15

1. Introduction

A consistent number of empirical works have shown the existence of a significant positive relationship between growth and innovation, especially stressing the role of innovation for economic growth, both at the macroeco-nomic and microeconomic levels (see Griliches, 1995; Brynjolfsson - Yang, 1996; Black - Lynch, 2004; Lipsey - Carlaw - Bekar, 2005; Pianta - Vaona, 2007; Hall - Lotti - Mairesse, 2009; Antonioli - Mazzanti - Pini, 2010, and Leoni, 2012 to cite a few). In these studies, the fact that innovation – not only technological, but also organisational1 – is an important driver of growth and competitiveness, of both countries and firms, emerges as a general re-sult, which holds true across different geographical contexts and industrial sectors. However, the same result is usually obtained, and theoretically dis-cussed, on the background of «normal» macroeconomic conditions, that is, without controlling for those severe global downturns which cyclically inter-rupt long-term growth trends (e.g. Reinhart - Rogoff, 2008; Lucchese - Pi-anta, 2011).

The severity and pervasiveness of the economic recession stimulated an analysis of the relation between innovation and firm resilience to the crisis challenges. For the Italian case, the state is even more peculiar than that of other developed countries: the economic crisis hit a stagnant economy, which had shown a rate of growth of the GDP next to zero for the period 2001-

Received March 2011, Approved January 2013.

We gratefully acknowledge the financial support of the Emilia-Romagna Region, the useful comments of the participants to the XVI AIEL Conference in Milan (15th-16th September 2011), to the SIE Conference in Rome (14th-15th October 2011), to the EAEPE Conference in Vienna (27th-30th October 2011) and the remarks and suggestions of two anonymous referees. All usual disclaimers apply.

1 See the recent literature discussed in the survey by Leoni (2013).

Innovation Strategies and Economic Crisis: Evidence from Firm-level Italian Data

by Davide Antonioli, Annaflavia Bianchi, Massimiliano Mazzanti, Sandro Montresor and Paolo Pini

ECONOMIA POLITICA / a. XXX, n. 1, aprile 2013

16

2008, with a brief positive interval in the period 2006-2007. The decline in productivity growth has opened a debate on its determinants (Faini - Sapir, 2005; Brandolini - Bugamelli, 2009). Several structural factors have been called into question to explain the stagnation of the Italian economy, ranging from the insufficient competitiveness of the Italian economic system, to the small size of the Italian firms, or the shortage of infrastructures or the ex-cessive rigidity of the labour market. Indeed, diverse determinants co-exist, leading to the widening of the gap between Italian economy and other de-veloped countries. The capacity of the economic actors, such as the firms, to react actively to the challenges brought by the crisis deeply influences the capacity of the whole economy to recover. For such a reason, it is of extreme importance to study Italian firms reaction to the crisis, with a special focus on the field of innovative activities as a response to the crisis.

A sample of 555 manufacturing firms located in the Emilia-Romagna region were surveyed, in order to collect information mainly concerning in-novative strategies, coupled with several other information about firm level industrial and labour relations and working conditions (Antonioli - Bianchi - Mazzanti - Montresor - Pini, 2011). The regional level choice is primarily motivated, because the regional nature of innovation strategies has been recognised as relevant, by the recent literature on the Regional Innovation Systems (RIS), because the role of regional subsidies to sustain innovation projects has grown for Emilia-Romagna firms in the last decade and since Emilia-Romagna is one of the two most innovative regions (with Lombardia) in Italy (Hollander-Tarantola-Loschky, 2009). Moreover, a rebalanced empha-sis that explicitly includes the role of «Regions» as drivers of development and growth is present in the re-launching of the redefined Lisbon agenda. Thus, EU growth policy is moving the attention from economic sectors and consequent sector specialisation, to a more balanced perspective that ac-counts for regional/sector specialisation. This implies a full recognition of regional idiosyncratic elements, such as industrial relations, local policy, insti-tutions and networks with specific characteristics that allow to distribute the risk of the crisis among all network members2 (Cainelli - Montresor - Vit-tucci, 2012).

2 The EU policy scenario is currently holding more attention and focus on regional growth, rather than country convergence, to fulfil a long-run economic development. It is, thus, based more on attention to within country heterogeneity of «regional economic stories». The peculiarities and idiosyncratic features of EU regions have been recognised to be a source that has to be empowered. The new «Smart Specialisation Strategy» based policies take in fact Regions as main actors of development, in line with the consolidation and suggestions coming from «economic geography» and geography of innovation realms (Foray - David - Hall, 2011). This new policy emphasis on spatial and regional issues for the future economic development of Europe is a pillar of strategies that aim at achieving 20-20-20 goals and future cohesion policy targets from 2014 onward (see the DG-REGIO paper by Barca - McCann, 2011). The EC DG JRC IPTS will support implementation of smart specialisation strategies that place regions at the heart of EU economic growth.

17

The work is organised as follows. The second section provides a brief overview of the background literature, which is useful in defining the re-search hypotheses. The third section reports the data and the methodology used in the empirical part. The fourth section provides a discussion of the econometric results, while the last section is left to conclusions and remarks.

2. Research background and hypothesis

Several authors have pointed out the linkage between innovation types and fluctuations of the business cycle. Focussing attention on microeconomic behaviour, Giedeman - Isley - Simons (2006) distinguish two broad catego-ries of theories linking innovation to the business cycle: countercyclical (Hall - Blanchard - Fischer, 1991; Aghion - Saint-Paul, 1998) and procyclical ones (Stiglitz, 1993; Barlevy, 2007). The former stream of works considers the «opportunity cost effect»3 as a main engine to become more efficient dur-ing recessions. In periods of growth, firms have less incentives to reorganise the production process and to contrast inefficiencies, because the amount of foregone profits due to reorganisation activities could be very high. However, during recessions, when the demand for goods is low, such an amount of foregone profits is lower and it is more convenient to reorganise production and labour activities. According to this position, investments in organisational changes and training, which yield benefits over a long period of time (Aghion - Saint-Paul, 1998), are likely to be more countercyclical than other types of investments, such as R&D, that are cash intensive. The procyclical position focusses on the cyclicality of R&D activities. As an example, Barlevy (2007) shows how combining growth data and R&D data the hypothesis of procy-clicality of R&D is supported. The explanation lies both on the constrained financial markets during recessions, that make it more difficult for firms to secure enough funds to conduct the planned R&D activities (Stiglitz, 1993) and on the likely short-run perspective of firms and entrepreneurs (Barlevy, 2005). Entrepreneurs know that the main gains from an innovation accrue in the short run, before the competitors learn and innovate on the original idea, securing for them a new stream of short-run returns; hence, entrepreneurs will be reluctant to invest resources during recessions, especially if recessions are expected to last in time, since the short-term gains would be lower than during booms.

Taking the move from this point, we shift our attention to a relatively un-explored set of firm level relations, in a short term perspective, during a reces-sion period. As firms are the actors responsible for the introduction, implemen-

3 Other mechanisms leading to high levels of productivity are also countercyclical accord-ing to Aghion - Saint-Paul (1998): clearing up or lame duck effect; disciplinary effect and ex-ternality effect.

18

tation and diffusion of innovations, they are the fundamental (micro) actors through which innovation triggers or slows down (macro) economic growth and they represent our natural unit of analysis. More emphasis ought to be granted to the microeconomic perspective through the analysis of the strategies and behaviours of micro economic agents, instead of giving too much atten-tion to macro aggregates. In our analysis, the crisis and the fluctuations of the cycle become the background over which firms operate their strategic choices, according to their specific characteristics and pre-crisis strategies.

Our main aim is to investigate whether the more innovative firms before the crisis are also those more active in the innovation fields during the cri-sis, using innovation as a way to exit from the crisis and to cope with chal-lenges brought by the latter. In fact, the more innovative firms could have planned (before the crisis) investments in R&D and other innovation activi-ties they are reluctant to cut, even if demand drops. Moreover, it can be ar-gued that the more firms are active before the crisis, in terms of innovation activities, the better they are equipped to pursue innovation activities, also during the crisis. Although we do not directly aim to test the persistence of innovation and to answer the question whether or not «success breeds suc-cess», our research hypothesis could be inscribed at the margin of a stream of literature interested in verifying such a persistence over time (Peters, 2009; Raymond - Mohnen - Palm - van der Loeff, 2010). Indeed, we aim to verify the persistence of intense innovative activities at firm level, but with a short-run perspective and on the background of an economic recession looking at innovation reaction to the crisis. We can argue that there is a dynamic cor-relation over time among innovation strategies; thus, the first hypothesis is accordingly:

Hp1 Innovation activities before the crisis are related to innovative strat-egies to react to the crisis, pointing to a dynamic correlation of innovation activities (innovation calls for innovation).

A further aspect related to the linkage between innovation strategies con-cerns the issue of complementary innovation activities. Several works have investigated the effect of a joint adoption of managerial and organisational practices on outcome variables, ranging from innovation (e.g. Hujer - Radic, 2003; Bondarouk - Looise, 2005; Pini - Santangelo, 2010; Lynch, 2012) to firm’s economic performance (e.g. Caroli - Van Reenen, 2001; Janod - Saint - Martin, 2004). As for the organisation sphere, the same may hold for other innovation spheres, for which the issue of complementary activities is less explored: technology, information and communications technologies (ICT), environment and strategies of internationalisation. Complementary activi-ties and practices within each of such innovation spheres are likely to exist. Whether or not such complementary nature emerges, when we consider the reaction to the crisis as the output variable, is an unexplored issue, as far as our knowledge is concerned; hence, the following hypothesis is put forward:

19

Hp2a Jointly adopting several innovation practices in each innovation sphere increases the capacity to react to the crisis through innovative actions (complementarities within innovation spheres).

The complementary nature of innovation activities is not limited to the boundaries of each innovation sphere (within sphere), rather, it is likely to hold also between innovation spheres. Even in this case, the literature mainly focusses on the impact of innovation complementarities on firm economic performance. The prevalent message stemming from the empirical contribu-tions (see among others Aral - Weill, 2005; Laursen - Foss, 2003) is that a consistent innovation strategy, whose building blocks are complementary technological and organisational activities, leads to productivity gains. The «productivity paradox» described by Brynjolfsson - Hitt (2000) is an illumi-nating example of what may happen when firms adopt innovations (ICT, in that case) without introducing complementary organisational changes that al-low for the full exploitation of the ICT, formerly introduced. Following Pini - Santangelo (2005), who clearly recognises in labour, organisational practices and innovative activity of mutually feeding phenomena, as organisational practices provides «the flexibility which allows firms to promptly respond to the drastic pace of technological change [and] similarly, innovative activ-ity feeds labour organisational practices by requiring greater organisational flexibility» (Pini - Santangelo, 2005, p. 254), we can argue that innovating in more than one sphere does provide the firms with the necessary capabilities, skill base and flexibility needed to cope with the crisis, through an intense and diversified innovation reaction. Accordingly, the following hypothesis is proposed:

Hp2b Jointly innovating on different innovation spheres increases the ca-pacity to react to the crisis through innovative actions (complementarities be-tween innovation spheres).

