Linking and Integrating Corporatism and Consensus Democracy: Theory, Concepts and Evidence Author(s): Markus M. L.

Crepaz and Arend Lijphart Source: British Journal of Political Science, Vol. 25, No. 2 (Apr., 1995), pp. 281-288 Published by: Cambridge University Press Stable URL: http://www.jstor.org/stable/194089 . Accessed: 13/01/2011 18:35
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=cup. . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Cambridge University Press is collaborating with JSTOR to digitize, preserve and extend access to British Journal of Political Science.

http://www.jstor.org

Notes and Comments

TABLE 4

281

Regression of the Nine Characteristics Corporatism on Consensus Democracy* Beta

-

plus

Variablet
MINWIN ELECDIS CABDUR ISSUEDIM NPARTIES

T
5.61 5.23 6.71 4.08 2.25

Sig.T
0.00 0.00 0.00 0.00 0.06

0.34 0.27 0.24 0.22 0.17

CORPORAL
CONFLEX

0.07
- 0.03

1.41
- 0.86

0.20
0.42

GOVTAX REFEREND
UNICAMER

0.02 0.01 0.01

(Constant)
* Adjusted R2 = 0.99

0.68 0.19 0.38 1.13

0.52 0.85 0.72 0.29

t See Table 3 for explanationof abbreviations.

consensus democracies plus the composite scale of corporatism we can determine the relative weight of corporatism in explaining the variance of consociationalism (Table 4). Although the bi-variate relationship between consensus democracy and corporatism is ratherstrong (r, = 0.64), the relative weight of corporatism is so small that it does not fit in the model. If compared to other factors, corporatism explains a negligible part in the variance of the features of consensus democracy as defined by Lijphart. Both concepts should be analysed independently, together with other political variables, in order to assess the extent to which each does indeed influence democratic decision making. We have demonstrated that both concepts are strongly intertwined with the type and the organization of the political system and also strongly related to the complexion of government and parliament. Although the concepts are apparently similar, they do not have the same meaning and can be distinguished empirically in relation to the features of liberal democratic decision making.

Linking and Integrating Corporatism and Consensus Democracy: Theory, Concepts and Evidence
MARKUS M. L. CREPAZ AND AREND LIJPHART* Hans Keman's and Paul Pennings's critique ('Managing Political and Societal Conflict in Democracies: Do Consensus and Corporatism Matter?', this Journal, preceding pages) of our attempt to link corporatism and consensus democracy falls essentially into three parts. Their first criticism deals with the way we measured corporatism. They reject our 'composite' approach on the basis that different experts have different conceptual understandings of corporatism. Hence, they argue, it is unwarranted to add up these various scores. Secondly, they claim that our central relationship between consensus
* Department of Political Science, The University of Georgia; Department of Political Science, The University of California, San Diego, respectively.

282

Notes and Comments

democracy and corporatism is a function of our particular measure of corporatism and, in addition, driven by two outlying cases: Italy and Austria. Thirdly, they claim that corporatism and consensus democracy are two different phenomena, and that therefore, corporatism should not be integrated into the concept of consensus democracy. We shall address these three main criticisms in the order described.
ON MEASURING CORPORATISM, 'EXPERT' JUDGEMENTS AND VALIDITY

