NPM and post-NPM in the view of European administrative elites: Towards understanding the...

24
IRSPM Annual Conference Session on “Modelling Public Service Change” 30 March-1 April 2015-03-30 Birmingham, UK NPM and post-NPM in the view of European administrative elites: Towards understanding the relationship of public management reform doctrines György HAJNAL 1 and Miklós ROSTA 2 Work in progress! The paper seeks to explore and understand the relationship between two major contemporary public management reform doctrines: the New Public Management (NPM) and the post-NPM doctrines. In particular, we wish to identify and test competing hypotheses about the possible causal relationships between these two doctrines. These hypotheses centre around two questions. Firstly, whether post-NPM reforms are triggered by earlier NPM reforms (and, in particular, by the perceived problems and failures brought about by them) or, rather, by other factors largely unrelated to NPM. Secondly, whether post-NPM reforms can be conceived of as an anti-thesis of NPM aimed at undoing the changes of the previous epoch or, rather, post-NPM elements add up to a new, additional “layer” of public management reforms, leaving the earlier ones largely untouched. The empirical basis of the analysis is a recent large-scale questionnaire survey of senior public administration executives working in sixteen European countries. 1 Introduction 3 Our broad ambition pursued in this paper is to gain a better understanding of how different reform doctrines relate to one another; whether the ‘tides of reform’ (Light 1997) can be understood as (partly different, partly recurring, but mostly independent) answers to the ever-changing challenges brought about by societal, technological, economic, political, or other sorts of developments or, rather, they have some kind of a circular, swing-of-the-pendulum nature, whereby earlier reforms themselves fabricate the ‘problems’ to be ‘solved’ by later reformers. 1 Corvinus University of Budapest and Institute for Political Science, the Centre for Social Science of the Hungarian Academy of Sciences (IPS CSS HAS) 2 Corvinus University of Budapest 3 The research leading to these results has received funding from the European Community’s Seventh Framework Programme under grant agreement No. 266887 (Project COCOPS), Socio-economic Sciences & Humanities.

Transcript of NPM and post-NPM in the view of European administrative elites: Towards understanding the...

IRSPM Annual Conference

Session on “Modelling Public Service Change”

30 March-1 April 2015-03-30

Birmingham, UK

NPM and post-NPM in the view of European administrative elites: Towards understanding the relationship of public management reform

doctrines

György HAJNAL1 and Miklós ROSTA2

Work in progress!

The paper seeks to explore and understand the relationship between two major contemporary public management reform doctrines: the New Public Management (NPM) and the post-NPM doctrines. In particular, we wish to identify and test competing hypotheses about the possible causal relationships between these two doctrines. These hypotheses centre around two questions. Firstly, whether post-NPM reforms are triggered by earlier NPM reforms (and, in particular, by the perceived problems and failures brought about by them) or, rather, by other factors largely unrelated to NPM. Secondly, whether post-NPM reforms can be conceived of as an anti-thesis of NPM aimed at undoing the changes of the previous epoch or, rather, post-NPM elements add up to a new, additional “layer” of public management reforms, leaving the earlier ones largely untouched. The empirical basis of the analysis is a recent large-scale questionnaire survey of senior public administration executives working in sixteen European countries.

1 Introduction3 Our broad ambition pursued in this paper is to gain a better understanding of how different reform doctrines relate to one another; whether the ‘tides of reform’ (Light 1997) can be understood as (partly different, partly recurring, but mostly independent) answers to the ever-changing challenges brought about by societal, technological, economic, political, or other sorts of developments or, rather, they have some kind of a circular, swing-of-the-pendulum nature, whereby earlier reforms themselves fabricate the ‘problems’ to be ‘solved’ by later reformers.

1 Corvinus University of Budapest and Institute for Political Science, the Centre for Social Science of the Hungarian

Academy of Sciences (IPS CSS HAS)

2 Corvinus University of Budapest

3 The research leading to these results has received funding from the European Community’s Seventh Framework Programme under grant agreement No. 266887 (Project COCOPS), Socio-economic Sciences & Humanities.

There is an abundance of literature examining different aspects of this question focusing on different geographical locations and historical epochs. In this study, we set out to examine this broad issue in the particular context of two doctrines: that of the New Public Management (NPM) on the one hand, and of the post-NPM doctrine, on the other. In the past decade or so, a rapidly growing body of literature conceptualising, describing and analysing the so-called post-NPM reforms emerged. It seems however that there are important gaps and contradictions in available knowledge regarding how post-NPM relates to NPM; whether it can be conceived of as a ‘counter-movement’, restoring the (much of) the virtues having characterised the pre-NPM world or, rather, it is, if not entirely but overwhelmingly, ‘something else’, giving different answers to largely different questions of the day.

In the following section we first (in Section 2) define the core elements of our conceptual universe and then proceed to give a systematic analysis of authoritative scholarly contributions carrying some implications regarding our focal ambition. Section 3 outlines the research questions derived from the preceding analysis, and describes the data and the method used to answer those questions. Section 4 presents the empirical analyses and findings, and Section 5 summarises and discusses the conclusions established.

2 A review of the literature: Core concepts and rival hypotheses

2.1 Conceptualization of NPM and Post-NPM The main difficulties with our core concepts – NPM and post-NPM – is that they are typical examples of “conceptual stretching” – a loosening up of the concept in order to cover empirical cases that would otherwise not easily fit into it. As Kinder (2012:421) notes: “When that passage [NPM – GH & MR] was written only God and Christopher Hood understood it. Now only God understands it.” That is why Goldfinch & Wallis (2010:1110) also state that “if the coherence of NPM can be questioned, perhaps the putative post-NPM consensus can be even more so, with its grab-bag of catch-phrases, its re-labellings of old ideas and questionable novelty, and the layering of different and possibly incompatible reform agendas and rhetorics”. The problem, to which Goldfinch and Wallis (2010) refer

can only be solved if the definitions of these notions are scientifically justifiable.4

Goertz’s (2006) methodology of concept building posits that a concept should involve “a theoretical and empirical analysis of the object or phenomenon referred to by the word. A good concept draws distinctions that are important in the behaviour of the object. The central attributes that a definition refers to are those that prove relevant for hypotheses, explanations, and causal mechanisms” (Goertz, 2006:4). He distinguished three levels of a concept: a basic level, which is used in theoretical

propositions, the secondary level which is used to give constitutive dimensions of the basic level5 and finally the indicator/data level which is used to categorize whether or not a specific phenomenon, individual, or event falls under the concept (Goertz, 2006:6).