A third set of firm level relations deals with the role of industrial rela-tions on the capacity of firms to innovate as a reaction to the crisis. As put forward by several authors, both from theoretical (Metcalf, 2003; Menezes - Filho - Van Reenen, 2003) and empirical (Boheim - Booth, 2004; Blundell - Griffith - Van Reenen, 1999; Della Torre, 2009) points of view, the impact of union presence and activity on firm innovations is not univocal. The link-age between industrial relations and innovation has to be tested case by case, and in recent papers (Antonioli, 2009; Antonioli - Mazzanti - Pini, 2010) we provided evidence of an overall positive link between good quality (par-ticipative) industrial relations and innovation activities, especially on the or-ganisational sphere4. Participative industrial relations contribute to a win-win

4 It is worth stressing the positive role that cooperative industrial relations also have for

20

strategic behaviour between management and unions, so that common goals and gains are pursued within a non-adversarial climate, favouring innovation activity. On the basis of the above consideration, we may argue that a «good climate» between management and unions is crucial in times of economic crisis for containing costs and improving efficiency, mainly through changes in production process and labour organisation, which need to be shared by a large number of employees.

However, the traditional model of indirect participation has been chal-lenged in recent years by the introduction of new human resource manage-ment practices, which delegate more responsibility and autonomy to the em-ployees (Santangelo - Pini, 2011). A model of direct participation, without the mediating role of unions, has grown side by side with the indirect par-ticipation model, questioning the relevance of union representatives at firm levels and lowering their appeal: management practices addressed to increase employees direct participation may increase job satisfaction and job empow-erment as well, reducing the scope for firm-level unions representatives, since the latter could become redundant, according to the employees (Antonioli - Pini, 2005; Machin - Wood 2005; Pini, 2005). Thus, it might be the case that a more direct participation model substitutes for the indirect one, making the relation between management and employees more relevant in sustain-ing innovation activities during the crisis than the relation between firm-level union representatives and management. This leads to the following two hy-potheses:

Hp3a A good quality industrial relation climate before the crisis is posi-tively related to the adoption of innovation strategies to overcome the crisis’ challenges, since a high level of participation is functional to adopt innova-tion activities that are shared by a large amount of workers («smoothly» im-plementable);

Hp3b A direct participation model overshadows the role of the tra-ditional indirect participation, emerging as the leading industrial relations driver of an intense innovation activity to tackle the challenges brought by economic downturns.

Finally, the last set of firm level relations concerns firm economic per-formance and innovation activity. This issue has attracted the attention of sev-eral scholars, interested in evaluating the impact of innovation on firm per-formance (see among others Cappelli - Neumark, 2001; Zwick, 2004, 2005; Arvanitis, 2005; Black - Lynch, 2001; Janod - Saint Martin, 2004). However, for the aim of the present work it is appropriate to reverse the question: are

workers’ well-being in Emilia-Romagna manufacturing firms, as pointed out by Antonioli - Mazzanti - Pini (2009, 2011).

21

the best performers before the crisis also those that intervene more inten-sively on innovation activities during the crisis? The reversed question is of relevance in the light of the credit crunch «imposed» by the financial crisis to economic agents. Because certain types of innovation, especially activities linked to product innovation, which usually rely on R&D investments, are cash intensive, we expect a positive link between performance and intensity in reaction to the crisis through innovation strategies. The «accumulation ef-fect», driven by a successful past that results in high levels of economic per-formance before the crisis, may influence the capacity to innovate during the crisis. This dynamic relation also results in a process of self-selection of the high performers as the group of most reactive firms5. The fourth hypothesis can be formulated as:

Hp4 The economic performance in the pre-crisis period is positively linked to the reaction to the recession through innovative action.

3. Empirical framework

In order to test our research hypotheses, we base the analysis on micro-level data coming from a unique dataset concerning a sample6 of 555 Italian manufacturing firms with at least 20 employees located in Emilia-Romagna region (tab. 1), NUTS 2 level of analysis, which counts the 4th regional GDP in Italy7 and creates the 3rd regional industrial value added, counting around the 7% of the Italian population8. Moreover, the region is one of the two most innovative regions in Italy, with Lombardia, as stated by the Regional Innovation Scoreboard (Hollander - Tarantola - Loschky, 2009), which clas-sifies Emilia-Romagna as a medium-high innovator region. In addition, the presence of industrial districts, mainly composed of small and medium sized enterprises strictly intertwined, well marked entrepreneurship spirits, conju-gated to a deep-rooted unionism and to common social and cultural values, constitute the elements of what has been called the «Emilian model» (Amin, 1999; Brusco, 1982). Such elements sum up to identify a Regional Innova-tion System (RIS), although with some peculiarities (Miceli, 2010) that lead

5 In a sense, the process underlying the firm decision to implement innovation activities with a high intensity, in order to react to the crisis mirrors the Melitz (2003) model on pro-ductivity and trade: in that model, the more productive firms are those that «self-select» as ex-porters, when exposed to trade, while the least productive firms are eventually forced to exit.

6 The random sample is stratified by size, province (geographic location) and sector. It is well representative of the population, showing only minor distortions (tab. 1).

7 Italian territory is subdivided into 20 regions.8 Data come from ISTAT (Italian National Institute of Statistics).

22

Evangelista - Iammarino - Mastrostefano - Silvani (2002) to define it as an informal learning system9.

The original set of information we use was collected through phone in-terviews with the firm managements in 2009, and was based on a structured questionnaire, subdivided into several sections: the firm in the crisis, organi-sational innovation, training, technological innovation, environmental inno-vations, ICT, internationalisation, firm performances and firms descriptive characteristics. Almost the totality of the data refers to the pre-crisis period (2006-2008), unless those concerning firm economic actions (innovations) adopted to react to the crisis, which refer to the 2009. Finally, as expressed

9 See also, Marzucchi - Antonioli - Montresor (2012), on the relationships the Emilia-Ro-magna manufacturing firms develop with research organisations and universities located both within and outside the regional borders.

Tab. 1. Population and sample distribution (%) by sector and size

Population distribution (%) Size Total Total(a.v.)

Sector 20-49 50-99 100-249 250+

Food 5,65 1,94 1,16 0,64 9,39 382textile 6,17 1,47 0,71 0,37 8,73 355wood, paper and other industries 7,79 1,67 0,79 0,42 10,67 434chemical and rubber 5,01 1,87 1,11 0,42 8,41 342non metallic mineral products 3,81 1,23 1,18 0,79 7,01 285metallurgy 16,99 3,29 1,18 0,25 21,71 883machinery 21,44 6,37 4,06 2,24 34,10 1387Total 66,86 17,85 10,18 5,11 100,00Total (a.v.) 2720 726 414 208 4068

Sample distribution (%) Size Total Total(a.v.)

Sector 20-49 50-99 100-249 250+

Food 2,88 3,78 1,62 0,54 8,83 49textile 2,70 1,44 1,62 0,54 6,31 35wood, paper and other industries 3,60 2,88 1,08 0,90 8,47 47chemical and rubber 3,78 3,42 1,80 1,08 10,09 56non metallic mineral products 1,62 2,16 1,62 2,16 7,57 42metallurgy 8,83 5,77 2,16 0,18 16,94 94machinery 14,05 15,32 7,39 5,05 41,80 232Total 37,48 34,77 17,30 10,45 100,00Total (a.v.) 208 193 96 58 555

Cochran TestMargin of error q*

Interviewed firms vs. Population

( )N nN

N1 11

i-

--

0.039

Margin of error q «usually» tolerated: 0.05. Restrictive test for small population: the smaller is N, the lesser the distance between N and n has to be in order to generate an acceptable q.

23

by a version of the Cochran test (tab. 1) for sample distortion, we have ac-ceptable results.

The survey data have been merged with balance sheet data coming from the Bureau Van Dijk-AIDA database. The time span covered by such data refers to the period 2006-2008. The choice of the period was made in or-der to homogenize the time of the accounting variables with the survey ones (see section 3.1). The firms covered by balance-sheet data for all of the three years, are significantly less than the 555 surveyed firms (tab. 2) and appropri-ate methods were used to solve the missing data problems in the empirical analysis. The chosen accounting variables, labour productivity (VA/EMP), profitability (PROFIT/EMP) and cash flow (CASHFLOW/EMP) per capita, aim to provide a picture of the overall performance for the period 2006-2008 that can, potentially, affect the capacity to react to the crisis with innovative activities.

3.1. Empirical model and variables

Provided that we have at our disposal a wide range of information and given the considerations reported in the second section, it is possible to set up the reduced form model as follows in order to test our research hypoth-esis:

(1) ACTION_INNOi,2009 = a0 + a1i,2006-2008(FIRM_SPEC) + + a2i,2006-2008(INNO) + a3i,2006-2008(INDREL) + a4i,2006-2008(PERF) + ei,t

where i identifies the single firm, and 2009 and 2006-2008 represent the pe-riods considered, FIRM_SPEC identifies the vector of firm-specific charac-teristics, INNO is the vector of innovation variable, INDREL encloses the industrial relations aspects and PERF is a vector of performance variables. A full description of all the covariates is provided below.

The convenient specifications to test the single hypotheses are reported in Section 4 where the results are discussed. The main econometric issues are tackled in the following way.

As far as the endogeneity problem due to simultaneity is concerned, it is possible to mitigate such a problem by exploiting the diachronic nature of the variables involved in the regressions, since the variables on the left-

Tab. 2. Number of firms covered by balance sheets data for each year of the period 2006-2008

Variable* 2006 (%) 2007 (%) 2008 (%)

VA/EMP 436 (78.5) 436 (78.5) 399 (71.9)PROFIT/EMP 433 (78.0) 436 (78.5) 390 (70.3)CASHFLOW/EMP 436 (78.5) 436 (78.5) 402 (72.4)

* See next section for a full description of the data.

24

hand side are measured on a different period, with respect to those on the right-hand side (Michie - Sheean, 2003). We are aware of the fact that test-ing our hypothesis amounts to testing dynamic correlations in a multivariate framework, rather than causal relationships. As for the unobserved hetero-geneity issue, we contend that the high number of firm structural variables coupled with relevant covariates capturing managerial attitudes allow to deal with large part of firm specific heterogeneity (Huselid - Becker, 1996), often attributable to the different managerial attitudes10.

Firm specific characteristics (FIRM_SPEC)

The structural variables used as controls aim at capturing some firm specific elements and also peculiar characteristics of the production con-text (FIRM_SPEC). Within this group (tab. 3) we have the usual sector dummies, size dummies and also the geographical location of the «regis-tered office» of the firm, the «openness» to international markets provided by a variable capturing the percentage of turnover made on international markets and the belonging to a national or international group. It is also known whether or not a firm is a supplier, and the percentage of turno-ver made as a supplier. The workforce structure, in terms of manual and non-manual workers, is another element potentially influencing both the absorptive capacity of the firm and its innovative propensity and, as such, is included among the controls. Such controls are thought to be partially exogenous elements influencing the propensity to innovate. Coupled with such controls, we also use other firm specific variables, strongly related to the strategic decisions of the firms, that help in explaining the propensity to adopt more or less intense innovative intervention. Indeed, we classify the firms in terms of their pre-crisis behaviour and of their propensity to prefer «active» policies to be implemented by policy makers to overcome the criticalities brought by the recession11.