Keman's and Pennings's first criticism is that producing a composite measure of corporatism is unwarranted,since the various experts offering scores on corporatism had different meanings of that elusive concept in mind. They point to the difference between the institutional aspect of corporatism within the strategic actors, labour and capital, and the concertation aspect of corporatism between the strategic actors - a distinction which Schmitter referred to as 'corporatism 1' and 'corporatism 2' . To this distinction, they add a third aspect: social partnership. We are well aware that our various experts operationalized corporatism differently based on different features of corporatism. In our original article, we explicitly mentioned Schmitter's distinction between the institutional aspects of corporatism such as hierarchical structures, centralization, organization, etc. (vertical corporatism), and 'concertation' which refers to the relationship among the strategic actors (horizontal corporatism).2While this is a useful distinction, the difference between 'concertation' and 'social partnership', which Keman and Pennings add as a third conceptually distinct class of corporatism, is less convincing.3 We believe that the 'social partnership'and 'concertation' aspects are very closely related. In creating a broadly based measure of corporatism, we wanted to produce a measure that was as impartial as possible. We did not want to rely on one individual measure because that would have invited the criticism that we chose a particularmeasure simply because it best fitted our argument, and because this would have made our results dependent on that individual's concept of corporatism. Since the level of measurement encompassed nominal, ordinal and interval data, standardization was a suitable statistical procedure to create a composite score. The important concept of validity is defined in the social sciences as 'the extent to which our measures correspond to the concepts they are intended to reflect'.4 Based on 1 Phillippe C. Schmitter, 'Reflections oni Where the Theory of Neo-Corporatism Has Gone and Where the Praxis of Neo-Corporatism May Be Going', in Gerhard Lehmbruch and Phillippe C. Schmitter, eds, Patterns of Corporatist Policy Making (London: Sage, 1982). 2 Arend Lijphart and Markus M. L. Crepaz, 'Corporatism and Consensus Democracy in Eighteen Countries: Conceptual and Empirical Linkages', British Journal of Political Science, 21 (1991), 235-46. 3 For example, Gerhard Lehmbruch, 'Concertation and the Structure of Corporatist Networks', in John H. Goldthrope, ed., Order and Conflict in Contemporary Capitalism (Oxford: Clarendon Press, 1984), pp. 60-80 at p. 62, defines corporatist concertation as having two essential features: first, 'it involves not just a single organized interest with privileged access to government but rathera plurality of organizations usually representing antagonistic interests; and second, these organizations manage their conflicts and co-ordinate their action with that of government expressly in regard to the systemic (gesamtwirtschaftliche) requirements of the national economy'. Peter Katzenstein's (Small Staitesin World Markets, Industrial Policy in Europe (Ithaca, NY: Comell University Press, 1985), p. 32) definition of 'social partnership' seems to be closely related to Lehmbruch's notion of concertation: 'This ideology [social partnership]mitigates class conflict between business and unions; it integrates differing conceptions of group interest with vaguely held notions of the public interest'. 4 Jarol B. Manheim and Richard C. Rich, Empirical Political Analysis: Research Methods in Political Science (New York: Longman, 1991), p. 62.

Notes and Comments

283

this definition we measured exactly what we wanted to measure, which was to provide an overall average score of corporatism based on the combined wisdom of different experts on that subject, notwithstanding differences among those experts with respect to the conceptual meaning of corporatism. Thus Keman's and Pennings's claim that our index is 'meaningless in terms of validity' is simply incorrect.5 If the countries in our Table 1 are ordered from most to least corporatist the following order emerges: Austria, Norway, Sweden, Netherlands, Denmark, Switzerland, Germany, Finland, Belgium, Japan, Ireland, France, Italy, United Kingdom, Australia, New Zealand, Canada, United States.6 This is very much in accordance with the conventional wisdom on the relative strength of corporatism in various countries. In fact, even Keman's own corporatism score,7 which is not included in our composite measure correlates very strongly with our measure (r = 0.870). Frederic Pryor's measure of corporatism, also not included in our composite measure, correlates even more strongly with our scores (r= 0.963).8 More importantly, both elements of corporatism, i.e. within institutions and between actors, can be observed simultaneously in practice. In order for concertation between strategic actors to be successful, a strong institutional structure (capacity to co-operate) organizing the interests within strategic actors is a prerequisite. We say as much in our original article.9 Moreover, for corporatism to be successful the willingness of the functional bodies to co-operate is equally crucial. Thus, both elements, institutional capacity and willingness, have to be present in order successfully to intermediate between the major functional groups in industrial societies. If this is the case, it does not matter much whether the scores of the individual expert judgements sometimes measure 'institutionalization' and sometimes measure 'concertation'. In fact, precisely because we were aware that various experts attributeddifferent features to the concept of corporatism, our aim was to create an overall average of the experts' combined wisdom. Lehmbruch seems to agree with this assessment when he, after describing various distinct forms of corporatism such as sectoral corporatism and tripartism, states that, 'I have earlier (1982) argued that, since these different developments [of
corporatism] are interrelated, they should be integrated into a pluridimensional crossnational comparison and for the investigation of empirical regularities.' concept We

of corporatism. This should permit a ranking of countries for the purpose of concur with Lehmbruch's statement.