Following the above logical framework we reviewed selected pieces of literature and examined the way they conceptualized NPM and post-NPM. The result can be found in Table 1. It is interesting to note that, despite there is a lively debate about their meaning and impact on public administration, there are barely any “basic level” definitions. Authors only use the secondary and indicator / data levels, which mean they enumerate the main constitutive dimensions of NPM and post-NPM. Although Pollitt and Bouckaert (2011) in their seminal work give a very detailed and comprehensive description on New

4 As Dunleavy – Margetts – Bastow & Tinkler (2006:468) states “The result is that ‘NPM is a slippery label’ (Manning

2000; Savoie 1995). Different conceptualizations of NPM all stress different things”. Note that defining concepts is not an easy task in public management; see for instance how “slippery” is the notion of

“policy” or “governance” as Kettl (2002:61) and De Vries & Nemec (2013:12) state. These notions are at the heart of our field, still we do not have a clear and common understanding of them.

5 „In classic philosophical logic to define a concept is to give the conditions necessary and sufficeint for something to fit into the category. Each of these necessary conditions is a secondary-level dimension…”(Goertz, 2006:6)

Public Management, Neo-Weberian State (NWS) and New Public Governance (NPG), they do not give

a basic definition of these notions, instead they go around them (see Chapters 1 and 2).6

6 Pollitt and Bouckaert (2011:10) define NPM as follows: „For the purposes of this book we will assume that the NPM

is a two-level phenomenon. At the higher level, it is a general theory or doctrine that the public sector can be improved by the importation of business concepts, techniques, and values. [...] Then, at the more mundane level, NPM is a bundle of specific concepts and practices...”

Table 1.:Secondary and indicator / data level definitions of NPM and post-NPM

Authors NPM or Post-NPM Constitutive dimensions of basic level definition (secondary level

definition)

Pollitt - Bouckaert 2011

NPM

"greater emphasis on ‘performance’, especially through the measurement of outputs a preference for lean, flat, small, specialized (disaggregated) organizational forms over large, multi-functional forms a widespread substitution of contracts for hierarchical relations as the principal coordinating device a widespread injection of market-type mechanisms (MTMs) including competitive tendering, public sector league tables, and performance-related pay” (2011:10)

Dunleavy, Margetts, Bastow & Tinkler

(2006) NPM

Disaggregation, Competition, Incentivization under these integrated themes: Purchaser-provider separation, Agencification, Decoupling policy systems, Growth of quasi-government agencies, Separation out of micro-local agencies, Chunking up privatized industries, Corporatization and strong single organization management, De-professional, Competition by comparison, Improved performance measurement, League tables of agency performance, Competition Quasi-markets, Voucher schemes, Outsourcing, Compulsory market testing, Intragovernment contracting, Public/private sectoral polarization, Product market liberalization, Deregulation, Consumer-tagged financing, User control, Incentivization, Respecifying property rights, Light touch regulation, Capital market involvement in projects, Privatizing asset ownership, Anti-rent-seeking measures, De-privileging professions, Performance-related pay, PFI (private finance initiative), Public-private partnerships, Unified rate of return and discounting, Development of charging technologies, Valuing public sector equity, Mandatory efficiency dividends.

Grüning 2001 NPM

"Unequivocal characteristics of the NPM": Budget Cuts, Voucher, Accountability For Performance, Performance Auditing, Privatization, Customer Concept (One-Stop-Shops, Case management), Decentralization, Strategic Planning/ Management, Separation of Provision and Production, Competition, Performance Measurement, Changed Management Style, Contracting Out, Freedom to Manage (Flexibility), Improved Accounting, Personnel Management (Incentives), User Charges, Separation of Politics and Administration , Improved Financial Management, More Use of Information Technology

Hood 1991 NPM

4 megatrends: slow down or reverse govemment growth; privatization and quasi-privatization; automation,; more intemational agenda. 7 doctrinal components: 'Hands-on professional management' in the public sector; Explicit standards and measures of performance; Greater emphasis on output controls; Shift to greater competition in public sector; Stress on privatesector styles of management practice; Stress on greater discipline and parsimony in resource use

Hood 1995 NPM 7 doctrinal components of NPM see Hood 1991

Osborne & Gaebler (1992)

NPM

catalytic government (steering rather than rowing), community-owned (empowering rather than serving); competitive by injecting competition into service delivery, mission-driven instead of rule-driven, results-oriented, customer-driven, enterprising, anticipatory, decentralized and market-oriented.

Christensen & Lægreid (2007/a)

Post-NPM restructure central political– administrative apparatuses (centralization)

collaborative cultures

Christensen & Lægreid (2011)

Post-NPM political and administrative control

institutional autonomy

Jun (2009) NPM

Structural devolution and decentralization, Vertical coordination and autonomy within single agency, Managerialism and management techniques, Contractualism, privatization, and entrepreneurship, Market-driven techniques, competition, and citizens as customers, Against the Weberian theory, Deregulation and market transactions, Performance management and output

Jun (2009) Post-NPM Reducing fragmentation through structural integration, Asserting recentralization and re-regulation, Whole-of-government or joined-up government initiatives, Eliminating role ambiguity and creating clear role

relationships, Private-public partnerships, Increased centralization, capacity building, and coordination, Strengthening central political and administrative capacity, Paying attention to environmental, historical, and cultural elements

Zafra-Gómez, Bolívar, & Muñoz (2012)

NPM agencification, depolitication, managers the freedom to manage

Zafra-Gómez, Bolívar, & Muñoz (2012)

Post-NPM

cross-origanizational processes / culture

reinforce control and coordination

governance capacity (central)

Goldfinch & Wallis (2010)

NPM

putative market and private sector business practices

efficiency

outputs’, outcomes or results, and the reporting of such

Decentralized structures

putatively flexible and innovative staff

Motivation of public servants based on financial incentives for more dimensions see pages 1105-1110

Goldfinch & Wallis (2010)

Post-NPM

rejection of ‘agencification’ participation (broader notion of citizenship) (in contrast NPM client’ or ‘customer’ )

wider focus on ethical dimensions (‘public value’)

focus on broader and longer-term ‘outcomes’

Chapman & Duncan (2007)

NPM

desire to apply new institutional economics

application of contractual models high degree of decentralization of control and the disaggregation of organization structure

Funding and accounting system

application of management and marketing concepts

Chapman & Duncan (2007)

Post-NPM

managing for outcomes

more co-ordinated systems approach

reintegration of organization

Abandonment of policy – operations and purchaser – provider splits (buy-in)

De Vries & Nemec NPM performance measurement

(2013) emphasis on output and controls

contracting and outsourcing

disaggregation of and competition within the public sector

emphasis on the quality of service delivery

e-government

De Vries & Nemec (2013)

Post-NPM

centralization - decentralization

rule of law

reliability

openness and transparency

accountability and responsibility

participation

effectiveness

long term effectiveness

outcome focus

8

In order to pursue our research ambition we do not need to give a basic level definition of NPM or Post-NPM; we can grab the main characteristic of these two doctrines from aggregating the information from Table 1 and set those indicators with which we are able to measure whether a specific phenomenon falls under the concept of NPM or Post-NPM.

