CC BY-NC-ND
26
оо
THE JOURNAL OF SOCIAL POLICY STUDIES_
ЖУРНАЛ
ИССЛЕДОВАНИЙ СОЦИАЛЬНОЙ
ПОЛИТИКИ • ••
ARTICLES IN ENGLISH
Daria Salnikova
DETERMINANTS OF SOCIAL SPENDING IN EUROPEAN UNION MEMBER-STATES: "GOVERNANCE MATTERS" VS "POLITICS MATTERS"
The contribution of this article is to test alternative approaches to explaining variation in the generosity of welfare spending across states. We use panel data from 27 European Union member-states over the period 1990-2011. We rely on the experience of previous research papers and use social expenditure (% of GDP) as a measure of welfare generosity. We also use data on left-wing parties as a percentage of parliamentary seats for all governmental parties and the Quality of Government indicator, provided by the International Country Risk Guide. Regression analysis, including mixed-effects modelling, demonstrates that there is a positive relationship between institutional and welfare performance. However, an analysis of subsamples within the dataset shows that the effect of the quality of government on welfare generosity varies across states: the strong positive effect holds only for post-communist states. Our study verifies that power resource theory is losing its explanatory potential, while the role of institutional performance in explaining the generosity of the welfare state is increasing.
Key words: welfare state, institutions, quality of government, governance, power resource theory
During the period of welfare state consolidation, power resource theory has been the main approach to explaining cross-national variations in social expenditure. According to this theory, the alignment of political forces determines the extent
Daria V. Salnikova - lecturer, Higher School of Economics, Department of Higher Mathematics, research-assistant, Higher School of Economics, Laboratory of Comparative Social Research Moscow, Russian Federation. Email: darysalnikova@yandex.ru; dsalnikova@hse.ru
© Журнал исследований социальной политики. Том 13. № 1
of welfare state generosity; within this concept political institutions do not have their own explanatory power. However, as Pierson noted, there is no evidence that the decline of left parties has had a significant effect on social spending (Pier-son 1996). This fact implies that power resource theory has lost its explanatory potential. Proponents of the alternative institutionalist approach state that what matters are "patterns of governance" (ibid.: 152). Empirical evidence from OECD countries was presented in a paper by Rothstein et al. (2012). The authors note that prior empirical studies on the welfare state did not address the problem of the low quality of government. However, it would be wrong to argue that the effect of institutions has been underesti-mated. The new institutionalist studies focus on the variation of government social expenditure across different types of political systems and electoral rules. A number of empirical research papers confirm that the share of social spending depends on the number of parties in government. For example, Bawn and Rosenbluth (2006) show this dependence on time-series cross-sectional data from European states (the period is 1970-1998): states with more parties in government tend to have a more generous welfare system. The findings of Pers-son and Tabellini (2004) show that in presidential regimes the distribution system targets minority groups. At the same time, parliamentary coalitions with a strong degree of cohesion in legislative voting tend to be more generous with social spending. In their paper (ibid.) they also address the effect of electoral systems on social expenditure. The authors show that elections based on proportional representation increase welfare state generosity. The study of the relationship between electoral systems and social expenditure by Milesi-Ferretti et al. (2002) emphasises spending composition. The authors single out transfer and universal public goods. Proportional electoral systems lead to a larger share of transfers, which are targeted across different social groups. In contrast, majoritarian electoral systems induce more public goods, which are distributed not to definite social groups, but along geographical units. There were also attempts to explain the power resource approach with reference to differences in electoral systems. Iversen and Soskice (2009) argue that proportional systems (in contrast to majoritarian electoral systems) promote centre-left coalitions and strengthen median voter support for welfare state generosity. To make it clear, prior research dealt with the classification of institutions rather than institutional performance. The lack of research on the relationship between the welfare state and the quality of government is caused by difficulties in defining the extent of good governance. The problem is that in a number of cases good governance is defined in an excessively broad way. Since there are no clear boundaries between good governance, on the one hand, and decentralisation as well as economic well-being, on the other, it is difficult to single out distinctive features of good governance. But despite the variety of interpretations, all conceptualizations of good governance focus on the institutional structure, the so-called "rules of the game". In our research we rely on the conceptualisation proposed by Rothstein and Teorell, who define governance as "the impartiality (namely, implementation of law does not depend on personal preferences) of institutions that exercise governance authority"
(2008: 165). We believe that this concept will help us avoid excessively broad definitions and, as a result, broad approaches to operationalisation of governance.