CONNECTING

CORPORATISM

WITH CONSENSUS

DEMOCRACY

The second criticism from Keman and Pennings deals with the empirical relationship between corporatism and consensus democracy and consists of two parts: first, they claim that our '(lump sum) measure [of corporatism] is among those that correlates strongest with [our] own index of consensus democracy'." Implicit in this criticism is the claim that the association between consensus democracy and corporatism is driven

Keman and Pennings, 'Managing Political and Societal Conflict', p. 5. Lijphart and Crepaz, 'Corporatism and Consensus Democracy', p. 239. 7 Latest scores used by Hans Keman, obtained by personal communication in 1994. x Frederic Pryor, 'Corporatism as an Economic System: A Review Essay', Journal of Comparative Economics, 12 (1988), 317-44, p. 326. 9 Lijphart and Crepaz, 'Corporatism and Consensus Democracy', p. 235-6. 10 Gerhard Lehmbruch, 'Concertation and the Structure of Corporatist Networks', p. 61 (emphasis added). l Keman and Pennings, 'Managing Political and Societal Conflict', p. 274. 6

284

Notes and Comments

by the fashion in which we measured corporatism. Secondly, they question the statistical robustness of the empirical model pointing out that Austria and Italy are outliers artificially enhancing the strength of the model. Let us first examine the strength of the relationship between corporatism and consensus democracy. A closer look at Table 1 shows that our composite measure does not correlate the strongest with consensus democracy. The individual measures of Crouch, Lehmbruch and Wilensky correlate more strongly than our composite measure of corporatism. Four of the twelve measurements display an association stronger than 0.5; another four observations display an association between 0.4 and 0.5 (among them Keman's own measure), while the last third (four observations) show associations smaller than 0.4.12 There is general agreement among social scientists that correlation coefficients between + / - 0.1 and + / - 0.4 denote 'weak' associations, between + / - 0.4 and + / - 0.7 denote 'medium' ones and between + / - 0.7 and + / - 0.9 denote 'strong' associations. Our corporatism measure does not stand out in any way in terms of its relation to consensus democracy. Eight of the twelve individual measures fall between the 0.4 and 0.7 range which is widely considered to be the 'medium' range of strength of association and three individual scores are actually higher than our composite score. Keman's own measure of corporatism, although not included in our composite measure, falls into the same 'medium' class as our own measure. Similarly, Pryor's score of corporatism, again not part of our original composite index, shows an association of almost the same magnitude as our own index of corporatism. The low correlation coefficient with the score of Marks is peculiar.13Marks looked mainly at neocorporatist incomes policy and created a weighted composite score according to the number of years in which countries displayed 'mixed', 'weak', 'moderate' or 'strong' incomes policies. His rankings indeed deviate from more conventional views on the strength of corporatist arrangements in different countries. For instance, he ranks Switzerland, France, Italy, the United States and Canada equally low on neocorporatist incomes policy. Great Britain is ranked before Denmark in neocorporatist incomes policy and Ireland is ranked between these two countries. Nevertheless, we decided to integrate Marks's score into our overall score, despite his 'unconventional' rankings which explain his relatively low correlation coefficient with consensus democracy. Keman's and Pennings's second point deals with the robustness of the empirical relationship between corporatism and consensus democracy. They argue that 'the location of the outliers Austria and Italy, being symmetrically aligned on either side of the regression line also strengthens the statistical outcome of the correlation coefficient ... The central relationship is less robust and overlapping than is claimed by Lijphart and Crepaz.'14 Their observation that two points symmetrically aligned on either side of a regression line should strengthen a relationship, indicates a misunderstanding of the concept of 'outliers'. They present an influence chart relying on two black 'dots' for identification of outliers. Fortunately, Belseley, Kuh and Welsch have devised a systematic way of not only detecting outliers but also leverage points.'5
12 Lijphart and Crepaz, 'Corporatism and Consensus Democracy', p. 238. 13 Gary Marks, 'Neocorporatism and Incomes Policy in Western Europe and North America', Comparative