Another set of less usual controls, which captures firm specific aspects, is given by the indexes concerning workers conditions. The higher the in-dex, the better is worker welfare, along several dimensions of the working environment (worker effort, economic incentives, degree of autonomy and responsibility, injuries, etc.). If worker welfare is high, it is more likely to ob-serve satisfied employees, which might translate into more productive and innovation-oriented employees (e.g. Bartel - Ichniowski - Shaw, 2005).

10 In synthesis, although we acknowledge the potential presence of endogeneity, we do not directly address the issue here. It certainly remains as an issue to cope with in future de-velopments of the analysis, related to the linkage between innovation and firms’ economic per-formances.

11 Information on this kind of propensity is given by specific questions in the question-naire. They are available from the authors upon request.

25

Finally, we also know whether or not each firm received subsidies to sup-port innovative programmes in the past (2004-2006), mainly related to R&D

Tab. 3. Construction and descriptive statistics of firm specific characteristics

Construction Mean Min Max

FIRM SPECIFIC CHARACTERISITCS (FIRM_SPEC)

PAVITT SECTORS (d) Dummies (5) identifying the sectors the firm belong to on the base of the OECD-Pavitt taxonomy

/ 0 1

GEO (d) Dummies of geographical location of the firm: NUTS 3 territorial units (9 provinces + extra regional) were grouped in 5 clusters

/ 0 1

SIZE (d) Size dummies by employees: 20-49; 50-99; 100-249; > 249

/ 0 1

EXPORT Percentage of turnover made on international mar-kets

0.33 0 1

GROUP_INTERNAT (d) Dummy: 1 firm is part of an international group; 0 otherwise

0.07 0 1

GROUP_NAT (d) Dummy: 1 firm is part of a national group; 0 other-wise

0.23 0 1

SUPPLIER Percentage of turnover made as supplier 0.28 0 1SKILL_SHARE Share of non-manual workers 0.85 0 5.1PROACTIVE Dummy variable: 1 if the firm is active in terms of

strategic innovation behaviour (strongly active on the innovation activities before the crisis and willing to see policies supporting training, innovation and human capital accumulation or policies directly ad-dressed to sustain the internal aggregate demand); 0 otherwise

0.40 0 1

DEFENSIVE Dummy variable: 1 if the firm is defensive in terms of strategic behaviour (strategic difficulty in front of competitors, especially from BRIC countries and/or structural «distress» for high labour, production and financial costs coupled with a willingness to see poli-cies aimed to cut labour costs through a reduction of taxation); 0 otherwise

0.14 0 1

MIX Dummy variable: 1 if the firm shows a mixed behav-iour in terms of strategic innovation behaviour (share both PROACTIVE and DEFENSIVE characteristics); 0 otherwise

0.13 0 1

OTHER Dummy variable: 1 if the firm shows neither PROAC-TIVE nor DEFENSIVE behaviours; 0 otherwise

0.33 0 1

WORK_COND_P Trend in working conditions focused on positive as-pects (workers effort; employees competencies; avail-able information on the production process for the employees; employees autonomy and control on their tasks; economic and non economic incentives) meas-ured on a 5 points scale going from –2 to +2 rescaled on the interval (0,1)

0.64 0 1

WORK_COND_N Trend in working conditions focused on negative as-pects (workload for single employees; job instability; rigidity of the working hours; diseases related to the job; work-related injuries) measured on a 5 points scale going from –2 to +2 rescaled on the interval (0,1)

0.56 0 1

INNO_SUB (d) Dummy: 1 firm has been publicly funded to support an innovative program 2003-2006; 0 otherwise

0.23 0 1

26

projects with specific characteristics12. The role of such subsidies may en-dure in time, leading the beneficiary firms to be more innovative, and better equipped to cope with the challenges brought by the recession.

Pre-crisis techno-economic profile of the sampled firms

The set of «core» covariates is composed of variables capturing past (2006-2008) a) innovation activities (INNO) of the firm in several spheres (technology, organisation, training, environment, ICT), as well as internation-alisation strategies, b) industrial relations quality (INDREL) as perceived by the respondents and c) economic performance (PERF) (tab. 4). Each subset of variables allows us to test the hypotheses listed in Section 2.

The first subset (INNO) of covariates is assumed to have relevant link-ages with the innovative activity in the crisis and it allows us to test Hp1, Hp2a and Hp2b. Specific and composite innovation indexes, as well as the interactions among the latter, are included in a first set of specifications, as fully described in Section 4. In so doing, we are able to depict whether the Emilia-Romagna industrial system copes with the crisis, widening the dis-tance between leading innovators and the other firms and, possibly, moving toward an even more pronounced regional specialisation.

With the second subset of core covariates (INDREL), we aim to test the expected positive linkage between cooperative industrial relations and the propensity to innovate, as has been demonstrated in our previous works (Antonioli, 2009; Antonioli - Mazzanti - Pini, 2010). In so doing, we are also able to test whether or not the indirect participation model is capable of pro-viding firms with the capacity to react more intensely to the crisis, or if it is the direct participation model to be more closely related to such a capacity (Hp3a and Hp3b).

Finally, with the last subset of covariates, given by performance indica-tors (PERF) for the period 2006-2008, which are constructed on the basis of balance sheet variables, we want to verify the relation between the pre-crisis economic performance and the capacity to robustly intervene with innova-tion during the crisis (Hp4).

Because of the number and nature of our covariates we, firstly, check for high levels of bivariate correlation, in order to exclude potential problems due to multicollinearity. The main relevant covariates do not show very high levels of correlations (tab. 5). However, the risk of multivariate collinearity cannot be excluded and a check for it is conducted in the econometric anal-ysis, through the use of the Variance Inflation Factor (VIF) indicator.

12 The regional subsidies have been provided in the framework of the PRRIITT program: Regional Program For Industrial Research, Innovation and Technology Transfer. For recent results on the impact of PRRIITT, see Antonioli - Marzucchi - Montresor (2013).

Tab. 4. Construction and descriptive statistics of innovation variable (period 2006-2008)

Construction Mean Min Max

INNO (2006-2008)Technological Innovation

INNO_TECH... Composite index of innovation intensity in the technological sphere. Values on the interval (0,1). Constructed on the basis of the following specific indexes:

0.22 0 0.60

...OUTPUT_TECH Index including innovation aspects belonging to the dimension of technological output

0.12 0 0.82

INPUT_TECH Index including innovation aspects belonging to the dimension of technological input

0.32 0 0.65

Organisational Innovation

INNO_ORG... Composite index of innovation intensity in the or-ganisational sphere. Values on the interval (0,1). Constructed on the basis of the following specific indexes:

0.26 0 0.75

...OUTSOURCING Index of outsourcing activities 0.11 0 0.8ORG_COLL Index of collaboration activities to carry out organ-

isational innovations0.20 0 1

PROD_PRACTICES Index as the average number of production organi-sation practices

0.48 0 1

LAB_PRACTICES Index as the average number of labour organisa-tion practices

0.25 0 1

Training

TRAIN... Composite index of intensity in training policies. Values on the interval (0,1). Constructed on the basis of the following specific indexes:

0.50 0 1

...TRAIN_TYPE Index of training typologies 0.42 0 1COV_PERM Percentage of permanent workers involved in train-

ing programs. Interval (0,1)0.38 0 1

COV_NONPERM Percentage of fixed-term workers involved in train-ing programs. Interval (0,1)

0.21 0 1

TRAIN_COMP Index of training competencies covered by train-ing programs (computing comp.; technical comp.; organisational/relational comp.; economic/legal comp.)

0.44 0 1

Environmental Innovations

INNO_ENV... Composite index of innovation intensity in the en-vironmental sphere. Values on the interval (0,1). Constructed on the basis of the following specific indexes:

0.13 0 0.89

...ENV_BEN Index of benefits due to environmental innovations (emission reduction, energy/material efficiency, CO2 reduction)

0.13 0 1

ENV_PROC Index of environmental innovation procedures (EMAS, ISO14001)

0.06 0 1

ICT

ICT... Composite index of innovation intensity in infor-mation and communication technologies sphere. Values on the interval (0,1). Constructed on the basis of the following specific indexes:

0.59 0 1

...INSTR_ICT Index of ICT instruments implemented 0.83 0 1SYS_ICT Index of ICT management systems implemented 0.29 0 1ACT_ICT Index of activities supported by ICT 0.69 0 1ROLE_ICT Index of types of role covered by ICT 0.55 0 1

Tab. 4. (follows)

Construction Mean Min Max

Internationalisation

INTERNAT... Composite index of internationalization activities. Values on the interval (0,1). Constructed on the basis of the following specific indexes:

0.08 0 0.59

...IDE (d) Dummy variable: 1 if foreign direct investments are done; 0 otherwise

0.16 0 1

IDE_TYPE Index of IDE typology 0.04 0 0.80IMPORT Dummy variable: 1 if the firm import intermediate

goods from abroad; 0 otherwise0.40 0 1

IMPORT_TYPE Typology of firms providing intermediate goods 0.12 0 1INT_PART Index capturing different typologies of interna-

tional participation0.02 0 0.83

INDREL (2006-2008)

UNION_INV... Composite index capturing the degree of union involvements constructed as the average of the fol-lowing three specific indexes of involvement:

0.28 0 1

...UNION_INF* Index: as average of union information about changes in the 6 innovation spheres

0.55 0 1

UNION_CONS* Index as average of union consultation about changes in the 6 innovation spheres

0.19 0 1

UNION_BARG* Index as average of union bargaining about changes in the 6 innovation spheres

0.07 0 1

EMP_INV... Composite index capturing the degree of employ-ees involvements constructed as the average of the following two specific indexes of involvement:

0.50 0 1

...EMP_INF Index as average of employees information about changes in the 6 innovation spheres

0.66 0 1

EMP_CONS Index as average of employees consultation about changes in the 6 innovation spheres

0.17 0 1

EMPINV_ORG Set of variables indicating the degree of employees involvement on each of the six innovations spheres: 0 no involvement; 1 informed; 2 consulted

1.06 0 2EMPINV_TRAIN 1.06 0 2EMPINV_TECH 1.07 0 2EMPINV_ICT 1.01 0 2EMPINV_ENV 0.92 0 2EMPINV_INTERNAT 0.86 0 2

UNIONINV_ORG* Set of variables indicating the degree of union rep-resentatives involvement on each of the six inno-vations spheres: 0 no involvement; 1 informed; 2 consulted; 3 bargained with

1.28 0 3UNIONINV _TRAIN* 1.25 0 3UNIONINV _TECH* 1.17 0 3UNIONINV _ICT* 1.11 0 3UNIONINV _ENV* 1.14 0 3UNIONINV _INTERNAT* 1.05 0 3

PERF (2006-2008)

GRVA/EMP Ln of value added per capita. Rate of growth over 2006-2008

0.05 –2.71 2.03

GRPROFIT/EMP Ln of profit per capita. Rate of growth over 2006-2008

–0.04 –2.51 3.25

GRCASHFLOW/EMP Ln of cash-flow per capita. Rate of growth over 2006-2008

0.11 –3.93 4.45

* These variables are only computed when union representative are present: 402 firms.

Tab. 5

. Pa

irw

ise

corr

elat

ions

am

ong

the

mai

n co

vari

ates

12

34

56

78

910

1112

1314

(1)

INN

O_O

RG

1(2

) T

RA

IN0.

301

(3)

INN

O_T

EC

H0.