Rothstein and his co-authors (Rothstein et al. 2012) underline the importance of testing whether the results of their empirical analysis are robust to changes in the sample, and hypothesise that the positive effect of the quality of government on the size of the welfare state is substantial in non-OECD countries, while political mobilisation loses its explanatory power. Our article attempts to find out if the alignment of political forces and institutional performance have an effect on the enerosity of the welfare state in EU member-states, and if they do, clarify whether this effect is positive or negative and whether it varies across subsamples. In contrast to prior research in this field, our sample is more heterogeneous and includes post-communist EU member-states (earlier studies focused on developed liberal democracies (Korpi, Palme 2003; Rothstein et al. 2012)), the time period under study is longer and allows us to track the relationship between the variables in the first decade of the 21st century. Moreover, the present study takes into consideration such factors as welfare models and the type of electoral system. We argue that this is important when running robustness checks because these factors can cause possible variation in the effect of the quality of government and political forces on social spending. In addition to these distinctions from prior studies, we use mixed-effects regression modelling. This method has a number of advantages when used to analyse dynamics.
Hypotheses
This study tests two main hypotheses:
H1. The quality of government and the generosity of the welfare state are
positively related in European Union member-states.
H2. The alignment of political forces has little explanatory power for the generosity of the welfare state in the European Union member-states.
We find it justified to posit hypotheses in accordance with Pierson and Roth-stein's theoretical statements. In particular, it is the presence of post-communist countries in our sample that corroborates the above hypotheses. The ongoing process of democratisation deserves special attention. As Orenstein (2008) states, democratisation in countries of Central and Eastern Europe weakened the party system. Both the needy and the wealthy expected a rise in social expenditure, but fluctuations in the alignment of political forces could not guarantee a steady increase in welfare generosity. The strengthening of democratic institutions has played a significant role in the reform process in respect of welfare policy. The other reason to suppose that power resource theory is not able to explain variation and changes in social spending in post-communist countries is the nature of exchange procedures between politicians and voters (Kitschelt 2000). The experience of parties in Eastern Europe gives evidence in support of the clientelist party model
in newly democratic countries (Biezen 2000 cit. in Ishiyama 2001). In contrast to programmatic parties, the clientelist strategy presupposes not universal but selective incentives: benefits are assigned only to those voters who support the party. As regards the factors influencing welfare generosity in non-post-communist countries, prior empirical research concludes that good governance matters more than the alignment of political forces. We do not have grounds to question these results. Our study aims to test whether they hold for our sample, model specifications and methods, which are distinctive from those applied in the previous research papers.
Data and measures
This study uses data on 27 European Union member-states (Austria, Belgium, Bulgaria, Cyprus, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden and the United Kingdom) during the period 1990-2011. We deal with this time period because the Comparative Political Data Set 3 does not provide the essential political and institutional variables before 1990.
Dependent variable
This research uses social expenditure (percentage of GDP) as an indicator of welfare generosity. The source of this variable is the Eurostat database. We are aware that this measure has some drawbacks. This indicator has been criticised for not reflecting the structure of individual entitlements, i.e. The effect of providing benefits on individual recipients (Scruggs, Allan 2006). Castles (1994) raises the concern that the expenditure approach to measuring the size of welfare states does not allow us to differentiate between social spending favouring privileged and middle-status groups and social assistance to vulnerable groups. However, due to its availability for a large number of countries and a long time period, social expenditure is widely used in empirical studies attempting to operationalise the size of the welfare state and to classify welfare regimes. We rely on this experience and believe social expenditure (% of GDP) to be a plausible measure of welfare generosity.
Key independent variables
To verify the power resource approach to explaining variation in social spending, we use data on left-wing parties as a percentage of parliamentary seats for all governmental parties. This variable is introduced in the Comparative Political Data Set 3.
The absence of consensus regarding theoretical conceptualisations of governance causes difficulties in measuring the quality of government. The construction of multidimensional indices allows us to take different conceptualisations into consideration and thus to some degree alleviates the problem of operati-onalising "governance". The broad country and time coverage of the Quality of
Government indicator, provided by the International Country Risk Guide (ICRG), explains the frequent use of this measure in empirical research. The QoG Basic Dataset (University of Gothenburg) contains data on the governance index, which is derived from the mean of three ICRG dimensions, namely "Law and Order" (an assessment of both the impartiality of the legal system and popular observance of the law), "Bureaucracy quality" (assessment of bureaucratic autonomy and institutional strength) and "Corruption" (the ICRG emphasises insidious forms of corruption, such as nepotism and close ties between business and political actors). The ICRG index scores range from 0 (the lowest value of QoG, high-risk observations) to 1 (the highest value of QoG, low-risk observations).