Politics, 18 (1986), 253-78 at p. 274. 14 Keman and Pennings, 'Managing Political and Societal Conflict', p. 6. 15 David A. Belseley, Edwin Kuh and Roy E. Welsch, Regression Diagnostics: Identifying Influential Data and Sources of Collinearity (New York: Wiley, 1980).

Notes and Comments

T A B LE

285

Correlation Coefficients 'r' of the Twelve Original Individual Measures of Corporatism with Consensus Democracy Compared to the Lijphart/ Crepaz (1991) Composite Measure of Corporatism and Keman 's (1994) and Pryor's (1988) Measure of Corporatism in Descending Order N 18 18 18 18 18 17 18 18 17 17 18 17 17 15 15 Pearson's r product moment coefficient 0.597 0.596 0.583 0.568 0.546 0.503 0.492 0.490 0.465 0.465 0.453 0.395 0.332 0.328 0.128

'Expert judgement' Crouch (1985) Lehmbruch (1984) Wilensky (1981) Lijphart/Crepaz (1991) Pryor (1988) Lehner(1987, 1988) Czada (1983) Keman (1989) Bruno & Sachs (1985) Schott (1984) Schmidt (1982, 1986) Paloheimo (1984) Cameron (1984) Schmitter(1981) Marks (1986)

Note: N = numberof countries for which expertjudgements were provided. Keman's (latest measure, obtained by personal communication in 1994) and Pryor's measure (Frederic Pryor, 'Corporatism as an Economic System: A Review Essay', Journal of Comparative Economics, 12 (1988), 317-44 at p.326) were not part of the twelve original scores. They are shown here only for comparative purposes. In order to reduce confusion we inversed the polarity of the Marks (Gary Marks, 'Neocorporatism and Incomes Policy in Western Europe and North America', Comparative Politics, 18 (1986), 253-78 at p. 274) and Schmitter (Phillippe Schmitter, 'InterestIntermediationand Regime Governability in Contemporary Western Europe and North America' in Suzanne Berger, ed., Organizing Interests in WesternEurope: Pluralism, Corporatism,and the Transformationof Politics (Cambridge: Cambridge University Press, 1981), 287-330 at p. 294) scores to achieve consistent positive correlation coefficients. Leaving them in their original form would have changed the sign, but not the strength of the association. Table 2 provides the studentized residuals (residuals divided by their estimated standard deviations), which obviously follow a t-distribution, and the 'leverage points' (the values of the diagonal 'hat'-matrix) for eighteen countries. Leverage points identify outliers in the independent variable space, i.e. these are observations to which parameter estimates are particularly sensitive. Contrary to the claim by Keman and Pennings, only Austria proves to be an 'outlier', as can be seen from Table 2.16 The removal of Austria from the dataset does not reduce the strength of the relationship, in fact its omission increases the strength of the bivariate relationship. Since the 'studentized residuals' follow a t-distribution, Table 2 indeed indicates that Austria (2.564) is indeed an outlier (at the 0.05 level, two-tailed test, 16 degrees of freedom). Outliers affect the dependent variable space and have a stronger effect on the
16 We used the conventional 0.05 (two-tailed t-test) as a cut-off point to determine outliers. Since residuals do not follow a hypothesized direction, two-tailed t-tests are appropriate. The two-tailed t-value for 16 degrees of freedom is + 2.120 at the 0.05 level. Only Austria's studentized residual exceeds this critical cut-off point.