The main constitutive dimensions of NPM are the following (secondary level of definition): efficiency, business-like management tools, market-type-mechanisms (MTMs), decentralization, incentivization.

The indicator level of definition of NPM includes the following elements.

• Efficiency: Ensuring efficient use of resources; Public sector downsizing, Internal bureaucracy reduction / cutting red tape, achieving results

• Business-like management tools: Business/strategic planning, Quality management systems, Treatment of service users as customers

• Decentralization: Decentralisation of financial decisions; Decentralisation of staffing decisions; Creation of autonomous agencies or corporatization

• Incentivization: Performance related pay

• MTMs: Privatisation, Contracting out.

The main constitutive dimensions of post-NPM are the following (secondary level of definition): centralization, network-type-mechanisms (NTMs), outcome-focus, public ethos as specific public culture.

Indicator level of definition of Post-NPM include

• Centralization: Mergers of government organisations; Extending state provision into new areas; Decision making in my organisation has become more centralized; The power of politicians (vs. non-elected public officials) in the decision making process has increased

• NTMs: Getting public organisations to work together; Sets up special purpose bodies; Sets up a cross-cutting work/project group; Sets up a cross-cutting policy arrangement or program; Collaboration and cooperation among different public sector actors; Consults civil society organisations or interest groups

There were no indicators of outcome-focus and public ethos in our questionnaire.

2.2 Relationship between NPM and Post-NPM – the rival hypotheses

There is no consensus in the literature on the relationship between the two doctrines. Most of the authors assume that existence of NPM at t1 is a precondition of post-NPM reforms at t2 (cf. Alford & Hughes 2008; for the notation see Figure 1). As Christensen & Lægreid (2011:408) state “post-NPM reform wave [can be] seen partly as a reaction to the negative effects of NPM”. Goldfinch, S., & Wallis, J. (2010:1109) phrase it more sharply: „Of course, where NPM has not been adopted in any substantive sense – France and Germany and much of South and South East Asia for example – it is problematic to propose a move to a post-NPM world, when there has never been a NPM one.” If Post-NPM is a reaction to NPM then we should also analyse whether the trail-blazer countries of NPM (UK, New-Zealand, Australia) or the moderate reformers (like Norway, Germany, France or CEE countries) will adopt more easily and more eagerly post-NPM reforms. Some authors suggest that the forerunners of NPM are to adapt post-NPM (Bach & Bordogna 2011, Chapman & Duncan 2007, Christensen & Lægreid 2007, Gregory 2003, Halligan 2007), because there is a swing-of-the-pendulum effect which means the more NPM reform steps were realized, the more Post-NPM is necessary to cure / debug (Halligan 2007:224) the public administration system from the side-effects of NPM reform.

However, others such as Christensen & Lægreid (2007/a:8) put forward the opposite standpoint: „If NPM reforms are less entrenched it may be easier to try out post-NPM reforms, particular those

9

elements that are similar to some basic features of the «old public administration»” (see also: Christensen & Lægreid, 2011/a)7.

In addition to these two – diagonally opposite – standpoints there are some other trails in the literature. Some authors do not posit a link between (earlier) NPM reforms and (subsequent) post-NPM reforms, but suggest that, rather, it is external factors that determine which countries adopt post-NPM, such as terrorism, technical development, ‘wicked’ problems like pandemics, and tsunamis or other social, political and environmental problems (see: Christensen & Lægreid 2007). Yet another stream of the literature suggests that not only post-NPM is more and more influential in public management, but the NPM doctrine is also getting stronger. As Kinder (2012:422) argues: “NPM tools and techniques, (such as those described in the listening and learning framework), will grow in importance of the next period of public service management”.

Whether or not post-NPM is a reaction to NPM, a next important aspect of the NPM–post-NPM relationship is whether post-NPM can be expected to undo NPM reforms or, rather, it forms another layer on top of earlier reform waves. Looking at the processes in public administration systems many scholars – like Dunleavy, Margetts, Bastow & Tinkler (2006:468) – conclude that “...new public management (NPM), has essentially died in the water. [...] ...key parts of the NPM reform message have been reversed because they lead to policy disasters, and other large parts are stalled”. The same conclusion is reached by Drechsler & Kattel (2009): “...NPM is certainly dead – not as dead as a door-nail, perhaps, but among scholars not a viable option anymore” (for much softer wording see De Vries & Nemec 2013:13).

Nevertheless the prevalence of “layering” or “supplementary” statements in the literature is higher than those which envision the death of NPM doctrine. This is a robust trend represented in the works of Christensen & Lægreid (2007/a, 2008, 2011). They (2007/a:32) state: “[w]hat we see is more supplementary reforms than a process in which post-NPM reforms are replacing NPM reforms. New measures have been added without a substantial reduction in the old ones”. Many scholars agree with

this statement (Verhoest & Jann, 20088; Chapman & Duncan, 20079; Zafra-Gómez, Bolívar, & Muñoz,

201210 Pollitt 2003).

To sum up the review we can conclude that some contradictory statements can be found in the literature on the relationship between the NPM and the Post-NPM doctrines. The explanation could be that the relationship between the two is analysed somewhat simplified manner.

3 Research questions, data and method

3.1 Research questions and hypotheses The above review and analysis of NPM and post-NPM literature revealed that the general ambition to test hypotheses regarding the nature of the relationship between NPM and post-NPM is somewhat over-simplified. In fact, questions, claims and hypotheses regarding the relationship between can, and should, be clustered into two, rather distinct groups.