The control variables and indicators which are used in our study for robustness checks are as follows:
• the logarithm of GDP per capita (World Development Indicators);
• the annual percentage growth rate of GDP per capita (World Development Indicators);
• the openness to trade indicator calculated as the sum of exports and imports divided by real GDP per capita (Penn World Table);
• the indicator of political regime, constructed as a result of averaging Freedom House and Polity variables, this indicator ranges from 0 (the least democratic countries) to 10 (corresponds to the most democratic countries);
• federalism (the Comparative Political Data Set 3) is presented as a dummy variable, where the value 1 stands for federalism, 0 - otherwise;
• the type of electoral system used in the state-level lower house legislative elections (Democratic Electoral Systems around the World, Bormann & Golder);
• a set of dummies, each with a value of 1, meaning that a given country relates to one of the five welfare regimes (social-democratic, corporatist, liberal, southern, and hybrid regime presented by post-communist countries).
Methods
In our paper we ran OLS regression models with panel-corrected standard errors and adjustment for autocorrelation of the first order. Panel-corrected standard errors were suggested by Beck and Katz (1995) as an alternative to the FGLS (feasible generalised least squares) estimator. FGLS for time-series cross-section data underestimate true parameter variability. Moreover, FGLS properties in finite samples are unknown. Panel-corrected standard errors account for contemporaneous correlation across units and heteroscedasticity. The generic model specification is as follows:
SocialSpendingit=$ o + PIXl(t_ I} + P2Ci(t_ I} + yt_ i + e i(t_
where index (i) stands for a country, and (t) - for a time period. Independent variables are 1-year lagged to alleviate the endogeneity problem. X is a vector of key
predictor variables (percentage of left-wing parties, quality of government), C is a vector of control variables, y is a set of year effects (7-9 model specifications account for year effects), and e is a vector of errors. Model 6 introduces an interaction term between the quality of government and the percentage of left-wing parties. Models 9-12 include additional dummies for groups of countries (post-communist countries in model 9 and welfare regimes in models 10-12) and corresponding interaction terms between these dum-mies and independent variables.
We also use mixed-effects modelling as a robustness check. Panel data can be viewed as hierarchical data with time periods nested within countries. Mixed-effects models adjust for possible differences in the relationship between explanatory variables and a dependent variable across countries. Therefore, mixed-effects models make it convenient to track individual trajectories, capture deviations from the average trend. Both time-varying and time-invariant independent variables can be used in these models. Besides, mixed-effects models do not require complete data and the same number of time points for all units. This advantage makes this class of models useful when working with missing and unbalanced datasets.
The following equation shows the mixed-effects model specification used in our study:
SocialSpendingti = y 00 + y10Trend(t_ 1)t + y^efiParties<t_ 1)i<t_1H + Y3C,QoG(t_ 1)t
+ Y 40C<t- 1), + U0, + u1,Trend<t_ 1), + U2LefiParties(t_ 1)i + u3lQoG<t_ 1)i + e <t- 1)h
where y stands for fixed effects, u stands for random effects (in addition to the random intercept; we include random effects for a trend variable, and key independent variables, namely the percentage of left-wing parties and quality of government); C is a vector of control variables; e is the first-level error (deviations of time periods from the predicted value).
Results
To test whether the alignment of political forces has an effect on welfare generosity we ran several regression models (panel-corrected standard errors and adjustment for autocorrelation of the first order) with different sets of control variables. Table 1 summarises the results of estimating these models. The political regime and logarithm of GDP per capita variables are not included in the model simultaneously due to high multicollinearity. Higher social expenditure is accompanied by the diversion of resources from productive private sectors and inefficient resource allocation, which explains the negative relationship between economic growth and the generosity of the welfare state. The regression coefficient for the variable "openness to trade" is expected to be negative. This indicator reflects the effect of globalisation on social spending. Globalisation erodes the state sovereignty and engages welfare states in the process of neoliberal restructuring to save their competitiveness in the global economy. Numerous empirical studies (e.g.,
Burgoon 2001; Garrett, Mitchell 2001) show that trade openness decreases social spending and makes the retrenchment of welfare policy more likely. The extent of resistance to globalisation effects explains the positive relationship between democratic political regimes and social spending. For example, the empirical analysis of panel data on 57 countries by Rudra and Haggard (2005) confirms that in comparison to authoritarian countries, democratic political regimes are less sensitive to the negative effect of economic openness on the generosity of the welfare state.