440.

371

(4)

INN

O_E

NV

0.19

0.20

0.26

1(5

) IC

T0.

430.

320.

440.

191

(6)

INT

ER

NA

T0.

370.

200.

350.

100.

281

(7)

UN

ION

_IN

F0.

030.

090.

090.

100.

130.

161

(8)

UN

ION

_CO

NS

0.14

0.09

0.08

0.00

0.05

0.03

– 0.

331

(9)

UN

ION

_BA

RG

0.16

0.08

0.09

0.10

0.11

0.06

– 0.

19–

0.09

1(1

0) E

MP

_IN

F0.

060.

140.

090.

060.

190.

060.

35–

0.01

– 0.

031

(11)

EM

P_C

ON

S0.

090.

100.

080.

04–

0.01

0.03

– 0.

250.

160.

11–

0.67

1(1

2) G

RVA

/EM

P–

0.05

– 0.

04–

0.03

– 0.

07–

0.11

0.00

– 0.

140.

05–

0.02

– 0.

03–

0.02

1(1

3) G

RC

ASH

FL

OW

/EM

P–

0.05

– 0.

07–

0.05

– 0.

06–

0.09

– 0.

02–

0.09

– 0.

08–

0.02

0.03

– 0.

050.

521

(14)

GR

PR

OF

IT/E

MP

0.01

– 0.

02–

0.04

– 0.

130.

00–

0.03

– 0.

08–

0.03

0.00

0.03

– 0.

040.

590.

791

30

The dependent variables: innovative action to react to the crisis in 2009

Our dependent variables are given by innovative reactions to the chal-lenges brought by the crisis (ACTION_INNO). Less than 1% of the inter-viewed firms declare a null reaction to the crisis13. In more detail, we asked the management whether or not, and with which degree of intensity, inno-vative interventions on product, process and organisation/HRM factors have been adopted to cope with the crisis (tab. 6). An average of the intensity of reaction of all the three dimensions is given by the ACTION_TOT index, which is an arithmetic average of the values of the three specific indexes of reaction.

Each index of reaction is the expression of the average intensity degree of the intervention. As shown in table 6, the average degree of intervention intensity is quite high and close to a «strong» degree for all the three ty-pologies of action. As a whole, the firms are slightly more privileged with an intervention on the process dimension, rather than on product and organisa-tion. This result is, somehow, expected if we consider, according to the coun-tercyclical linkage between recessions and innovation, the bad times of the business cycle as windows of process/organisational changes implemented by the firms, which they are reluctant to undertake in periods of growth, be-cause of the higher opportunity costs. Overall, we may say that firms of our sample are not inactive in front of the crisis, rather, they are active in re-acting to it by implementing different strategies, ranging from the expected cost saving/efficiency gaining strategies (ACTION_PROC)14, to the less likely strategies aimed at increasing profits, due to higher level of competitiveness secured by further product innovations (ACTION_PROD)15 and by the con-struction of a more skilled/satisfied workforce (ACTION_ORG_HRM)16.

13 The answers could be affected by a potential bias of cognitive nature. In brief, if a manager has a self-image of being active and an innovator, it is likely he/she behaves as an active and innovating manager, in order to avoid dissonance between his/her self-image and his/her behaviours. The same manager when asked about the innovation activity within the firm could tend to over-estimate it, in order to remain consistent with the image he/she has of him/herself, even if the level of innovation activity is low. In synthesis, we could have what is called a problem of cognitive dissonance here: if real events (low innovation activity) do not prove deeply-held manager beliefs (being an innovating manager) this generate a psychologi-cal discomfort so that he/she will respond through distortion and denial (when asked, he/she still confirms a high level of innovation activity within the firm). Unfortunately, because of data shortcomings, we are not able to control for the actual presence of such a problem of cognitive dissonance, for which we refer to Akerlof - Dickens (1982); Hosseini (1997); Oxoby (2004) and Smith (2009).

14 Absence of any action in the process dimension for 13 firms.15 Absence of any action in the product dimension for 16 firms.16 Absence of any action in the HRM/ORG dimension for 7 firms.

31

4. Results

Our discussion of the results develops in order to test each hypothesis at a time on the basis of the general framework defined in equation (1).

Since firm specific characteristics are included in all the specifications, it is convenient to discuss their evidence here. Only the significant firm specific characteristics are discussed, but their results are not reported in tables 7, 8 and 9 for space constraint. A first general evidence is that the significant controls maintain their level of significance across the diverse specifications. It is also worth reminding that the set of firm specific characteristics we use largely outperform the standard controls used in the empirical literature, al-lowing us to capture much more firm heterogeneity, with respect to other empirical works.

Starting with the discussion of the results, they show that being a small- or medium sized firm (50-99 employees) might help coping with the crisis through the implementation of innovation actions, especially those actions related to human resource management and organisation (ACTION_ORG_HRM). The flexibility granted by the small size might be at the basis of a quick reaction, in the way that labour is managed and the firm is organ-ised, to the challenges brought by the crisis. Firms belonging to an indus-trial group show a less innovative behaviour in the crisis, especially on the

Tab. 6. Innovative actions to react to the crisis

Innovations Construction Mean Min Max

Degree of the intervention intensity (Null = 0; Very fee-ble = 1; Feeble = 2; Strong = 3; Very strong = 4)ACTION_TOT Constructed as the average of the following three

indexes:0.63 0 1

ACTION_PROC(Underlying strategic behav-iour: efficiency gains)

Index as sample average of the answers on five dimensions of process innovation: designing of product and service; efficiency/productivity/costs; flexibility in product variety; productive capacity; quality of productive process. Values normalised on the interval (0-1)

0.66 0 1

ACTION_PROD(Underlying strategic behav-iour: high competitiveness and future rent exploitation)

Index as sample average of the answers on five di-mensions of product innovation: new products and services; quality of product and service; access to new markets; marketing activities; logistics and dis-tribution activities. Values normalised on the inter-val (0-1)

0.62 0 1

ACTION_ORG_HRM(Underlying strategic behav-iour: skill base construction and efficiency gains)

Index as sample average of the answers on five dimensions of competitive factors: increased em-ployees competencies; increased employees respon-sibility and satisfaction; increased security and de-creased injuries; environmental impact reduction; adjustment to laws and quality standards. Values normalised on the interval (0-1)

0.62 0 1

32

side of product innovation (ACTION_PROD). A tentative explanation is that groups might internally redirect their innovation efforts in order to ad-dress them towards markets where such efforts are potentially more valu-able. Moreover, an additional strategic behaviour for groups could be that of delocalising the production process instead of reacting through innova-tion. In synthesis, the different options the groups have at their disposal to cope with the crisis challenges possibly reduce the probability of innovation intervention.

A negative coefficient is also associated with the suppliers. In this case, we may argue that suppliers are likely to be the link of the production chain that suffers more from the cash constraint, deriving both from the finan-cial sector and from the deferred payments of their clients, represented by other – frequently larger – firms (Cainelli - Montresor - Vittucci, 2012). An-other general result concerns the positive significant coefficients usually asso-ciated to the firms «proactive behaviour» (PROACTIVE). The latter means that a firm is dynamic in the pre-crisis and that it has a vision of what the policy makers should do in coping with the crisis that involves stimulus to the aggregate demand and to the development of human capital. Those firms sharing such characteristics seem to have a higher capacity to innovate dur-ing the crisis. Finally, another «pillar» among the firm characteristics is given by the importance of providing the employees with a good quality working environment (WORK_COND_P). This index, that includes positive elements of the working environment, is positively related to the innovation reaction of the firms: especially with product and process elements.

4.1. Innovation before and during the crisis: does innovation call for innova-tion?

The first specification we use to test Hp1 and Hp2a,b shows the results as reported in table 7 and assumes the following forms:

(2a) ACTION_INNOi,2009 = a0 + a1i,2006-2008(FIRM_SPEC) + + a2i,2006-2008(SPECIFIC_INNO_INDEXES) + ei,t

(2b) ACTION_INNOi,2009 = a0 + a1i,2006-2008(FIRM_SPEC) + + a2i,2006-2008(COMPOSITE_INNO_INDEXES) + ei,t

(2c) ACTION_INNOi,2009 = a0 + a1i,2006-2008(FIRM_SPEC) + + a2i,2006-2008(INNO_INTERACTIONS)17 + ei,t

17 The composite indexes we chose to interact were centred around their mean in or-der to construct multiplicative interacting variables (INNO × INNO) that are not correlated to each other, so that we can include them in the same specification avoiding multicollinearity problems.

33

where the dependent is in turn, for each specification, an overall measure of reaction to the crisis (ACTION_TOT), a reaction on process innovation (ACTION_PROC), a reaction on product innovation (ACTION_PROD) and a reaction on human resources and organisational aspects (ACTION_ORG_HRM). The FIRM_SPEC vector encloses the firm specific character-istics, the SPECIFIC_INNO_INDEXES vector is given by the innovation indexes capturing specific elements of the innovation activity within each innovation sphere, the COMPOSITE_INNO_INDEXES vector encloses indexes for each innovation sphere, which are constructed on the basis of the specific innovation elements, and finally, the vector INNO_INTERAC-TIONS is given by the multiplicative interaction between couples of com-posite innovation indexes.

Looking now more closely to each specification, we notice the relevant correlation of several «before-crisis» innovative practices with the intensity of reaction (Hp1). Two pillars of the organisational sphere, changes in both labour and production organisation (LAB_PRACTICES and PROD_PRAC-TICES), are positively linked to the overall reaction intensity and to the AC-TION_ORG_HRM. A mixed evidence is instead associated to the training sphere. On the one hand, we have a positive and significant coefficient for permanent employees covered by training programmes (COV_PERM), point-ing to the importance of new skills and competencies to sustain innovative organisational/HRM activities during the crisis. On the other hand, we find a negative sign associated to the variable capturing the wideness of compe-tencies covered by training programmes. The result is quite puzzling, but it might be due to a sort of threshold effect: widening too much the competen-cies in the pre-crisis turns out to be an obstacle to further innovation actions in the recession phase18.

For the technological sphere, INPUT_TECH is positively linked both with the overall reaction and with product and process reaction. Finally, also the activities covered by ICT (ACT_ICT) are positively related to process in-novation in the crisis. This list of relations outlines the importance of past in-novation for the firm capacity to implement innovation activities during the crisis as a way to answer to the challenges brought by the crisis itself, con-firming the validity of Hp1 for the covariates belonging to several innovation spheres.

We now question whether or not some forms of synergy exist among pre-crisis innovations that exert an effect on the capacity to implement innova-tion activities during the crisis (tab. 7): Hp2a and Hp2b. With Hp2a we aim

18 Such a result holds for specific types of competencies and innovative reactions. A check on the specific types of competencies shows that «technical competencies» and «or-ganisational/relational competencies» are the two elements out of four (see tab. 4) to drive the sign and the level of significance. On ACT_PROD the «technical competencies» and on ACT_PROC the «organisational/relational competencies» show negative and significant coef-ficients. Results are not reported for space constraint, but they are available upon request.