286
TABLE

Notes and Comments

2 Studentized Residuals and Leverage Points for Eighteen Countries residuals Studentized 'outliers'* - 0.615 2.564
- 0.123

Country Australia Austria

Belgium

Leveragepointst 0.105 0.064

0.079

Canada
Denmark Finland France Germany Ireland Italy Japan

- 0.631
0.099 0.519 0.785 0.616 0.133 1.963 0.087

0.185
0.090 0.182 0.057 0.056 0.087 0.119 0.056

Netherlands New Zealand Norway Sweden Switzerland UnitedKingdom UnitedStates

0.265 - 0.448 1.81 1.736 - 0.580 - 0.188 - 0.993

0.171 0.157 0.059 0.056 0.220 0.147 0.110

* 'Outliers' residuals residuals Sincestudentized standard deviation. dividedby theirestimated are the follow a t-distribution, criticalcut-offpointis + 2.120 (0.05 level, two tailed)for 16 degrees of freedom. Valueslargerthan0.222 indicate t Leveragepointsarethe diagonalelementsof the 'hat' matrix. Thereareno withhigh leverage,i.e. the slope will be sensitiveto theseobservations. observations for that observations exceed thatthreshold leveragepoints. intercept than on the parameter estimate.17 However, outliers do not exert sufficient leverage on the parameter estimates to change the coefficients as opposed to leverage points drastically. Belseley, Kuh and Welsch suggest as a critical threshold for leverage points 2p/N, where p stands for the number of estimated parameters (including the constant), N for the number of cases. 8 The critical threshold, then, in our case is 0.222. Table 2 indicates that none of the leverage points displays values larger than the critical threshold, indicating that none of the observations have leverage enough to sufficiently disturb the parameter estimates. Table 3 provides further evidence that the omission of Austria does not drastically change the effect on the parameterestimates. The table also shows that the central relationship between consensus democracy and corporatism holds (in fact, its strength and significance improves) if not only Austria but also Italy are removed from the dataset. Contrary to Keman's and Pennings's claim, inclusion of the Austrian and Italian case actually weakens, rather than strengthens, the relationship. It is not at all the case that Austria and Italy 'drive' the statistical relationship as they claim. In fact, removing either Austria or Italy, or both, strengthens the parameter estimates.'9 However, we decided
Diagnostics,p. 7. Belseley, KuhandWelsch,Regression Diagnostics,p. 17. Belseley, KuhandWelsch,Regression value( - 1.963)does not exceedthe criticalcut-offpoint residual 19As indicated above,Italy'sstudentized at 16 of ? 2.12 (two-tailed, degreesof freedom, the0.05 level),whichdoes notqualifyit foranoutlier.However, since KemanandPennings arguethatItalyis an outlier,andsince Italyis close to the cut-offpoint,we decided estimates. on with and the to report effectsof Italy'somissionindividually in conjunction Austria theparameter
18
17

Notes and Comments

TABLE

287

Change in the Parameter Estimates after Omitting Italy, Austria and Both Full model 0.568*
0.206 2.76

Statistics Coefficient
Standarderror t-test

Italy removed 0.668**

0.196 3.41

Austria removed 0.611**

0.178 3.43

ItalyandAustria removed 0.705**

0.163 4.33

R2 Adjusted Degreesof freedom

*

0.281 16

0.398 15

0.403 15

0.542 14

Note: Consensusdemocracyis the independent is variable. variable; corporatism the dependent

Significant at 0.05 level, two-tailed test. ** Significant at 0.01 level, two-tailed test.

to reportthe findings with all eighteen countries and not to omit observations selectively so as to create a stronger fit artificially. Tables 2 and 3 provide strong evidence for the robustness of our central relationship and invalidate Keman's and Pennings' s claim that the association between corporatism and consensus democracy is a statistical artefact.
CORPORATISM AS AN ELEMENT OF CONSENSUS DEMOCRACY