Table 1.
Relationship between social spending and alignment of political forces
Dependent variable - Social expenditure (% of GDP) (1) (2) (3) (4)
left-wing parties (% of all parliamentary seats) -0.0002 (0.002) -0.001 (0.002) -0.0003 (0.002) -0.001 (0.002)
logarithm of GDP per capita 2 517*** (0.54) 3.136*** (0.564) 2.967*** (0.549)
growth of GDP per capita -0.095*** (0.029) -0.083*** (0.029) -0.076** (0.031) -0.081*** (0.029)
openness to trade -0.017*** (0.004) -0.011*** (.004) -0.018*** (0.004)
political regime 0.815*** (0.254)
federalism 2.231*** (0.299) 1.449*** (0.366)
intercept -10.31* (5.36) -14.805*** (5.576) 7.643*** (2.433) -13.241** (5.386)
R-squared 0.6598 0.6743 0.6559 0.6829
Number of observations 525 525 523 523
Note. * - p < 0.1; ** - p < 0.05; *** - p < 0.01; panel-corrected standard errors are given in brackets.
The results concerning the relationship between federalism and the generosity of the welfare state are in conflict with the conclusions made in comparative studies on the "old politics" (the "old politics" involves the period of welfare state consolidation and the subsequent growth of social expenditure). According to these studies, federalism is an impediment to the increase in the generosity of the welfare state. Veto points in federal systems create more opportunities to block redistributi-
ve legislation. However, this argument becomes less unambiguous in the context of "new politics" (to use Pierson's term). Country-specific characteristics explain the observed positive relationship between federalism and social spending in our sample. For example, Austria and Germany demonstrate the coexistence of generous welfare policy and federalism. It is important to point out that the establishment of the welfare state in Austria preceded federalism. Besides, the Länder did not have formal opportunities to exercise vetoes (Obinger 2005), which prevented the impeding effect of federalism on the welfare state. Manow (2005) emphasises the strong feedback effects of the welfare state on federalism in Germany. The welfare state in Germany was established to support further unification. Thus, substantial redistribution prevents inter-regional competition.
Table 2.
Alignment of political forces and quality of government as factors of welfare generosity
Dependent variable - Social expenditure (% of GDP) (5) (6)
left-wing parties (% of all parliamentary seats) 0.0004 0.014
(0.002) (0.009)
quality of government: ICRG index 5.185*** 5.818***
(1.324) (1.364)
growth of GDP per capita -0.085** -0.086**
(0.035) (0.035)
openness to trade -0.011*** -0.011***
(0.004) (0.004)
federalism 2.064*** 2.049***
(0.265) (0.265)
interaction term -0.017
(QoGxpercentage of left-wing parties) (0.011)
Intercept 11.425*** 10.952***
(1.324) (1.112)
R-squared 0.6763 0.6776
Number of observations 504 504
The effect of left-wing parties (as a % of all parliamentary seats) on social spending turns out to be statistically insignificant. As we see, these results demonstrate consistency across different specifications of the model.
To test whether institutions matter for the variation in social expenditure, we include the quality of government (ICRG index) as a predictor variable (see Table 2). Since the quality of government is highly correlated with the logarithm of GDP per capita and the political regime, which causes substantial multicollinearity, we did not use these indicators as control variables in the fifth model specification. The estimates in Table 2 demonstrate that the quality of government has a positive and significant effect on social expenditure. At the same time the relationship between welfare state generosity and the alignment of political forces remains statistically insignificant, which is consistent with our hypothesis.
Estimates of model (6), presented in Table 2, and marginal effects1 show that the effect of government composition on social expenditure does not depend on the values of the quality of government, and vice versa.
Next, we address the question concerning the robustness of our findings to changes in model specification, sample units and methods. We are aware that our results can be driven by time dynamics. Estimates of model specification (7) demonstrate robustness to the inclusion of year effects (see Table 3).