Tab. 7

. R

esul

ts t

o te

st H

P1 a

nd H

P2

2a

2b2c

AC

TIO

N_

TO

TA

CT

ION

_P

RO

CA

CT

ION

_P

RO

DA

CT

ION

_ O

RG

_HR

MA

CT

ION

_T

OT

AC

TIO

N_

PR

OC

AC

TIO

N_

PR

OD

AC

TIO

N_

OR

G_H

RM

AC

TIO

N_

TO

TA

CT

ION

_P

RO

CA

CT

ION

_P

RO

DA

CT

ION

_ O

RG

_HR

M

HP

1H

P2a

HP

2b

FIR

M_S

PE

C^

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

PR

OD

_ P

RA

CT

ICE

S0.

043*

*0.

054*

*0.

013

0.06

2**

(0

.020

)(0

.027

)(0

.027

)(0

.026

)

L

AB

_ P

RA

CT

ICE

S0.

109*

**0.

057

0.08

10.

190*

**

(0.0

37)

(0.0

49)

(0.0

55)

(0.0

46)

CO

V_P

ER

M0.

035*

0.03

10.

009

0.06

3***

(0

.020

)(0

.026

)(0

.028

)(0

.023

)

T

RA

IN_

CO

MP

– 0.

056*

*–

0.08

0**

– 0.

047

– 0.

039

(0

.027

)(0

.035

)(0

.036

)(0

.030

)

IN

PU

T_

TE

CH

0.19

4***

0.17

7***

0.31

5***

0.08

8

(0

.048

)(0

.062

)(0

.068

)(0

.058

)

A

CT

_IC

T0.

042

0.07

2**

0.04

60.

011

(0

.028

)(0

.036

)(0

.039

)(0

.035

)

IN

NO

_OR

G§

0.16

6***

0.12

4*0.

107

0.26

3***

0.18

0***

0.13

3*0.

116

0.28

6***

(0

.049

)(0

.065

)(0

.068

)(0

.066

)(0

.052

)(0

.068

)(0

.072

)(0

.067

)T

RA

ININ

G§

0.02

5–

0.00

1–

0.01

10.

089*

**0.

032

0.00

9–

0.00

30.

090*

**

(0.0

26)

(0.0

33)

(0.0

35)

(0.0

31)

(0.0

26)

(0.0

33)

(0.0

35)

(0.0

31)

INN

O_T

EC

H§

0.21

5***

0.21

1***

0.38

8***

0.04

10.

216*

**0.

205*

*0.

407*

**0.

029

(0

.064

)(0

.082

)(0

.092

)(0

.081

)(0

.067

)(0

.085

)(0

.095

)(0

.084

)IN

NO

EN

V§

0.00

60.

002

– 0.

039

0.05

4*0.

018

– 0.

014

– 0.

062

0.02

4

(0.0

24)

(0.0

27)

(0.0

35)

(0.0

31)

(0.0

3)(0

.032

)(0

.047

)(0

.04)

Tab. 7

. (f

ollo

ws)

ICT

§

0.

091*

*0.

083

0.15

4***

0.03

90.

095*

*0.

097*

0.14

6**

0.04

4

(0.0

44)

(0.0

58)

(0.0

58)

(0.0

54)

(0.0

42)

(0.0

56)

(0.0

57)

(0.0

51)

TR

AIN

xEN

V§

0.21

2**

0.17

0*0.

192

0.25

3*

(0.0

90)

(0.0

99)

(0.1

61)

(0.1

33)

TE

CH

xEN

V§

– 0.

191

– 0.

144

– 0.

534*

0.12

2

(0.1

84)

(0.2

35)

(0.2

82)

(0.2

84)

OR

GxT

EC

H§

– 0.

807*

– 0.

903

– 0.

579

– 0.

944

(0

.478

)(0

.628

)(0

.675

)(0

.587

)

TE

CH

xIC

T§

0.33

5–

0.02

7–

0.26

71.

241*

*

(0.3

93)

(0.5

21)

(0.5

69)

(0.4

87)

Con

stan

t0.

300*

**0.

289*

**0.

298*

**0.

313*

**0.

336*

**0.

333*

**0.

332*

**0.

343*

**0.

497*

**0.

460*

**0.

529*

**0.

500*

**

(0.0

72)

(0.0

91)

(0.0

95)

(0.0

84)

(0.0

67)

(0.0

88)

(0.0

88)

(0.0

77)

(0.0

63)

(0.0

83)

(0.0

84)

(0.0

72)

Obs

erva

tions

555

555

555

555

555

555

555

555

555

555

555

555

Adj

. R2

0.21

30.

118

0.18

90.

139

0.20

40.

114

0.19

50.

120

0.20

00.

107

0.19

00.

120

F5.

474

3.26

14.

840

3.36

27.

274

4.37

06.

957

3.67

56.

037

3.52

55.

244

3.56

1

Not

es:

* p

< 0

.10,

**

p <

0.0

5, *

** p

< 0

.01;

Rob

ust

to h

eter

oske

dast

icity

sta

ndar

d er

rors

in

pare

nthe

ses;

for

spa

ce c

onst

rain

t on

ly s

igni

fican

t va

riab

les

are

re-

port

ed; e

mpt

y ce

lls m

ean

the

vari

able

s ar

e no

t in

clud

ed in

the

spe

cific

atio

n; t

he V

aria

nce

Infla

tion

Fac

tor

(VIF

) do

es n

ot s

how

any

rel

evan

t m

ultic

ollin

eari

ty p

robl

em.

^ A

ll fir

m s

peci

fic c

hara

cter

istic

s ar

e in

clud

ed a

nd t

heir

mai

n re

sults

are

dis

cuss

ed a

t th

e be

ginn

ing

of S

ectio

n 4,

but

we

do n

ot r

epor

t th

em f

or s

pace

con

stra

int.

§ In

spe

cific

atio

ns 2

c th

e va

riab

les

are

cent

red

arou

nd t

heir

mea

n in

ord

er t

o re

duce

pro

blem

s of

mul

ticol

linea

rity

in t

he s

peci

ficat

ions

.

36

to test the existence of within innovation sphere complementarities. For both INNO_ORG and INNO_TECH we may claim that the nature of innovation activities within each of the two spheres is synergic: indeed such composite indexes are significant and have relatively larger coefficients with respect to their specific elements, which also emerged as significant.

For the ICT sphere, the presence of potential complementarities seems to emerge more robustly, since the composite index has a positive and sig-nificant coefficient for ACTION_TOT and ACTION_PROD, while the specific indexes are not related to such dependents. In this case, a complex strategy of joint introduction of different types of ICT, accounting for several managerial activities and objectives (Electronic Data Interchange, Enterprise Resource Planning, etc.), seems to create a background over which a more intensive reaction to the crisis is feasible.

Also noteworthy is the (weak) relation of the environmental innovation (INNO_ENV) composite index with ACTION_ORG_HRM. This result may be interpreted as the capacity to integrate environmental innovation with other organisational factors, mainly involving the management of human re-sources and organisational aspects. Whether such an integration is capable of bringing the firms out of the crisis along a sustainable path is a matter of interest for future works.

The last set of results is provided in order to verify the Hp2b, which is only marginally supported by our evidence. A quite robust evidence support-ing this type of complementarities emerges only for training and environmen-tal innovation (TRAIN × ENV), for the overall reaction and both for proc-ess and organisational/HRM aspects. We may argue that the small fraction of firms engaged in environmental related innovation activities is dynamic and possibly more flexible than the counterpart not involved in green strategies. Such dynamism is supported by investment in workers’ human capital and it persists over time, even in a period of deep recession, leading these firms to react more intensively to the crisis challenges. Also, the interaction between technological innovation and ICT (INNO_TECH × ICT) is robustly related to the reaction intensity in organisational/HRM aspects. Firms involved in complementary technological and ICT innovations before the crisis seem to be aware of the need for the introduction of further complementary organi-sational and human resource practices to complement the technological/ICT ones, also in time of crisis.

The two interacted terms slightly and negatively significant INNO_TECH × INNO_ENV and INNO_ORG × INNO_TECH seem to point to a sort of trade off between high levels of innovation intensity just before the crisis on different innovation spheres and the capacity to innovate during the crisis for product innovation and for the overall index of reaction19. Although

19 It is worth stressing the low level of significance of the coefficients (10%), especially in a cross section environment.

37

the evidence is very weak, a tentative explanation might be that the intense innovative effort in such specific innovation spheres makes firms equipped only to tackle the crisis challenges in virtue of the innovation activity made before the crisis. Another possible explanation may lie on the fact that most dynamic firms, in terms of these specific couples of innovations, could have been more strongly displaced by the crisis, because they were in a moment of change and they were possibly financially stressed.

In synthesis, the evidence points out how the innovation persistence, before and during the crisis, can be attributable to a tradition of incremen-tal innovation and accumulation of competencies, over time (see Antonioli - Bianchi - Mazzanti - Montresor - Pini, 2011 for a detailed description of Emilia-Romagna manufacturing firm innovation strategies): the more innova-tive firms had been in the past, the more innovative during the crisis they tend to be. It is likely that such «path dependency» builds on the basis of existing competencies and capabilities that had been widened and developed through organisational changes and technological innovation before the re-cession. The role of synergic innovation activities is also confirmed, especially within the innovation sphere, while the complementary nature of innovation activities belonging to different spheres (between innovation spheres comple-mentarities), that usually emerges with reference to economic performance, is less evident.

4.2. The industrial relations climate: how do firm-level employment relation-ships influence the reaction to the crisis?

In order to test the Hp3a Hp3b hypotheses we set up the following spec-ifications:

(3a) ACTION_INNOi,2009 = b0 + b1i,2006-2008(FIRM_SPEC) + + b2i,2006-2008(COMPOSITE_INNO_INDEXES) + + b3i,2006-2008(INDREL_COMPOSITE_INDEXES) + ui,t

(3b1) ACTION_INNOi,2009 = b0 + b1i,2006-2008(FIRM_SPEC) + + b2i,2006-2008(COMPOSITE_INNO_INDEXES) + + b3i,2006-2008(INDREL_SPECIFIC_INDEXES) + ui,t

(3b2) ACTION_INNOi,2009 = b0 + b1i,2006-2008(FIRM_SPEC) + + b2i,2006-2008(COMPOSITE_INNO_INDEXES) + + b3i,2006-2008(INDREL_SPECIFIC_INDEXES_BY_INNOVATION) + ui,t

where ACTION_INNO and FIRM_SPEC vectors are defined as before (see specifications 2a, b, and c), the COMPOSITE_INNO_INDEXES vector is the same as in specification 2b, INDREL_COMPOSITE_INDEXES is made

38

of two main industrial relations variables, one (UNION_INV) capturing the cooperative nature of the relation between unions and management (indirect participation), and the other (EMP_INV) capturing the same participative relations between employees and management (direct participation). The IN-DREL_SPECIFIC_INDEXES encloses more specific participative variables as described in table 4. Finally, the INDREL_SPECIFIC_INDEXES_BY_INNOVATION is a further refined set of variables that captures the degree of unions and employees «participation» in the decisions concerning each in-novation sphere.

We, thus, capture both the correlation of cooperative industrial relations with the reaction to the crisis and the potential substitution effect between indirect and direct participation.

It is worth stressing that the analysis is conducted for the firms with un-ion representatives20, a sample of 402 firms21.