7 Christensen & Lægreid (2011/a:143) state: “First, Nordic countries were NPM laggards, so based on this the depth of

NPM reform will be more shallow and the use of post-NPM reforms probably more extensive.” 8 “Over the past decade, a second generation of administrative reforms with a post-NPM ‘flavour’ has emerged and

supplemented the NPM reforms” (Verhoest & Jann, 2008:94). 9 “At this stage (late-2005), while there have certainly been changes in wide areas of social and economic policy, the

NPM reforms have largely been left intact. There has been no suggestion of undoing the reforms ....” (Chapman & Duncan, 2007:12)

10 “…the appearance of new management models such as the post-NPM approach should not be aimed at replacing the postulates established under NPM, but rather at complementing them…” (Zafra-Gómez, Bolívar & Muñoz, 2012:736)

10

a) The first cluster relates to the extent to, and the ways in, which NPM contributed to or otherwise influenced the emergence of post-NPM reforms.

b) The second cluster of questions, claims and hypotheses regards the extent to, and the ways in, which post-NPM reforms influenced (in particular, eliminated) core features of earlier NPM reforms.

These two sets of relationships are graphically depicted in the below figure.

Figure1.: The analytical framework for examining the relationship between NPM and post-NPM

Our research questions target both of these clusters. Our approach followed is similar in that, on the basis of the literature review and analysis, we formulate specific and competing hypotheses regarding the given relationship, and we use our survey data to corroborate or disconfirm them

Research question #1

[H1:] The overwhelming majority of contributions on post-NPM posit that (i) NPM reforms have brought about a multitude of consequences, both intended and unintended, detrimental to the government’s capacity to tackle new and increasingly complex societal problems (Christensen & Lægreid, 2007).; that (ii) post-NPM reforms should be overwhelmingly seen as a response to these problems (Christensen & Lægreid, 2011; Halligan, 2007); and finally that (iii) post-NPM reforms, while contain new elements and cover new dimensions in comparison with NPM, are intended to ‘undo’ the extremes created by earlier NPM reforms (Drechsler 2005; Dunleavy, Margetts, Bastow & Tinkler, 2006). This view – underlying, among some other possibilities, the ‘swing-of-the-pendulum’ interpretation of post-NPM – implies a positive association between NPM and post-NPM reforms. That is, one can expect stronger post-NPM reforms where the amount/scope/depth of earlier NPM reforms had been larger.

[H2:] Contrary to the above view, other authors maintain that post-NPM reforms constitute a response to qualitatively new sorts of societal problems and challenges (see: Christensen & Lægreid, 2007). If this was the case then (earlier) NPM reforms would not be associated in either a positive or a negative way with (later) post-NPM measures; rather, the latter would reflect other factors unrelated to previous NPM reforms such as the severity of the new kinds of problems and the government’s capacity and willingness to react and adjust itself to the new circumstances.

[H3:] Finally, yet another set of contributions claim that post-NPM reforms can be expected in context where previous NPM reforms were modest or non-existent. These ‘old public administration’ type settings are the primary homes of true post-NPM reforms, whereas ‘NPM-ised’ context are overly alien to, and thus hinder the introduction of, post-NPM measures (Christensen & Lægreid, 2007/a and 2011/a). Such a causal pattern would imply a negative association between NPM and post-NPM reforms.

NPM reforms

Time and causation T1 T2 T3

post-NPM reforms

NPM-type features

RQ#1 RQ#2

11

Research question #2

Conceptualisations and interpretations of post-NPM diverge in terms of the second causal relationship (RQ #2) as well. These are somewhat related to, but are definitely neither identical with, nor fully determined by, the standpoints occupied in relationship to the first (#1) set of questions.

[H4:] The ‘swing-of-the-pendulum’ conceptualisation of post-NPM [H1] implies that post-NPM reforms strive to eliminate the main bulk of earlier NPM reform results. Whether they manage to do so or not is, however, another question. A reasonable hypothesis in this regard is that the more post-NPM reforms we have, the less one can expect an increase of NPM-type elements in public management practices (even to the contrary: the more one would expect a decrease in NPM type features). In other words, one possible hypothesis derived from theory is a negative association between post-NPM features on the one hand, and the increase in NPM-type features, on the other.

[H5:] If, however, post-NPM reforms are largely unrelated to NPM, targeting different problems and using different instruments, then one could expect little or zero association between the presence of post-NPM features and the change in NPM-type elements.

[H6:] Finally, it is conceivable that reforms of whatever type (i.e., whether NPM or post-NPM) reinforce each other, or are correlated positively as a result of some third set of factors causing administrative systems and system elements to reform (Brusis, 2003).. In other words, it may be that the ‘reformability’ or, possibly, the ‘inclination-to-reform-oneself’ – causing different administrative systems to reform themselves relatively easily, no matter what the actual direction and content of the reforms are – is a key factor explaining reform effects. If this was the case then we could expect a positive association between post-NPM features on the one hand, and the increase in NPM-type elements, on the other.

The research questions, the associated hypotheses, and the operationalisation of those hypotheses are summarised in the below table.

Variables involved H

ypo

thes

is #

Rel

ati

on

ship

im

pli

ed

Variable #1 (independent) Variable #2 (dependent)

Type time perspective

temporal focus

Type time perspective

temporal focus

...of the variable ...of the variable

RQ #1 NPM static ‘long ago’

(t1) post-NPM

static (system features)

‘now’ (t3)

H1 r>0

H2 r≈0

H3 r<0

RQ #2 post-NPM

static

‘nowadays’ (t2 and t3, supposing no change)

NPM dynamic

(changes in features)

‘nowadays’ (between t2 and t3)

H4 r<0

H5 r≈0

H6 r>0

Table 2.: Research questions and hypotheses, and their operationalisation

3.2 Data The empirical basis of this study is a questionnaire survey that took place in 2012 and 2013 as part of a large European comparative questionnaire survey implemented in the framework of the COCOPS FP7

12

research project. The response set contains data from 16 countries: Austria, Denmark, Estonia, France, Germany, Hungary, Ireland, Italy, Lithuania, Norway, Portugal, Serbia, Spain, Sweden, The Netherlands and the UK. The target populations were comparable sets of top level civil servants working in central government, and certain territorial offices of the health and the employment sector. The analyzed data is a subset of the total data set. In order to improve cross-country comparability respondents working in subnational level agencies were filtered out. The resulting data set contains 5999 observations in total, and 250 for Hungary. The number of observations per country is in the 147 to 880 range. Data includes, with some exception noted here, only respondents from central government ministries and central government agencies, at the first two hierarchical levels, and in order to ensure similar large sample sizes, also at third levels in the case of some countries. Additionally, in the case of Germany and Spain, given their rather low share/size of central government, we have also included ministries on state level (Germany), and ministries on regional level (Spain). (For a full documentation of the survey see Hammerschmid et al. 2013.)

