraries over more than 100 years; it is a reference point in
understanding the mission and responsibilities of the present time and, above all, it is a call
to the future generations of students and teachers to honour the memory of the past and to
continue the ancestors’ endeavour to increase the competitiveness and attractiveness of the
institution and to have its value recognised at national and international level.
AE Book Review: The Statistical Monograph of the Bucharest University of Economic Studies. 100 Generations of Graduates
314 Amfiteatru Economic
About the authors
A team of the teaching staff of the Bucharest University of Economic Studies – BUES (in
Romanian: ASE) has produced this impressive work concerning the history of our
university:
Professor NICOLAE ISTUDOR, PhD
The Bucharest University of Economic Studies,
Romania
E-mail: [email protected]
Nicolae Istudor – coordinator of the team of authors – is Rector of ASE since 2016,
He was reelected by the university senate for a second term (2020-2023). He is a 1993
graduate of the ASE Faculty of Economics and Management of Agricultural and Food
Production. He got in 1999 his PhD degree in economics at ASE. Currently he is tenured
professor at the Department of Agrifood Economics and Environment. Areas of excellence:
Logistics of Agrifood Companies, Agrifood commodities exchanges, Regional and rural
development, University management. He is a fervent supporter of those who are writing
and publishing studies and thematic monographs which bring to the consideration of
younger generations personalities and defining events for the history of Alma Mater of the
Romanian higher education in economics and business administration. For his professional
and academic merits the President of Romania decorated him in 2004 as Chevalier of
National Order of Faithful Service. He is holder of the UN Food and Agriculture
Organization’s 2010 Diploma for merits in the agrifood sector, as well as of the 2010
Excellence Diploma of the Romanian Society of Statistics for university management
performance.
Amfiteatru Economic recommends AE
Vol. 23 • No. 56 • February 2021 315
Professor EMILIA GOGU, PhD
The Bucharest University of Economic Studies, Romania
E-mail: [email protected]
Emilia Gogu is a 1995 graduet of the ASE Faculty of Cybernetics, Statistics and Economic
Informatics. PhD degree in economics at ASE in 2002. Currently she is full time professor
at the Department of Statistics and Econometrics. Areas of excellence: Statistics,
Macroeconomic statistics, Econometrics, Quality assessment of higher education. Her
mastery in applying statistical methods in the reconstruction of historical time series is fully
proven by personal involvement in the last 4-5 years in the publication of a series of
encyclopedic works dedicated to the Romanian higher education and especially to the
higher education in the fields of economics and business education.
Professor DUMITRU MIRON, PhD
The Bucharest University of Economic Studies, Romania
E-mail: [email protected]
Dumitru Miron is the President of the university’s academic senate. He is a 1981 graduate
of the ASE Faculty of Commerce. PhD degree in economics at ASE in 1993. He is tenured
professor at the Department of International Economic Relations. Areas of excellence:
Economics of international trade, Economics of European integration, Quality
mamanement in higher education, University management. In 2020 he received the
Romania’s Academy Virgil Madgearu Award for the 2018 published book „100 years of
Romania’s foreign trade”. He is the co-editor of the book „Virgil Madgearu – the professor,
the man of the city, the doctrinaire” which was published in 2020 by the BUES Printing
House – Editura ASE. This book brings back to the attention of the academic and scientific
circles the great personality of this professor who was teaching in our university between
1916 and 1940 and is considered to be a model of academic excellence for his
contemporaries and for the present and following generations.
AE Book Review: The Statistical Monograph of the Bucharest University of Economic Studies. 100 Generations of Graduates
316 Amfiteatru Economic
Professor ALEXANDRU ISAIC-MANIU, PhD
The Bucharest University of Economic Studies, Romania
E-mail: [email protected]
Alexandru ic-Maniu is a 1971 graduate of the ASE Faculty of Economic Computation and
Economic Cybernetics. PhD degree in economics at ASE in 1977. He is an emeritus
professor of the BUES Department of Statistics and Econometrics and a principal
researcher at the Centre for Industry and Services Economy of the Romanian Academy’s
National Economic Reserach Institute. Areas of interest: Statistics, Products’ quality and
realiability statistics, Business management statistics, Small and medium sized company
statistics, Romania’s national economy. In 1996 he received the Romanian Academy’s
Petre S. Aurelian Award for the 1994 published book on „Statistical problems of products’
reliability”.
Professor ION VOROVENCI, PhD
The Bucharest University of Economic Studies, Romania
E-mail: [email protected]
Ion Vorovenci is a 1992 graduate of the Bucharest University’s Faculty of history. PhD degrre
in economics at ASE in 2000. He is associate professor at the ASE Department of Philosophy
and Scio-Humanistic Sciences. Areas of excellence: History of the world economy,
Comparative history of the Romanian economy, Histroy of small industrial companies, Histroy
of the Academy for Higher Commercial and Industrial Studies in Bucharest.
More than 70 consultants working in the faculties and functional services of BUES have
offered advice to the authors in the design and writing of the Statistical Monograph.