As we can see in table 8, the main composite indexes of both direct and indirect participation, which capture less adversarial employment relation-ships, are not significant. This means that a general involvement of unions and employees is not related to a high level of innovation reaction to the crisis. This first evidence would leave us to consider a participative indus-trial relations climate substantially neutral. However, if we refine our analysis by looking at the specific degree of involvement, we find that having imple-mented a process of bargaining with unions (on changes over the innovation spheres) before the crisis is correlated with a high reaction capacity. The link is more robust for the overall reaction to the crisis and less for the other specific strategies, but if we couple this evidence with the quite robust and positive linkage of employees involvement, although through simple informa-tion, with product innovation we can argue that Hp3a is not rejected. The interpretation behind this evidence could be that the mediating role of the unions is important, in order to smooth the adoption of innovation activities that may have a possible negative outcome for the employees (e.g., process changes in order to reduce the labour costs).

Notwithstanding the challenges brought by the diffusion of managerial practices addressed to increase the employees involvement without the medi-ating role of the unions, it appears quite clear how a participative industrial relation climate favours firms addressing innovation strategies.

20 In such a way, we are not forced to arbitrarily impute a value in the industrial relations variables concerning the unions/management relationships for those firms with no union rep-resentatives.

21 Because the employees involvement may be present, also where union representative are not present, and possibly be even wider where unions are absent, we also run regressions for the 153 firms without unions, in order to verify the influence of employees involvement. The results tell us that employees’ involvement does not have any significant impact on the innovative reaction in the crisis. The results are not reported due to space constraint but they are available from the authors upon request.

Tab. 8

. R

esul

ts t

o te

st H

P3

3a

3b1

3b2

AC

TIO

N_

TO

TA

CT

ION

_P

RO

CA

CT

ION

_P

RO

DA

CT

ION

_O

RG

_ H

RM

AC

TIO

N_

TO

TA

CT

ION

_P

RO

CA

CT

ION

_P

RO

DA

CT

ION

_O

RG

_ H

RM

AC

TIO

N_

TO

TA

CT

ION

_P

RO

CA

CT

ION

_P

RO

DA

CT

ION

_O

RG

_ H

RM

HP

3aH

P3b

HP

3b§

FIR

M_S

PE

C ^

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

CO

MP

OS

ITE

_IN

NO

_IN

DE

XE

S^Y

esY

esY

esY

esY

esY

esY

esY

esY

esY

esY

esY

es

UN

ION

_IN

V0.

026

0.04

80.

019

0.00

9

(0.0

24)

(0.0

34)

(0.0

37)

(0.0

30)

EM

P_I

NV

0.00

6–

0.05

70.

033

0.04

1

(0.0

31)

(0.0

42)

(0.0

43)

(0.0

41)

UN

ION

_BA

RG

0.05

9**

0.05

8*0.

066*

0.05

4*

(0.0

25)

(0.0

35)

(0.0

38)

(0.0

31)

EM

P_I

NF

0.04

60.

027

0.07

8**

0.03

0

(0

.028

)(0

.035

)(0

.038

)(0

.033

)E

MP

INV

_EN

V

0.

049*

*0.

052*

*0.

052

0.04

7*

(0.0

23)

(0.0

26)

(0.0

32)

(0.0

26)

EM

PIN

V_I

NT

ER

NA

T

–

0.01

6–

0.05

1**

0.02

4–

0.02

6

(0.0

22)

(0.0

25)

(0.0

27)

(0.0

30)

UN

ION

INV

_EN

V

–

0.01

8–

0.05

3**

0.02

7–

0.02

9

(0.0

21)

(0.0

26)

(0.0

29)

(0.0

30)

UN

ION

INV

_IN

TE

RN

AT

– 0.

019

0.02

4–

0.04

8*–

0.03

0

(0.0

25)

(0.0

29)

(0.0

28)

(0.0

38)

Con

stan

t0.

366*

**0.

373*

**0.

399*

**0.

325*

**0.

346*

**0.

358*

**0.

371*

**0.

308*

**0.

373*

**0.

395*

**0.

394*

**0.

329*

**

(0.0

76)

(0.0

90)

(0.1

01)

(0.0

88)

(0.0

75)

(0.0

91)

(0.1

00)

(0.0

88)

(0.0

77)

(0.0

94)

(0.1

03)

(0.0

89)

Obs

erva

tions

402

402

402

402

402

402

402

402

402

402

402

402

Adj

. R2

0.20

60.

125

0.19

10.

105

0.22

70.

131

0.21

10.

113

0.20

40.

118

0.20

10.

101

F4.

727

3.26

34.

738

2.50

04.

813

3.09

44.

598

2.60

43.

800

2.93

44.

112

2.28

6

Not

es:

* p

< 0

.10,

**

p <

0.0

5, *

** p

< 0

.01;

Rob

ust

to h

eter

oske

dast

icity

sta

ndar

d er

rors

in

pare

nthe

ses;

for

spa

ce c

onst

rain

t on

ly s

igni

fican

t va

riab

les

are

repo

rted

; em

pty

cells

mea

n th

e va

riab

les

are

not

incl

uded

in t

he s

peci

ficat

ion;

the

Var

ianc

e In

flatio

n F

acto

r (V

IF)

does

not

sho

w a

ny r

elev

ant

mul

ticol

linea

rity

pro

blem

, but

in s

peci

-fic

atio

ns 3

c.^

All

firm

spe

cific

cha

ract

eris

tics

are

incl

uded

and

the

ir m

ain

resu

lts a

re d

iscu

ssed

at

the

begi

nnin

g of

Sec

tion

4, b

ut w

e do

not

rep

ort

them

for

spa

ce c

onst

rain

t; th

e co

mpo

site

inno

vatio

n in

dexe

s in

clud

ed in

ord

er n

ot t

o om

it re

leva

nt v

aria

bles

sho

w t

he s

ame

resu

lts a

s re

port

ed in

tab

. 7.

§ T

his

set

of s

peci

ficat

ion

slig

htly

suf

fers

fro

m m

ultic

ollin

eari

ty p

robl

ems,

how

ever

the

re i

s no

t a

clea

r cu

t so

lutio

n to

mul

ticol

linea

rity

(K

enne

dy,

2001

). D

ropp

ing

spec

ific

vari

able

s of

inte

rest

cou

ld n

ot b

e a

good

cho

ice

if w

e ar

e in

tere

sted

on

thei

r re

sults

, so

we

deci

ded

to k

eep

all t

he v

aria

bles

, with

the

cav

eat

that

the

res

ults

may

be

affe

cted

by

wea

k m

ultic

ollin

eari

ty.

40

Looking at the specific indexes of unions and employees involvement on each innovation sphere, we notice that the higher the employees involvement on environmental related innovation, the higher is the reaction capacity of the firm, while the union representatives involvement on two single innova-tion spheres, environmental innovation and internationalisation, shows some negative relations with the capacity to innovate in the crisis. The same is also true for the employees’ involvement on internationalisation strategy when the dependent is innovation in product related activities. It seems possible to infer that, for the share of firms undertaking environmental innovation, the employees’ involvement represents an element that improves firm capacity to react to the crisis through further innovation. Hypothesis Hp3b finds weak partial support. We are looking, although through a static picture provided by our data, to a likely process of «relevance transfer», from indirect to di-rect participation of the employees in the decisions related to environmental innovation activities.

In synthesis, we may argue that the strategic reaction through intense in-novative actions is also supported by a pre-crisis participative industrial rela-tions climate, when participation has to be intended as indirect participation. The role of unions acquires importance in a phase of deep production con-traction. We argue that the strategic choices adopted to deal with the cri-sis need more consensus and cohesion between social parties, in a period of economic downturn than in a stable or growing economic environment and non-adversarial relationships between management and unions may provide such a consensus, more than employees’ direct participation. However, the latter emerges as important especially for firms involved in environmental in-novations, pointing to a process of substitution between indirect and direct participation, that might have been delayed but not stopped by the economic crisis. This has revived the role of unions as institutions to deal with, faced with grievances and conflicts, more frequent during an economic downturn than in a stable phase of the economic cycle.

4.3. Does past economic performance act as a base upon which to rely in order to innovate during the crisis?

Because both innovation and industrial relations add knowledge to our investigation of innovation reaction, we keep them in this last set of specifi-cations, to which we add past economic performances22.

22 Here, we return to the 555 firms, although industrial relations variables are used. Since we are no longer interested in the relation between industrial relations and innovative reaction (we analysed such a relation in the preceding section), but we do not want to lose relevant information, we decided to fill in the missing values in the industrial relations variables with zeroes (0s).

41

(4) ACTION_INNOi,2009 = c0 + c1i,2006-2008(FIRM_SPEC) + + c2i,2006-2008(COMPOSITE_INNO_INDEXES) + + c3i,2006-2008(INDREL_SPECIFIC_INDEXES) + + c4i,2006-2008(ECONOMIC_PERF) + vi,t

where the vectors FIRM_SPEC, COMPOSITE_INNO_INDEXES and IN-DREL_SPECIFIC_INDEXES are as in the previous sections and the vector ECONOMIC_PERF is given by three account indicators that are expressed in rate of growth for the period 2006-2008: value added per employee, cash flow per employee and profit per employee.

In order to overcome23 missing values flaws, we implement the Multiple Imputation (MI) procedure24. The MI procedure is reasonable if we assume that the missingness pattern is at least a Missing At Random (MAR) pattern, that is to say, the «probability that data are missing does not depend on un-observed data, but may depend on observed data» (Stata Manual, 2009, p. 6). In our case, we are able to control for the variables likely to induce the miss-ingness pattern, such as the firm dimension: the smaller the firm, the higher the probability of having missing values in the account data. Hence, we are confident in defining the missingness pattern of account data as a MAR.

We notice, in table 9, that there is no relationship between pre-crisis growth of performance indicators and firm capacity to react in an intense way, hence, Hp4 is not supported by our evidence. The absence of any sig-nificant relation is a signal of the potential disrupting effect of the recession on the «traditional» linkage between innovation and economic performance, as pointed out by many scholars, notwithstanding the causal nexus and the causality direction (see among others the recent contributions by Coad-Rao, 2010 and Hall, 2011, for an assessment of the linkage between innovation and economic performance). Here, we find a different evidence; it is as dif-ferent as the innovative reaction to the crisis, with respect to the business-as-usual innovation activity during stable periods of the economic cycle. This evidence means that the reaction of high performers is not dissimilar to the reaction of poor performers, before the crisis. The disruptive nature of the crisis breaks the positive linkage between economic performance and innovation.

23 With missing values, originally in the account dataset and those generated by the pro-cessing of the data, the remaining observations are 233. The analysis run on this limited subset of observations provides similar results with respect to those presented in table 9. Evidence is available from the authors upon request.

24 The procedure creates multiple datasets on the basis of the original one. We decided to generate 20 of them, in order to reduce the sampling error. The original missing values are replaced by «plausible» values. All the covariates used in the regression model (the completed model), as well as the outcome variables of such model, are also used as predictors in the imputation model, in order to avoid misspecification in the imputation model. The analysis of the completed-data model is then performed over the 20 datasets and the results are com-bined in a single result (see the Sata11 Manual, 2009, for further information).

42

The specific evidence reported in tables 7-9 is synthesised in table 10, in order to provide a meaningful set of results that can be easily referred to the hypotheses.