The survey instrument contains data represented by more than 200 variables. The questionnaire items were analysed, compared and coded using the Qualrus computer assisted qualitative data analysis software. The data subset used in the current study contains – in addition to a number of socio-demographic and context variables – those items that were found to measure related either to NPM or to post-NPM. The number of items related to either NPM or post-NPM was a total of 38. (Eight of these variables contained information on various minute aspects of managerial autonomy. Since these variables were quite redundant on the one hand, and relatively numerous on the other – carrying the danger of disproportionately distorting scales – we collapsed the eight autonomy variables into two

factors: one representing administrative autonomy and the other one policy autonomy11.)

Thus altogether 32 variables (including the two autonomy factors) were used to build various scales measuring various features of NPM and post-NPM reforms (for a list of items see Annex I.). In a next step of analysis we grouped the variables according to three criteria:

1) whether they grasp NPM or post-NPM features (these categories were treated as not mutually exclusive);

2) their time perspective (static, measuring state and dynamic, measuring change unfolding in time; categories were treated as mutually exclusive and jointly exhaustive); and

3) level of analysis (individual level, the level of the respondent’s ‘own organisation’, and the level of the respondent’s ‘own policy area’; categories were mutually exclusive but jointly not exhaustive as one item was deemed to refer to an unspecified level of analysis).

Content-wise, the various items together seemed to give a good coverage of most reform features and elements customarily associated with NPM and post-NPM. However we have to keep in mind that the scales built from these items could be of limited validity and/or limited reliability for various reasons. Firstly, albeit theory may suggest that a given set of elements – say, centralisation, extension of state role, organisational aggregation and increasing political control – go hand in hand to constitute so-called ‘post-NPM’ reforms. But this may or may not be the empirical reality at the given geographical and time coordinates. Secondly, respondents’ perceptions may be imperfect because of, for example, the lack of information, limited ability to judge, or various sorts of respondent biases. In order to limit (though not exclude) the scope and possible severity of these problems we calculated Cronbach’s Alpha measures of scale reliability for each (sub)set of variables12 (see: Meier, Brudney & Bohte, 2012:21).

The results are summarised in the below table.

11 KMO=0.81, Varimax rotation. Total variance explained by the two factors: 63.85%

12 Cronbach’s Alpha can be understood as a sort of ’average correlation’ between scale items that supposedly measure various aspects of the same underlying construct – in our case, the NPM / post-NPM reforms. As a rule of thumb we used the following – conventional – threshold values: α<0.5 – unacceptable; 0.5< α < 0.6 – poor; 0.6< α<0.7 – acceptable; 0.7< α<0.8 – good.

13

NPM Post-NPM

Static Dynamic

All

perspectives Static Dynamic

All

perspectives

Level Alpha N Alpha N Alpha N Alpha N Alpha N Alpha N

Individual 0.68 2 - 0 0.68 2 - 1 - 0 - 1

Organizational 0.83 12 - 0 0.83 12 0.72 3 - 1 0.57 4

Policy - 0 0.7 9 0.55 9 - 0 0.41 3 0.41 3

All Levels 0.83 14 0.65 10 0.82 24 0.64 4 0.41 5 0.57 9

Table 3.: Features of scale measuring NPM and post-NPM reforms

As the data show the NPM scales behave quite well; the reliability of overall, 24-item scale is 0.82. It is only the policy level scale that is characterised by a poor scale reliability. The picture is markedly different on the right-hand, post-NPM side of the table. Firstly, the number of items is much lower here (a total of 9 items are related to post-NPM reforms). Secondly, the policy level scale – which happens to be almost identical with the dynamic, change oriented scale – is unacceptably unreliable. As a result, the overall scale reliability is 0.57, not achieving the ‘acceptable’ range.

In the analyses that follow we will use the following scales:

- NPMstatic: this scale measures the NPM features present at the time of the survey. It is composed of 14 items. These items happen to be include individual and organizational level items only.

- ∆NPMdynamic13:this scale measures the extent, to which the presence of NPM-type features has

increased (~7) or decreased (~1) in the years preceding the time of the survey. This scale is composed of 9 items exclusively referring to policy level phenomena.

- PNPMstatic: this scale measures the extent, to which post-NPM type features are present at the time of the survey. It is composed of 4 items.

Since available data are unbalanced (not all levels appear in all time perspectives and vice versa), in order to test the sensitivity of the results two additional scales were calculated and used:

- NPMAll: This scale contains all 24 NPM related items.

- PNPMAll: This scale contains all 9 post-NPM related items.

All scales are calculated as an unweighted mean of the component items.

3.3 Unit(s) of analysis Although ideally it would have been more appropriate to define the primary unit of analysis in the sub-section on research questions it is easier to do so after the above summary of available data.

Data in our data set represent perceptions and experiences of senior public sector executives working and living in different European countries. These perceptions and experiences refer, as we showed above, to different entities ranging from personal through organizational to policy level constructs.

13 The ∆ (delta) sign denotes, throughout the paper, that the variable measures a change taking place between two –

more or less defined – points in time, rather than a state characterising a given point in time. Occasionally we use the subscripts ‘dynamic’ to refer to the same feature; where both are used, as here, this leads to redundancy but we hope it makes the argument less easy to miss for the reader.

14

Researchers – especially those researching multiple countries – typically treat public management reforms, even if implicitly (e.g. analysing individual cases – organisations, sectors or areas – within a country) as a country level phenomenon. While we admit that this country-level focus is usually justified we emphasise that it is not exclusive. To the contrary: diverse directions and qualities of reforms may co-exist in a given country and a given time (period) at a lower, policy or organisational level analysis. Even more to that, within a large organisation different functional areas or hierarchical levels etc. may be characterised by different reform mixes.

Keeping all this as well as our core ambition – making a move towards a better understanding of how public management reform doctrines relate to one another – in mind in the current, first phase of the research (reported in this paper) our primary unit of analysis is the individual level. That is, we conceptualise and then operationalise public management reforms as events, experiences, judgements, perceived trends appearing in the life worlds of senior public administration executives. Of course, these can be aggregated at different levels such as organisation, policy area, country, or even country group. While such an aggregation opens new and very exciting avenues for investigation it implies however a broad range of additional conceptual and methodological difficulties too. Therefore – observing the volume limitations of this paper – we restrain ourselves, with the exception of some minimal, passing remarks, to the individual senior civil servant as the primary unit of analysis.