Proportional electoral systems represent a broader range of interests than majoritarian ones and increase the likelihood of alliances between working and middle class voters, which favours generous redistributive policies (Immergut 2010). The findings of empirical studies (Persson, Tabellini 2004; Milesi-Ferret-ti et al. 2002) confirm that proportional elections produce more generous welfare states. In contrast to majoritarian elections, proportional electoral systems are characterised by high party discipline. Given these differences between electoral systems we find it important to test whether the absence of a significant relationship between the alignment of political forces and social expenditure holds for the subsample consisting only of proportional system cases. Table 3 (model 8) displays regression estimates for this subsample (the model is estimated controlling for year effects). These results are similar to those for the whole sample. We do not have a reason to think that the insignificant effect of left-wing parties on the generosity of the welfare state is driven by majoritarian electoral system cases with low party discipline and low social expenditure.
For the next robustness check, we generate two interaction terms: between a dummy variable for post-communist countries and each of our key independent variables, namely, the quality of government predictor and the left-wing party percentage of all parliamentary seats. The estimates of model 9 in Table 3 report that the preceding results concerning the significant positive relationship between good governance and the generosity of the welfare state were driven by post-communist countries. In model 9, the coefficient for the quality of government inde-
1 Calculations of marginal effects are available on request.
pendent variable is interpreted as the "effect" of good governance on social expenditure for non-post-communist countries only, holding other variables fixed. As we see, this coefficient is positive but insignificant. In post-communist countries one unit increase in the "quality of government" variable is associated with a 6.4 unit increase in social expenditure, holding other variables fixed. The conclusion that power resource theory has minor explanatory potential turns out to be true for both post-communist and non-post-communist members of the European Union.
Table 3.
Robustness checks (1)
Dependent variable -Social expenditure (% of GDP) (7) (8) (9)
left-wing parties (% of all parliamentary seats) 0.0016 (0.001) 0.0017 (0.0018) 0.0004 (0.002)
quality of government: ICRG index 4145*** (0.911) 5.079*** (1.404) 1.508 (1.46)
growth of GDP per capita -0.069*** (0.021) -0.076** (0.031) -0.083** (0.033)
openness to trade -0.015*** (0.003) -0.012*** (.004) -0.008** (0.004)
federalism 1.762*** (0.226) 1.798*** (0.463) 1.782*** (0.295)
dummy variable: postcommunist -5.797*** (1.742)
interaction term (QoGxpostcommunist) 6.377*** (2.42)
interaction term (percentage of left-wing partiesxpostcommunist) -0.003 (0.003)
year effects yes yes yes
intercept 13.376*** (0.651) 12.128*** (1.105) 14.692*** (1.24)
R-squared 0.7781 0.7939 0.7892
Number of observations 504 391 503
To capture possible variation in the effect of the alignment of political forces on social spending we introduce a set of variables for welfare regimes. The social-democratic regime includes Denmark, Finland, the Netherlands and Sweden. The con-servative-corporatist regime is realised in Austria, Belgium, France, Germany and Luxembourg. Anglo-Saxon EU member-states, namely, Ireland and the UK, make up the liberal cluster with developed private welfare systems and a relatively low share of social spending. The southern model (Bonoli 1997; Ferrera 1996) is represented by Cyprus, Greece, Italy, Malta, Portugal and Spain. We separate Bulgaria, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia out into the fifth cluster. The present literature does not suggest any clear classification of social systems in post-communist countries.