Tab. 9. Results to test HP4

ACTION_TOT

ACTION_PROC

ACTION_PROD

ACTION_ORG_HRM

HP4

FIRM_SPECIFIC_CHARACTERISTICS^ Yes Yes Yes YesCOMPOSITE_INNO_INDEXES^ Yes Yes Yes YesINDREL_SPECIFIC_INDEXES^ Yes Yes Yes Yes«GR_VAEMP_0608» – 0.010 0.014 – 0.028 – 0.015

(0.022) (0.030) (0.031) (0.029)«GR_CASHFLOWEMP_0608» 0.009 0.016 – 0.011 0.020

(0.024) (0.034) (0.034) (0.026)GR_PROFITEMP – 0.006 – 0.022 0.016 – 0.012

(0.019) (0.027) (0.027) (0.021)Constant 0.304*** 0.285*** 0.317*** 0.311***

(0.070) (0.092) (0.093) (0.082)Observations 555 555 555 555Adjusted R-squared§§ 0.222 0.126 0.213 0.126F 5.63 3.29 5.09 3.03

Notes: * p < 0.10, ** p < 0.05, *** p < 0.01; Robust to heteroskedasticity standard errors in paren-theses; for space constraint only significant variables are reported; empty cells mean the variables are not included in the specification; the Variance Inflation Factor (VIF) does not show any relevant multicolline-arity problem.

^ All firm specific characteristics are included and their main results are discussed at the beginning of Section 4; the composite innovation indexes and the industrial relation specific indexes included in order not to omit relevant variables show the same results as reported in tab. 7 and tab.8 but we do not report them for space constraint.

§ All the «performance» variables are reported although they are not significant. The results are based on the Multiple Imputation strategy (Sata11 Manual, 2009) which has been used in order to over-come the problem given by missing values.

§§ The adjusted R-squared has been obtained using the mibeta command (Stata Manual, 2009).

Tab. 10. Summing up the empirical evidence

Dependent variables

ACTION_TOT

ACTION_PROC

ACTION_PROD

ACTION_ORG_HRM

INNOVATION Hp1 ++ ++ ++ ++Hp2a ++ ++ ++ ++Hp2b +/– + – +INDUSTRIAL RELATIONSHp3a + + + +Hp3b +/– +/– +/– +/– ECONOMIC PERFORMANCEHp4 / / / /

Note: + + (– –) means a strong support to (rejection of) Hp; + (–) means a weak support to (rejec-tion of) Hp; +/– means mixed evidence; / means no evidence.

43

Overall, it is clear how the first set of hypotheses, related to the link-ages among innovations before and during the crisis, is largely supported by our evidence. The same holds, although with a lesser degree of robustness, for the relationships between participative industrial relations and innovation reactions, while there is no evidence of a nexus between economic perform-ance in the pre-crisis, and innovation during the recession.

5. Conclusions

The present work shows the reaction of Emilia-Romagna manufacturing firms faced with the economic downturn, by means of econometric analysis, based on data from a structured survey carried out in 2009.

The econometric exercise provides evidence for the main research hy-potheses concerning the linkages between pre-crisis innovative strategies, economic performances and industrial relations and the capacity to react to the crisis through further innovation activities. In a sense, we are investigat-ing whether the Emilia-Romagna manufacturing firms have the resistance to survive the economic crisis and to resurge in the medium run as one of the leading industrial manufacturing system. On the basis of our evidence we can provide some tentative considerations.

The reduced-form model, with innovative activities in the recession as dependent variables, clearly shows how some firm specificities are important in sustaining innovation as an instrument to react to a downturn. Because it is not straightforward to think about innovation as a means to overcome the recession criticalities, we consider the characteristics that positively im-pact on the innovative strategies in the crisis as extremely relevant. Indeed, it seems that the crisis has to be considered, to some extent, more as a pe-riod of change (Mensch, 1979) rather than inertia for Emilia-Romagna man-ufacturing firms, given the declared intensity of innovation. Furthermore, a sustained innovative activity may contribute in a crucial way to the survival and to the competitive capacity of the firms at the end of the slowdown: the higher the innovation intensity in the crisis, the shorter will, probably, be the shadow cast by the crisis upon the future. The capacity of firms to maintain a certain coherence between innovation strategies, before and during the cri-sis, is a point of strength of the Emilia-Romagna manufacturing firms: the recession phase does not interrupt the innovation path.

Globally, the analysis of the dynamic relations between pre-crisis and during-crisis innovation activities clarifies the complex nature of the innova-tion strategies. Complementarities among innovation activities within each in-novation sphere clearly emerge, while the crisis nature and strength seem to weaken the complementary link between innovation spheres.

A further important element for the implementation of all the innovation strategies in the crisis is the connotation of the industrial relations climate: the more participative the relations are, the better this will be for the inno-

44

vation intensity during the crisis. The analysis on industrial relations aspects points out the importance of indirect participation of the employees through union representatives, but also the relevance of direct employees’ participa-tion. The dialogue between management and unions turns out to be impor-tant in times of crisis, as «driver» of a superior capacity to react to the crisis through innovation. Although the social dialogue and the industrial relations system have been undermined in the last decade, for several reasons – rang-ing from critical government orientation towards the social dialogue, to the lack of capacity of the unions to provide a unified front and, at firm level, to the diminished quality of the union representatives, which hampers the capacity of the union to resolve, in an efficient and efficacious manner, the manifested grievances – the firm level industrial relations climate, oriented to cooperation, rather than to conflict, is a point of strength of the Emilia-Romagna production model.

The same does not hold for the pre-crisis economic performance effect. The substantial lack of significance of economic performance indicators cou-pled with the results of past innovation strategies tell us that it is not the economic trend of a firm which supports innovation activities in the crisis, rather it is its innovation dynamism before the recession: in a sense, inno-vation calls for innovation. Hence, the lack of economic performance lock-in emphasises the role of various innovative drivers and encourages firms to adopt innovations to tackle downturn periods.

Our main results translate into clear recommendations to policy mak-ers and social actors. The first one concerns the strategic choice to innovate. This should be a priority in positive phases of the economic cycle, because it helps in sustaining the firm competitiveness, and also because it creates the capacity to cope with the crisis challenges, in a way that can lead to a smoother exit from the economic downturns. Thus, on the one hand, the firm management is called upon to be dynamic on the innovation side and, on the other hand, regional policy makers should be aware of the importance of policy programmes aimed at sustaining not only firm innovation activities, but also those elements that help to connote a regional innovation system. One of these elements is the social dialogue among the economic actors. In-deed, the main second insight concerns the importance of fostering social dialogue. The latter turns out to be a powerful instrument to accommodate different innovation strategies.

The evidence suggests that designing a consistent and coherent innova-tion strategy could help firms in constructing their capacity to react to the downturns through innovation. This can be thought to be an exit strategy, potentially capable of creating post-crisis competitive advantages.

45

References

Aghion P. - Saint-Paul G. (1998), Uncovering Some Causal Relationships Between Productivity Growth and Structure of Economic Fluctuations: A Tentative Sur-vey, Labour, vol. 12, no. 2, pp. 279-303.

Akerlof G.A. - Dickens W.T. (1982), The Economic Consequences of Cognitive Dis-sonance, American Economic Review, vol. 72, no. 3, pp. 307-319.

Amin A. (1999), The Emilian Model: Institutional Challenges, European Planning Studies, vol. 7, no. 4, pp. 389-406.

Antonioli D. (2009), Industrial Relations, Techno-Organizational Innovation and Firm Economic Performance, Economia Politica. Journal of Analytical and Insti-tutional Economics, vol. XXVI, no. 1, pp. 21-52.

Antonioli D. - Bianchi A. - Mazzanti M. - Montresor S. - Pini P. (2011), Strategie di innovazione e risultati economici. Un’indagine sulle imprese manifatturiere dell’Emilia-Romagna, Milano, FrancoAngeli.

Antonioli D. - Marzucchi A. - Montresor S. (2013), Regional Innovation Policy and Innovative Behaviour. Looking for Additional Effects, European Planning Stu-dies, forthcoming.

Antonioli D. - Mazzanti M. - Pini P. (2009), Innovation, Working Conditions and Industrial Relations. Evidence for a Local Production System, Economic and In-dustrial Democracy, vol. 30, no. 2, pp. 157-18.

— (2010), Productivity, Innovation Strategies and Industrial Relations in SME. Em-pirical Evidence for a Local Manufacturing System in Northern Italy, Internatio-nal Review of Applied Economics, vol. 24, no. 4, pp. 453-482.

— (2011), Innovation, Industrial Relations and Employee Outcome, Journal of Eco-nomic Studies, vol. 38, no. 1, pp. 66-90.

Antonioli D. - Pini P. (2005), Partecipazione diretta, partecipazione indiretta ed innovazioni tecno-organizzative. Alcuni risultati per il sistema locale di Reggio Emilia, Quaderni di Rassegna Sindacale, vol. 6, no. 2, pp. 195-226.

Aral S. - Weill P. (2004), IT Assets, Organizational Capabilities and Firm Performan-ce: Asset and Capability Specific Complementarities, MIT Sloan School of Ma-nagement Working Paper No. 451604, mimeo.

Arvanitis S. (2005), Modes of Labor Flexibility at Firm Level: Are There Any Impli-cations for Performance and Innovation? Evidence for the Swiss Economy, Indu-strial and Corporate Change, vol. 14, no. 6, pp. 993-1026.

Barca F. - McCann P. (2011), Outcome Indicators for the Thematic Priorities Ad-dressing the Europe 2020 Objective «Improving the Conditions for Innovation, Research and Development», European Commission, http://ec.europa.eu/regio-nal _policy/ sources/docgener/evaluation/performance_en.htm, Bruxelles.

Barlevy G. (2005), Why Don’t Recessions Encourage More R&D Spending?, Chicago Fed Letter (220), Chicago.

— (2007), On the Cyclicality of Research and Development, The American Econo-mic Review, vol. 97, no. 4, pp. 1131-1164.

Bartel A. - Ichniowski C. - Shaw K. (2005), How does Information Really Affect Productivity? Plant-Level Comparisons of Product Innovation, Process Improve-ment and Worker Skills, NBER Working paper 11773, National Bureau of Eco-nomic Research, mimeo.

Black S. - Lynch L. (2001), How to Compete: The Impact of Workplace Practices and Information Technology on Productivity, The Review of Economics and Stati-stics, vol. 83, no. 3, pp. 434-445.

46

— (2004), What’s Driving the New Economy: The Benefits of Workplace Innova-tion, Economic Journal, vol. 114, no. 493, pp. 97-116.

Blundell R. - Griffith R. - Van Reenen J. (1999), Market Structure and Innovation: Evidence from British Manufacturing Firms, Review of Economic Studies, vol. 66, no. 3, pp. 529-544.

Boheim R. - Booth A. (2004), Trade Union Presence and Employer-Provided Trai-ning in Great Britain, Industrial Relations, vol. 43, no. 3, pp. 520-544.

Bondarouk T. - Looise J. (2005), HR Contribution to IT Innovation Implementation: Results of Three Case Studies, Creativity and Innovation Management, vol. 14, no. 2, pp. 160-168.

Brandolini A. - Bugamelli M.B. (2009), Rapporto sulle tendenze nel sistema produt-tivo italiano, Questioni di Economia e Finanza, no. 45, Bank of Italy, Rome.