3.4 Method Apart from the multivariate statistical procedures used in creating the scales described above in the main part of our analysis we rely on simple bivariate statistical procedures, mostly bivariate correlations. These analyses were performed using IBM SPSS 22.0.

Specifically, we are going to calculate two correlation coefficients:

1) One between NPMt1 and ∆PNPMt2...t3 (representing the first, #1 arrow in Figure 1 above), and

2) One between PNPMt2...t3 and ∆NPMt3 (representing the second, #2 arrow in Figure 1).

While statistically the analysis is simple it involves certain – non-statistical – elements making it somewhat more complex than one would expect. At the heart of the problem there is our intention to examine, on the basis of a single cross-section data set, phenomena and their relationships having taken place in various points in time. Some of the data – the ones we called ‘dynamic’ variables and scales – do create a basis for making inferences regarding intertemporal changes. However, answering our first research question – exploring the relationship between NPM reforms having taken place ‘then’ and post-NPM reform elements we have ‘now’ – requires something more than this. Simply put: as we don’t have a measure for NPMt1 we have to use some proxy.

The only candidate for this purpose we have is NPMt3. However, the ‘amount’ of NPM type features might have changed between t1 and t3 – so much that one of our research questions aims exactly at finding out how much. There is one trick here however. Let’s assume

- that since the beginning of the post-NPM era (t2) NPM type reforms have not occurred, and therefore

- that the current amount of NPM type elements either equals to, or is an underestimation of, NPM type elements present prior to the post-NPM era (in mathematic symbols: NPMt3≤NPMt1).

These are, by and large, rather safe assumptions as we found no evidence in the literature that additional significant (or even noticeable) NPM type reforms have happened in Europe since the second half of the 2000’s.

15

If this assumption is true then our correlation analysis between NPMt3 and PNPMt2...t3 will be a conservative (lower) estimate of the real correlation between NPMt1 and PNPMt2...t3. In other words:

a) If correlation between the proxy NPMt1 measure and PNPM is found to be positive then the ‘true’ correlation is probably positive too but weaker (closer to 0);

b) If correlation between the proxy NPMt1 measure and PNPM is found to be negative then the ‘true’ correlation can only be even ‘more negative’ (closer to -1);

c) Finally, between the proxy NPMt1 measure and PNPM is found to be (close to) zero then the ‘true’ correlation is likely to be close to zero, as well.

By ways of example the argument for the first case (a) is illustrated in the below figure (the argument is analogous for the other two cases).

Figure 2.: The uni-directional character of bias in estimating NPMt1 using NPMt3 as a proxy measure (illustration)

That is, data points might have moved since t1 only (i) horizontally and (ii) to the ‘left’ (less NPM) direction. The probability / possibility of such a movement is predominantly present at those observations (data points) where the initial (t1) value of NPM had already been high (where it was low all the way there was simply not much room for further decrease).

This means that using the cross-sectional data available different hypotheses can be supported or rejected with different degrees of soundness: H2 and H3 can be tested and supported/rejected with a relatively high extent of certainty. In the case of H1, however, a positive correlation found between the NPMt1 proxy and PNPM may possibly conceal a much weaker correlation which is in reality would be closer to zero and/or statistically less significant (or even insignificant). This possibility will have to be taken into consideration when interpreting the results.

4 Findings

4.1 The role of NPM in triggering post-NPM reforms The research question targets one aspect of the relationship between NPM and post-NPM: whether stronger post-NPM measures are associated with stronger or, rather, weaker (or non-existent) NPM reform elements.

The result of the correlation analysis is presented below.

NPMt1 & NPMt3

PNPM

Legend

Actually observed data point using the NPMt3 proxy

Hypothetical data point using the ’true’ NPMt1 measure

Hypothetical (’real’) regression

Actually observed regression

16

Correlations

NPM NPM2 PNPM PNPM2

NPM / NPM Metric (14-item,

static)

Pearson Correlation 1 ,898** ,367** ,281**

Sig. (2-tailed) ,000 ,000 ,000

N 3362 2336 2910 2342

NPM2 / Extended NPM

metric (includes all NPM

related items)

Pearson Correlation ,898** 1 ,424** ,479**

Sig. (2-tailed) ,000 ,000 ,000

N 2336 2336 2262 2223

PNPM / post-NPM Metric (4-

item, static)

Pearson Correlation ,367** ,424** 1 ,758**

Sig. (2-tailed) ,000 ,000 ,000

N 2910 2262 4777 3808

PNPM2 /Extended post-PM

metric (includes all PNPM

related items)

Pearson Correlation ,281** ,479** ,758** 1

Sig. (2-tailed) ,000 ,000 ,000

N 2342 2223 3808 3808

**. Correlation is significant at the 0.01 level (2-tailed).

Table 4.: Correlations between static measures of NPM and post-NPM

From the point of view of RQ#1 the key finding is in the shaded part of the correlation matrix: there is a highly significant and unambiguous positive correlation (r=0.367 ) between the presence of NPM and post-NPM. Within the overall limitations of our data we can establish the following findings.

- One can reject, with a relatively high confidence, H2 as well as H3. That is, the true correlation between the (unobserved though approximated) ‘old-time’ NPM reforms on the one hand, and subsequent post-NPM reforms, on the other, are quite unlikely to be either negative or (close to) zero.

- Further, this result suggests that H3 is likely to be well-founded: we find evidence of stronger post-NPM type measures where NPM reforms had been stronger. However, at this point we have to exert caution because of the possibly biased proxy we use. As Figure 2 above shows positive correlation between PNPM and the NPMt3 proxy we use to approximate NPMt1 is possibly overstated. Therefore the true (unobserved) correlation may be weaker and/or statistically less significant than the one reported here.

In addition to the above key findings we can conclude that our results are relatively robust – that is, insensitive to the actual scale we use as a metric measuring NPM and post-NPM type reforms. The correlations calculated with the extended NPM and post-NPM and scales (involving substantially more items) are largely similar to the one found above both in terms of their direction, size and statistical significance14.