Table 4 summarises the results of estimating regression models with a set of dummy variables for the welfare regimes and a set of interaction terms between the percentage of left-wing parties and dummies for these welfare regimes. To provide a more parsimonious model, we do not control for year effects in models 10-12. The social-democratic cluster is used as a reference category in model 10. The regression coefficients for interaction terms are insignificant; in other words, there are no grounds to state that the effect of government composition on welfare generosity varies across welfare regimes. However, some clusters have few observations, which may be not enough to detect significant differences between groups. It makes sense to try other groupings which could be more theoretically grounded. Josifidis and his co-authors (2011) find that political orientation does not have a significant effect on social expenditure in social-democratic and conservative-corporatist welfare regimes. The authors explain this finding through structural factors. They underline that these two welfare regimes have developed welfare institutions, which allow institutional reforms to be an effective mechanism to reduce poverty. Other welfare regimes with underdeveloped redistributive insti-tutions face higher costs of implementing institutional reforms; therefore, changes in government composition in these regimes are manifested mainly through changes in social expenditure. To find out if this explanation is empirically confirmed for our sample, we ran the regression model 11 (see Table 4). For this model specification, sample units are divided into two groups: social-democratic and conservative-corporatist countries make up the first group, while the second cluster is represented by liberal, southern and hybrid (post-communist) regimes. We choose the first cluster as a reference category. Model 11 does not reveal any differences in the effect of government composition between the EU member-states with more and less developed redistributive institutions (the interaction term coefficient is insignificant). Next, we address path dependence theory, which explains why welfare regimes are different in terms of their resistance to the alignment of political forces. The conservative-corporatist welfare regime has survived despite numerous attempts by left-wing parties to impede the differentiated system of redistribution and high solidarity exclusively within occupational groups. The high persistence of this welfare regime is explained by the high costs of change (Korpi 2001). Fluctuations in social expenditure and reforms of the labour market
would be detrimental to the better-off occupational groups. In compliance with path dependence theory, one can expect that the effect of government composition on the generosity of the welfare state is less evident in conservative-corporatist regimes than in others. We generate a dummy variable for a conservative-corporatist welfare system and an interaction term between this dummy variable and the percentage of left-wing parties. As the estimates of model 12 indicate, the results of our study remain robust to changes in country grouping. Government composition does not affect social expenditure, and this finding is true for different welfare regimes.
Table 4.
Robustness checks (2): welfare regimes
Dependent variable -Social expenditure (% of GDP) (10) (11) (12)
left-wing parties (% of all parliamentary seats) -0.0003 (0.0017) -0.0002 (0.0018) -0.0003 (0.002)
corporatist welfare regime 0.683 (0.595) 3.304*** (0.432)
liberal welfare regime -3.334*** (0.674)
southern welfare regime -2.506*** (0.617)
hybrid (postcommunist) welfare regimes -3.335*** (0.655)
leftxcorporatist regime -0.0001 (0.003) -0.0002 (0.003)
leftxliberal regime -0.001 (0.004)
leftxsouthern regime 0.0018 (0.003)
leftxhybrid regime -0.004 (0.003)
dummy for corporatist and social-democratic regimes 3.486*** (0.478)
leftxdummy for corporatist and social-democratic regimes -0.003 (0.003)
controls yes yes yes
intercept 17.698*** (0.641) 14.599*** (0.529) 15.426*** (0.482)
R-squared 0.6913 0.6852 0.6627
Number of observations 522 522 522
Note. * - p < 0.1; ** - p < 0.05; *** - p < 0.01; panel-corrected standard errors are given in brackets.
Results of mixed-effects modelling confirm our main findings1. The quality of government and social expenditure are positively related, while there is no significant effect of political orientation on the generosity of the welfare state.
Conclusion
This paper contributes to welfare studies by testing alternative power resource and institutional theories on expanded data; in contrast to prior studies this research provides empirical evidence from EU member-states, including post-communist countries. Both regression models with panel-corrected standard errors and mixed-effects models confirm that government composition does not have a significant effect on the generosity of the welfare state. This result remains robust to changes in model specification. We also suggested subsampling to take into account possible variation in the effect of left-wing parties across different electoral systems and welfare regimes. However, our conclusion holds for different subsamples. The hypothesis about the positive relationship between the quality of government and social expenditure was partially confirmed. The empirical analysis shows that post-communist countries demonstrate a significant positive effect of the quality of government on the generosity of the welfare state, while this effect turns out to be insignificant for non-post-communist European Union members. The detected tendency implies that the ongoing process of democratisation accompanied by the increasing quality of government offsets the clientelist-oriented weak party system. To sum up, our study verifies that power resource theory is losing its explanatory potential, while the role of institutional performance in explaining the generosity of the welfare state is increasing.
Further research could be conducted into testing the result robustness to changes in the operationalisation of governance. Moreover, it would be interesting to digress from the left-party force. Power resource theory does not deny the role of other parties. As Korpi states, along with the left parties there are other significant actors such as conservative and Catholic parties (Korpi 1989). Thus, it would make sense to test if general changes in the alignment of political forces (emphasising not only left parties) have an effect on the generosity of social spending and social spending structure.
References
Arts W., Gelissen J. (2010) Models of Welfare State. Castles F., Liebfried S., Lewis J., Obinger H., Pierson C. (eds.) The Oxford Handbook of the Welfare State, Oxford: Oxford University Press: 570-584.