Brusco S. (1982), The Emilian Model: Productive Decentralization and Social Inte-gration, Cambridge Journal of Economics, vol. 6, no. 2, pp. 167-184.

Brynjolfsson E. - Hitt L. (2000), Beyond Computation: Information Technology, Or-ganizational Transformation and Business Performance, The Journal of Economic Perspectives, vol. 14, no. 4, pp. 23-48.

Brynjolfsson E. - Yang S. (1996), Information Technology and Productivity: A Re-view of the Literature, Advances in Computers, vol. 43, no. 1, pp. 179-214.

Cainelli G. - Montresor S. - Vittucci G. (2012), Firm Exit and Spatial Agglomera-tion. Evidence on the Resilience of Italian Provinces, Journal of Evolutionary Economics, forthcoming.

Cappelli P. - Neumark D. (2001), Do «High-Performance» Work Practices Improve Establishment-Level Outcomes?, Industrial and Labor Relations Review, vol. 54, no. 4, pp. 737-775.

Caroli E. - Reenen J.V. (2001), Skilled-biased Organizational Change? Evidence from a Panel of British and French Establishments, The Quarterly Journal of Economi-cs, vol. 116, no. 4, pp. 1449-1492.

Coad A. - Rao R. (2010), R&D and Firm Growth Rate Variance, Economics Bulletin, vol. 30, no. 1, pp. 1-6.

DellaTorre E. (2009), High Performance Work Practices e caratteristiche delle im-prese innovative, Economia Politica. Journal of Analytical and Institutional Econo-mics, vol. XXVI, no. 2, pp. 315-352.

Evangelista R. - Iammarino S. - Mastrostefano V. - Silvani A. (2002), Looking for Re-gional Systems of Innovation. Evidence from the Italian Innovation Survey, Re-gional Studies, vol. 36, no. 2, pp. 173-186.

Faini R. - Sapir A. (2005), Un modello obsoleto? Crescita e specializzazione dell’eco-nomia italiana, in Ichino A. - Pisauro G. - Scarpa C. - Boeri T. - Faini R. (Eds.), Oltre il declino, Bologna, Il Mulino.

Foray D. - David P. - Hall B. (2011), Smart Specialization. From Academic Idea to Political Instrument, the Surprising Career of a Concept and the Difficulties In-volved in Its Implementation, MTEI-Working Papers-2011-001, École Polytech-nique Fédérale de Lausanne, mimeo.

Giedeman D. - Isley P. - Simons G. (2006), Innovation and the Business Cycle: A Comparison of the U.S. Semiconductor and Automobile Industries, International Advances in Economic Research, vol. 12, no. 2, pp. 277-286.

Griliches Z. (1995), R&D and Productivity: Econometric Results and Measurement Issues, in Stoneman P. (Ed.), Handbook of the Economics of Innovation and Tech-nological Change, New Jersey, Blackwell.

47

Hall B. (2011), Innovation and Productivity, NBER Working Papers 17178, National Bureau of Economic Research, mimeo.

Hall B. - Lotti F. - Mairesse J. (2009), Innovation and Productivity in SMEs: Empiri-cal Evidence for Italy, Small Business Economics, vol. 33, no. 1, pp. 13-33.

Hall R. - Blanchard O. - Fischer S. (Eds.) (1991), NBER Macroeconomics Annual, Massachusetts, MIT Press.

Hollander H. - Tarantola S. - Loschky A. (2009), Regional Innovation Scoreboard RIS (2009), Technical report, PRO INNO EUROPE, European Commission.

Hosseini H. (1997), Cognitive Dissonance as a Means of Explaining Economics of Irrationality and Uncertainty, Journal of Socio-Economics, vol. 26, no. 2, pp. 181-189.

Hujer R. - Radic D. (2003), Holistic Innovation Success? Complementarities betwe-en Flexible Workplace and Human Resource Management Practices in the Inno-vation Process, paper presented at the ZEW Workshop «Empirical Economics of Innovation and Patenting» at the Centre for European Economic Research, Mannheim, 14th-15th March, mimeo.

Huselid M. - Becker B. (1996), Methodological Issues in Cross-Sectional and Panel Estimates of the Human Resource-Firm Performance Link, Industrial Relations, vol. 35, no. 3, pp. 400-422.

Janod V. - Saint-Martin A. (2004), Measuring the Impact of Work Reorganization on Firm Performance: Evidence from French Manufacturing, Labour Economics vol. 11, no. 3, pp. 785-798.

Kennedy P. (2001), A Guide to Econometrics, New Jersey, Blackwell.Laursen K. - Foss N. (2003), New Human Resource Management Practices, Com-

plementarities and the Impact on Innovation Performance, Cambridge Journal of Economics, vol. 27, no. 2, pp. 243-263.

Leoni R. (2012), Workplace Design, Complementarities among Work Practices and the Formation of Key Competencies. Evidence from Italian Employees, Indu-strial and Labor Relations Review, vol. LXV, no. 2, pp. 316-349.

— (2013), Organization of Work Practices and Productivity: an Assessment of Re-search on World-Class Manufacturing, in Grandori A. (ed.), Handbook of Eco-nomic Organization. Integrating Economic and Organization Theory, Cheltenham, Edward Elgar, forthcoming.

Lipsey R. - Carlaw K. - Bekar C. (2005), Economic Transformations: General Pur-pose Technologies and Long Term Economic Growth, Oxford, Oxford University Press.

Lucchese M. - Pianta M. (2011), Cycles and Innovation, Working Papers 1102, Uni-versity of Urbino, mimeo.

Lynch L. (2012), The Evolving Nature of High Performance Workplace Practices in the United States, in Bryson A. (Ed.), Advances in the Economic Analysis of Par-ticipatory and Labor-Managed Firms, Bingley (UK), Emerald Group Publishing Limited, pp. 267-309.

Machin S. - Wood S. (2005), Human Resource Management as a Substitute for Tra-de Unions in British Workplaces, Industrial and Labor Relations Review, vol. 58, no. 2, pp. 201-218.

Marzucchi A. - Antonioli D. - Montresor S. (2012), Research Cooperation Within and Across Regional Boundaries. Does Innovation Policy add Anything?, IPTS Working Paper on Corporate R&D and Innovation, no. 04/2012, JRC-IPTS, Se-villa, mimeo.

48

Melitz M. (2003), The Impact of Trade on Intra-Industry Reallocation and Aggregate Industry Productivity, Econometrica, vol. 71, no. 6, pp. 1695-1725.

Menezes-Filho N. - Van Reenen J. (2003), Unions and Innovation: A Survey of the Theory and Empirical Evidence, in Addison C. - Schnabel J.T. (Eds.), Internatio-nal Handbook of Trade Unions, Cheltenham, Edward Elgar.

Mensch G. (1979), Stalemate in Technology: Innovations Overcome the Depression, Cambridge, Ballinger Publications.

Metcalf D. (2003), Unions and Productivity, Financial Performance and Investment: International Evidence, in Addison J.T. - Schnabel C. (Eds.), International Hand-book of Trade Unions, Cheltenham, Edward Elgar.

Miceli V. (2010), Distretti Tecnologici e Sistemi Regionali di Innovazione. Il caso Italia, Bologna, Il Mulino.

Michie J. - Sheehan M. (2003), Labour Market Deregulation, «Flexibility» and Inno-vation, Cambridge Journal of Economics, vol. 27, no. 1, pp. 123-143.

Oxoby R.J. (2004), Cognitive Dissonance, Status and the Growth of the Underclass, Economic Journal, vol. 114, no. 498, pp. 727-749.

Peters B. (2009), Persistence of Innovation: Stylised Facts and Panel Data Evidence, The Journal of Technology Transfer, vol. 34, no. 2, pp. 226-243.

Pianta M. - Vaona A. (2007), Innovation and Productivity in European Industries, Economics of Innovation and New Technology, vol. 16, no. 7, pp. 485-499.

Pini P. (2005), Dinamiche innovative, partecipazione e risultati d’impresa in un siste-ma locale di produzione, Economia Politica. Journal of Analytical and Institutio-nal Economics, vol. XXII, no. 1, pp. 3-23.

Pini P. - Santangelo G. (2005), Innovation Types and Labour Organisational Practi-ces: A Comparison of Foreign and Domestic Firms in the Reggio Emilia Indu-strial Districts, Economics of Innovation and New Technology, vol. 14, no. 4, pp. 251-276.

— (2010), The Underlying Internal Learning Processes of Incremental and Radi-cal Innovations. An Analysis in the Emilia-Romagna Region, Economia Politica. Journal of Analytical and Institutional Economics, vol. XXVII, no. 1, pp. 55-81.

Raymond W. - Mohenen P. - Palm F. - van der Loeff S. (2010), Persistence of Inno-vation in Dutch Manufacturing: Is It Spurious?, The Review of Economics and Statistics, vol. 92, no. 3, pp. 495-504.

Reinhart C. - Rogoff K. (2008), Is the 2007 Sub-prime Financial Crisis so Different? An International Historical Comparison, American Economic Review: Paper & Proceedings, vol. 98, no. 2, pp. 339-344.

Santangelo G. - Pini P. (2011), New HRM Practices and Exploitative Innovation: A Shopfloor Level Analysis, Industry and Innovation, vol. 18, no. 6, pp. 611-630.

Smith J. (2009), Cognitive Dissonance and the Overtaking Anomaly: Psychology in the Principal-Agent Relationship, Journal of Socio-Economics, vol. 38, no. 4, pp. 684-690.

Stata Manual (2009), Multiple-Imputation Reference Manual, Stata Press, College Sta-tion.

Stiglitz J. (1993), Endogenous Growth and Cycles, NBER Working Paper No. 4286, National Bureau of Economic Research, mimeo.

Zwick T. (2004), Employee Participation and Productivity, Labour Economics, vol. 11, no. 3, pp. 715-740.

— (2005), Continuing Vocational Training Forms and Establishment Productivity in Germany, German Economic Review, vol. 6, no. 2, pp. 155-184.

49

Summary: Innovation Strategies and Economic Crisis: Evidence from Firm-level Italian Data (J.E.L. L1, L23, L6, J53, O3)

The disruptive economic downturn of the period 2008-2009, forced industrial firms to implement strategies, in order to survive and to generate new competiveness sources. One of such strategic behav-iours regards the way of intervention on several innovation areas through different strategies. Disentan-gling the effect of pre-crisis adopted innovations, industrial relations quality and economic performance on such strategies may be of extreme relevance to analyse the nexus between the reactions to the chal-lenges brought by the crisis and the capacity of the firms to proactively tackle economic downturns.

The present work provides an empirical analysis on the basis of more than 500 Italian manufactur-ing firms located in Emilia-Romagna region. The results suggest the existence of strong relationships be-tween pre-crisis innovative activities and the capacity to react to the challenges brought by the crisis. This happens through innovative strategies, whose contents are mainly product (technological competitiveness), process and organisation/HRM innovative dimensions (cost competitiveness/efficiency gaining). Comple-mentary innovative activities emerge as a key factor. Industrial relations quality is also related to the stra-tegic reaction to the crisis: more participative industrial relations support the adoption of diversified types of innovation strategies. There is, instead, no evidence of a relation between past economic performance and innovation actions in the crisis.