14 In order to check the consistency/robsutness of the pattern revealed (or the lack thereof) the correlation between the

NPM and the post-NPM variable was performed on other subpopulations as well. (i) A statistically significant, positive correlation between the two variables is present in 14 of the 16 countries (p-values usually below 0.01). The Netherlands and Denmark feature no significant correlation. (ii) The positive correlation is even stronger if data are aggregated on the country level (n=16 observations). For some further details see Annex II.

17

4.2 The effect of post-NPM reforms on pre-existing NPM-type elements The second research question asks what the effect of post-NPM reforms on NPM type elements is; whether they weaken/reverse earlier NPM reforms or, rather, simply add on top of earlier reforms as kind of additional ‘layer’.

The correlations between post-NPM reforms and the change in NPM type elements (that is, ∆NPM) are presented below.

Correlations

PNPM PNPM2 NPM_D

PNPM / post-NPM metric (4-

item, static)

Pearson Correlation 1 ,758** ,280**

Sig. (2-tailed) ,000 ,000

N 4777 3808 3663

PNPM2 / Extended post-

NPM metric (includes all

PNPM related items)

Pearson Correlation ,758** 1 ,553**

Sig. (2-tailed) ,000 ,000

N 3808 3808 3605

NPM_D / ∆NPM Metric

(dynamic time perspective)

Pearson Correlation ,280** ,553** 1

Sig. (2-tailed) ,000 ,000

N 3663 3605 4002

**. Correlation is significant at the 0.01 level (2-tailed).

Table 5.: The correlation between post-NPM and the change-in-NPM (∆NPM) scale

The key finding regarding our second research question (shaded yellow) is the highly significant and positive (r=+0.28) correlation between post-NPM reforms and the change in NPM type elements over the period preceding the time of the survey. Note that higher figures in the NPM_D (that is, ∆NPM) variable mean a larger increase (or smaller decrease) in the amount of NPM type elements. This finding means that more post-NPM reforms are associated with not a decrease but, rather, a (relative) increase if NPM type elements.

The validity of our key finding is corroborated by the ‘backup’ post-NPM scale (PNPM2). Using this extended measure of post-NPM reforms (which include dynamic variables/items as well) the correlation becomes even stronger.

Putting these pieces of information together one may conclude that post-NPM reforms did not weaken or eliminate NPM type elements installed in previous reform waves signified by New Public Management. H4 therefore – operationalising the swing-of-the-pendulum nature of the relationship between NPM and post-NPM – can be rejected with a large degree of certainty. Likewise, H5 – positing the independence of NPM and post-NPM reforms – can be rejected with a significant extent of certainty. Instead, our data support hypothesis H6 suggesting that subsystems have a capacity or inclination to embrace / absorb public management reforms either in smaller or in higher quantities, practically irrespective of their actual content (or, more exactly, this is the situation at least with regards

to two specific reform waves: NPM and post-NPM).15

15 Just like it was the case with RQ#1, in order to check the consistency/robsutness of the pattern revealed (or the lack

thereof) we analysed the correlation between the two variables – post-NPM and ∆NPM – on other subpopulations as well. (i) A statistically significant, positive correlation between the two variables is present in 15 of the 16 countries (p-values in all

18

5 Discussion and conclusions Our first research question explored one aspect of the relationship between NPM and post-NPM reforms: namely, the effect of (earlier) NPM reforms on (subsequent) post-NPM reforms. Our finding in this regard disconfirmed hypotheses H2 and H3 and – to an extent limited by the nature of our data – seems to have supported hypothesis H1. This finding agrees with the – probably – most frequent and accepted view of post-NPM whereby it is seen, at least to a significant extent, as a reaction to earlier NPM reforms and, in particular, the problems caused by them such as coordination (in particular, fragmentation and ‘siloisation’) problems, ‘hollowing out’ the state and weakening its capacity to govern, and built-in limitations of market type coordination instruments as well as markets themselves.

The pattern we found seems robust as it is present, when analysed individually, in practically all countries; moreover and possibly more importantly, our preliminary, rough analysis suggests that the relationship characterises not only individual level perceptions but countries as such too.

Our answer to the second research question is, especially in the view of the above finding, surprising. Contrary to what the – rather popular – ‘swing-of-the-pendulum’ interpretation of post-NPM reforms suggest stronger presence of post-NPM is associated with not a decrease but, rather, a (relative) increase in NPM-type features. (Note that the term ‘relative’ increase is used in order to refer to the fact that the perceptions and the way of their measurement do not allow for making a judgement whether NPM features ‘objectively’ increased or diminished. What we can conclude on their basis is that respondents perceiving the presence of more or stronger post-NPM features perceive, at the same time a dynamic, or change, leading to definitely more or stronger presence of NPM type features, compared to those respondents who perceived weaker post-NPM features.)

To us, this finding suggests that ‘reformability’ in general rather than leaning towards a particular mode of reform is a key feature of politico-administrative systems influencing reform trajectories.

15 cases are below 0.01), the only exception being Denmark. (ii) However, the pattern is not present if data are analysed at the country level (n=16 observations). For some further details see Annex II.

19

6 References

Alford, J., & Hughes, O. (2008). Public value pragmatism as the next phase of public management. The American Review of Public Administration. Volume 38 Number 2, 130-148

Bach, S., & Bordogna, L. (2011). Varieties of new public management or alternative models? The reform of public service employment relations in industrialized democracies. The International Journal of Human Resource Management, 22(11), 2281-2294.

Brusis, M. (2003). Developing Governance Capacity. A Review of Causes and effects. Paper for the Transformation Thinkers Conference (Vol. 30). Berlin, 30 November – 5 December 2003, Downloaded: http://dev.bertelsmann-transformation-index.de/fileadmin/pdf/Governance_Capacity.pdf

Chapman, J., & Duncan, G. (2007). Is there now a new ‘New Zealand model’?. Public Management Review, 9(1), 1-25.

Christensen, T., & Lægreid, P. (2007). The whole‐of‐government approach to public sector reform. Public Administration Review, 67(6), 1059-1066.

Christensen, T., & Lægreid, P. (2007/a). NPM and beyond–leadership, culture, and demography. Working Paper 3, Stein Rokkan Centre for Social Studies Unifob AS

Christensen, T., & Lægreid, P. (2011). Complexity and hybrid public administration—theoretical and empirical challenges. Public Organization Review, 11(4), 407-423.

Christensen, T., & Lægreid, P. (2011/a). Democracy and administrative policy: contrasting elements of New Public Management (NPM) and post-NPM. European Political Science Review, 3(01), 125-146.