Thirdly, Keman and Pennings argue against integrating corporatism into the concept of consensus democracy by pointing out that these are two different phenomena. As evidence they present correlation coefficients indicating that corporatism and consensus democracy correlate distinctively with the complexion of government and parliament, and with cabinet duration. While this may be the case, we clearly did not intend to show, in the words of Keman and Pennings, that 'corporatism and consensus democracy were in fact and in theory two sides of the same coin'.20 Instead, we claimed that 'corporatism is the interest group system that goes together with the consensual type of democracy and its opposite, the "pluralist"interest group system, goes together with majoritarian democracy. ' After we had established an empirical link between corporatism and consensus democracy, the next logical step was to integrate the corporatist/pluralist system of interest representation into the cluster of consensus/majoritarian political institutions. Undoubtedly, the inclusive, accommodative and consensus-oriented corporatist policymaking process is conceptually related to the five more 'political' elements of consensus democracy: the sharing of executive power; balanced executive-legislative relations; a high degree of electoral proportionality; multi-party systems and multiple issue dimensions of partisan conflict. Similarly, the competitive, atomistic and exclusive pluralist policy-making style is conceptually related to political structures which encourage little sharing of executive power (minimal winning cabinets), strong executives, electoral disproportionality, and two-party systems characterized by a single issue dimension. We believe that the affinity between corporatism and consensus democracy, and between pluralism and majoritarianism is compelling, and there is no reason why the interest group system should not be linked to the political and
20
21

Keman and Pennings, 'Managing Political and Societal Conflict', p. 6. Lijphart and Crepaz, 'Corporatism and Consensus Democracy', p. 235.

288
TABLE

Notes and Comments

4 Correlation Matrix of the Five Constitutional Democracy with Corporatism Elements of Consensus

1 1. % minimal winningcoalitions 2. Executive dominance 3. Electoral disproportionality of 4. Number issue dimensions 5. Effectivenumber of parties
6. Corporatism (L/C)

1 0.377 0.305 -0.618 -0.735

- 0.447

1 0.405 -0.394 -0.619

- 0.126

1 -0.167 -0.486
- 0.661

1 0.717
0.488

1
0.451 1

elementsof consensusdemocracyare based on ArendLijphart, Note: The five constitutional measure. YaleUniversity Press,1984).L/Cis theLijphart/Crepaz Haven,Conn.: ( 1991)corporatism
Democracies: Patterns of Majoritarian and Consensus Governmentin TwentyOne Countries (New

constitutional structures. In our original article, we presented the connection between consensus democracy and corporatism in its aggregated form. Do we also find meaningful associations between corporatism and its disaggregated elements of consensus democracy? How do the five original individual elements of consensus democracy, described above, associate with corporatism? Table 4 indicates that corporatism correlates with each individual element of consensus democracy in the hypothesized way, i.e. it is consistent with the argument made above that corporatism can logically be integrated into the cluster of consensus democracy. The negative correlation between corporatism and the percentage in which minimal winning coalitions were in power ( - 0.460) indicates that corporatism can be thought of as an element of consensus democracy. The same logic applies to the other negative correlation coefficients between corporatism and executive dominance and electoral disproportionality. The more corporatist a country, the less dominant is the executive (although the coefficient is a weak one, - 0.126). Also, the more corporatist a country, the less disproportional is its electoral system (- 0.661). Moreover, the number of issue dimensions in partisan conflict and the effective number of parties also correlate in a predictable way with corporatism. Corporatist countries display a higher number of issue dimensions of partisan conflict (0.488) while also showing a higher effective number of parties (0.451). Overall, the correlation coefficients between corporatism and the other five elements of consensus democracy (in the bottom row of Table 4) are about as strong as the ten correlation coefficients among these other five elements; the means of the absolute values are 0.435 and 0.482 respectively, and the medians are 0.451 and 0.446 respectively. Table 4 shows a consistent and predictable relationship of corporatism with each of the disaggregated elements of consensus democracy. Given this affinity between corporatism and the five constitutive elements of consensus democracy we perceived it as logical to integrate it as an additional element of consensus democracy, and we believe that integrating all six into a more comprehensive concept of consensus democracy leads to a richer understanding of the process by which private desires are turned into actual public policies.