Bawn K., Rosenbluth F. (2006) Short versus Long Coalitions: Electoral Accountability and the Size of the Public Sector. American Journal of Political Science, 50 (2): 251-265. Beck N., Katz J. (1995) What to do (and not to do) with Time-Series Cross-Section Data. The American Political Science Review, 89 (3): 634-647.
1 Estimates of mixed-effects modelling are available on request.
Bonoli G. (1997) Classifying Welfare States: a Two-Dimension Approach. Journal of Social Policy, 26(3): 351-372.
Burgoon B. (2001) Globalization and Welfare Compensation: Disentangling the Ties that Bind. International Organization, 55: 509-551.
Castles F. (1994) Is Expenditure Enough? On the Nature of the Dependent Variable in Comparative Public Policy Analysis. Journal of Commonwealth and Comparative Politics, 33 (3): 349-363.
Ferrera M. (1996) The "Southern Model" of Welfare in Social Europe. Journal of European Social Policy, 6: 18-36.
Garrett G., Mitchell D. (2001) Globalization, Government Spending and Taxation in the OECD. European Journal of Political Research, 39: 145-177.
Immergut E. (2010) Political Institutions. Castles F., Liebfried S., Lewis J., Obinger H., Pierson C. (eds.) The Oxford Handbook of the Welfare State, Oxford: Oxford University Press: 228-240.
Ishiyama J. (2001) Party Organization and the Political Success of the Communist Successor Parties. Social Science Quarterly, 82 (4): 844-864.
Iversen T., Soskice D. (2009) Distribution and Redistribution: the Shadow of the Nineteenth Century. World Politics, 61 (3): 438-486.
Josifidis K., Hall J., Supic N., Ivancev O. (2011) European Welfare Regimes: Political Orientations versus Poverty. Panoeconomicus, 5: 651-674.
Kitschelt H. (2000) Linkages between Citizens and Politicians in Democratic Polities. Comparative Political Studies, 33(6/7): 845-869.
Korpi W. (2001) Contentious Institutions: an Augmented Rational-Action Analysis of the Origins and Path Dependency of Welfare State Institutions in Western Countries. Rationality and Society, 13(2): 235-283.
Korpi W. (1989) Power, Politics, and State Autonomy in the Development of Social Citizenship: Social Rights during Sickness in Eighteen OECD Countries since 1930. American Sociological Review, 54: 309-328.
Korpi W., Palme J. (2003) New Politics and Class Politics in the Context of Austerity and Globalization: Welfare State Regress in 18 countries, 1975-95. American PoliticalScien-ce Review, 97(3): 425-446.
Manow P. (2005) Germany: Cooperative Federalism and the Overgrazing of the Fiscal Commons. Obinger H., Leibfried S., Castles F. (eds.) Federalism and the Welfare State. New World and European Experiences, New York: Cambridge University Press: 222-263. Milesi-Ferretti G. M., Perotti R., Rostagno M. (2002) Electoral Systems and Public Spending. Quarterly Journal of Economics, 117(2): 609-657.
Obinger H. (2005) Austria: Strong Parties in a Weak Federal Polity. Obinger H., Leibfried S., Castles F. (eds.) Federalism and the Welfare State. New World and European Experiences, New York: Cambridge University Press: 181-221.
Orenstein M. (2008) Poverty, Inequality, and Democracy. Postcommunist Welfare States. Journal of Democracy, 19(4): 80-94.
Persson T., Tabellini G. (2004) Constitutional Rules and Fiscal Policy Outcomes. The American Economic Review, 94 (1): 25-45.
Pierson P. (1996) the New Politics of the Welfare State. World Politics, 48(1): 143-179. Scruggs L., Allan J. (2006) Welfare-state decommodification in 18 OECD countries: a replication and revision. Journal of European Social Policy, 16 (1): 55-72. Rothstein B., Samanni M., Teorell J. (2012) Explaining the Welfare State: Power Resources vs. The Quality of Government. European Political Science Review, 4 (1): 1-28. Rothstein Bo, Teorell J. (2008) What is Quality of Government? A Theory of Impartial Government Institutions. Governance: an International Journal of Policy, Administration, and Institutions, 21(2): 165-190.
Rudra N., Haggard S. (2005) Globalization, Democracy, and Effective Welfare Spending in the Developing World. Comparative Political Studies, 38 (9): 1015-1049.