Christensen, Tom, and Per Laegreid. 2008. “NPM and beyond: Structure, Culture and Demography.” International Review of Administrative Sciences 74 (1): 7–23.

De Vries, M., & Nemec, J. (2013). Public sector reform: an overview of recent literature and research on NPM and alternative paths. International Journal of Public Sector Management, 26(1), 4-16.

Drechsler, W. (2005). The rise and demise of the new public management. Post-autistic economics review, 33(14), 17-28.

Drechsler & Kattel (2009): Conclusion: Towards the Neo-Weberian State? Perhaps, but certainly Adieu, NPM! Downloaded: http://www.nispa.org/page_print.php?sid=860

Dunleavy, P., Margetts, H., Bastow, S., & Tinkler, J. (2006). New public management is dead—long live digital-era governance. Journal of public administration research and theory, 16(3), 467-494.

Goertz, G. (2006). Social science concepts: a user's guide. Princeton University Press.

Goldfinch, S., & Wallis, J. (2010). Two myths of convergence in public management reform. Public Administration, 88(4), 1099-1115.

Gregory, R. (2003). All the king's horses and all the king's men: Putting New Zealand's public sector back together again. International Public Management Review, 4(2), 41-58.

Grüning, G. (2001). Origin and theoretical basis of New Public Management. International public management journal, 4(1), 1-25.

Halligan, J. (2007). Reintegrating government in third generation reforms of Australia and New Zealand. Public Policy and Administration, 22(2), 217-238.

Hood, C. (1991). A public management for all seasons?. Public administration, 69(1), 3-19.

20

Hood, C. (1995). The “New Public Management” in the 1980s: variations on a theme. Accounting, organizations and society, 20(2), 93-109.

Jun, J. S. (2009). The Limits of Post‐New Public Management and Beyond. Public Administration Review, 69(1), 161-165.

Kettl, D. F. (2002). The transformation of governance: Public administration for twenty-first century America. JHU Press.

Kinder, T. (2012). Learning, innovating and performance in post-new public management of locally delivered public services. Public Management Review, 14(3), 403-428.

Manning, Nick. 2000. The new public management and its legacy. World Bank discussion note. Downloaded: http://www.mh-lectures.co.uk/npm_2.htm

Meier, K., Brudney, J., & Bohte, J. (2012). Applied statistics for public and nonprofit administration. Wadsworth, Cengage Learning.

Osborne, D., & Gaebler, T. (1992). Reinventing government: How the entrepreneurial spirit is transforming government. Reading Mass. Adison Wesley Public Comp.

Pollitt, C. (2003). The essential public manager. McGraw-Hill International.

Pollitt, C., & Bouckaert, G. (2011). Public Management Reform: A comparative analysis-new public management, governance, and the Neo-Weberian state. Oxford University Press.

Savoie, Donald. 1995. What is wrong with the new public management? Canadian Public Administration 38 (1): 112–21.

Verhoest, K., & Jann, W. (2008). The Governance, Autonomy and Coordination of Public Sector Organizations. Public Organization Review, 8(2), 93-96

Zafra-Gómez, J. L., Bolívar, M. P. R., & Muñoz, L. A. (2012). Contrasting New Public Management (NPM) Versus Post-NPM Through Financial Performance: A Cross-Sectional Analysis of Spanish Local Governments. Administration & Society, vol. 45 no. 6. 710-747

21

7 Annex I.: The list of variables / questionnaire items used to build scales measuring NPM and post-NPM reforms

NPM related items

I mainly understand my role as public sector executive as: 1 – Strongly disagree ... 7 – Strongly agree)

q5.3 Achieving results

q5.8 Ensuring efficient use of resources

To what extent are the following instruments used in your organisation? (1 – Not at all ... 7 – To a large extent)

q7.1 Business/strategic planning

q7.2 Customer/ user surveys

q7.4 Quality management systems

q7.6 Internal steering by contract

q7.7 Management by objectives and results

q7.9 Cost accounting systems

q7.10 Decentralisation of financial decisions

q7.11 Decentralisation of staffing decisions

q7.12 Performance related pay

When my organisation’s responsibility or interests conflict or overlap with that of other organisations, my organisation typically (1 – Strongly disagree ... 7 – Strongly agree)

q13.7 Consults civil society organisations or interest groups

How important are the following reform trends in your policy area? (1 – Not at all ... 7 – To a large extent)

q17.1 Public sector downsizing

q17.2 Citizen participation methods/initiatives

q17.3 Creation of autonomous agencies or corporatization

q17.4 Contracting out

q17.5 Focusing on outcomes and results

q17.7 Treatment of service users as customers

q17.9 Internal bureaucracy reduction / cutting red tape

q17.11 Privatisation

q17.13 External partnerships and strategic alliances

As a result of the fiscal crisis... (1 – Strongly disagree ... 7 – Strongly agree)

q22.5 The power of politicians (vs. non-elected public officials) in the decision making process has increased

In my position, I have the following degree of decision autonomy with regard to... (1 – Very low autonomy ... 7 – Very high autonomy).

F_Autonomy_administrative

22

F_Autonomy_policy

Post-NPM related items

I mainly understand my role as public sector executive as: 1 – Strongly disagree ... 7 – Strongly agree)

q5.2 Getting public organisations to work together

q13.7 Consults civil society organisations or interest groups

q13.3 Sets up special purpose bodies (more permanent)

q13.4 Sets up a cross-cutting work/project group (ad hoc, temporary)

q13.5 Sets up a cross-cutting policy arrangement or program

How important are the following reform trends in your policy area? (1 – Not at all ... 7 – To a large extent)

q17.6 Extending state provision into new areas

q17.8 Collaboration and cooperation among different public sector actors

q17.14 Mergers of government organisations

As a result of the fiscal crisis... (1 – Strongly disagree ... 7 – Strongly agree)

q22.2 Decision making in my organisation has become more centralized

q22.5 The power of politicians (vs. non-elected public officials) in the decision making process has increased

23

8 Annex II.: Country level analysis

8.1 NPM and post-NPM

Figure 3.: The relationship between static measures of NPM and post-NPM (data aggregated on the country level)

The correlation coefficient between the two variable it r=0.61 (p<0.01).

8.2 Post-NPM and the change in NPM (∆NPM)

24

Figure 4.: The relationship between the static post-NPM reform scale and the change-in-NPM (∆NPM) scale; data aggregated on the country level

The correlation between the two variables is – although positive but – statistically not significant, even if the outlier case of Portugal is removed from the analysis.