On 22 June 2017, the Appeals Court in Varna, Bulgaria found a former senior accountant at the local Social Protection Agency in Vulchi Dol guilty of expropriating 390,102 BGN ($222,559) allocated to social benefits for low-income citizens. This type of crime typically carries a prison sentence of 10–20 years. Nevertheless, the defendant, the granddaughter of a former local police director, was found not guilty by the Regional Court in Varna. It was only after the Appeals Court heard the case that she was sentenced to 6 years and 8 months in prison.
Two years earlier, two towns in Romania – Iaşi and Ţibana – suspended social benefits payments for three months after an investigation uncovered pervasive fraud. More than a quarter of the beneficiaries in Ţibana had received state support although they earned more than the legally determined threshold or no longer resided in the country. After a string of signals, a 2018 Court of Auditors Report estimated that Agency for Payments and Social Inspection branches in 17 cities across Romania had paid close to $5.2 million in family allowance and childcare benefits to recipients who were not legally entitled to social assistance in 2017.
These cases raise important questions about Bulgaria and Romania's ability to monitor their bureaucracy, ensure the proper use of their resources and enforce the rule of law. Such problems are pervasive across the developing world in general, and in post-communist societies in particular. Indeed, many of these countries suffer from endemic corruption, low government effectiveness and weak law enforcement (Holmes Reference Holmes2006; Kornai and Rose-Ackerman Reference Kornai and Rose-Ackerman2004). Prior studies have shown that these problems have serious implications for economic growth, economic development, foreign direct investment and public participation (Bubbico, Elkink, and Okolikj Reference Bubbico, Elkink and Okolikj2017; Rothstein Reference Rothstein2011; Rothstein, Samanni and Teorell Reference Rothstein, Samanni and Teorell2012; Valeriani and Peluso Reference Valeriani and Peluso2011). The anecdotes above indicate that bad government can also damage the state's ability to effectively alleviate inequality. Yet this effect has received less – and, arguably, insufficient – attention. This article fills this gap by examining the relationship between redistribution and the quality of government.
I argue that the quality of government powerfully shapes the supply of economic redistribution. Defined in terms of the corruptness of the state apparatus, the enforcement of the rule of law and the efficiency of the civil service (Rothstein, Samanni, and Teorell Reference Rothstein, Samanni and Teorell2012), the quality of government affects the availability of resources allocated to social protection and the effective targeting of social benefits recipients. Poor government can decrease the real amount of spending on social benefits for the poor if the state bureaucracy regularly expropriates funds from the system. The same effect occurs when the civil service is not competent enough to reach citizens who are eligible for social transfers. In the absence of a legal system that punishes offenses, malpractices persist. The state therefore fails to effectively alleviate income inequality.
The argument outlined above rests on several insights from the literature on comparative welfare states and state capacity, but proposes a new causal mechanism that previous studies have thus far overlooked. Research has found that clientelism, inefficiency, and corruption delegitimize national authorities (Gilley Reference Gilley2006; Seligson Reference Seligson2002), decrease mobilization in support of the welfare state (Rothstein, Samanni, and Teorell Reference Rothstein, Samanni and Teorell2012) and prompt citizens to turn to private markets for social services (Skocpol Reference Skocpol1992). Existing work has also established that, by satisfying basic needs, such practices serve as substitutes for the formal state infrastructure and perpetuate an underperforming social protection regime (Calvo and Murillo Reference Calvo and Murillo2004; Levitsky Reference Levitsky2003). In either case, the quality of government meaningfully affects the development of social policy, either by shaping individual preferences for social protection or by altering political actors’ calculations about the costs and benefits of social reforms. Neither of these arguments, however, looks exclusively at states’ capacity to properly implement redistribution. Ignoring this direct effect of the quality of government on the supply of redistribution neglects a meaningful alternative causal mechanism.
In order to evaluate this mechanism, I focus on Central and Eastern Europe. Partly due to the historical legacy of socialism, post-communist countries exhibit higher aggregate levels of support for state-sponsored redistribution (Pop-Eleches and Tucker Reference Pop-Eleches and Tucker2017). During the post-war years, these countries experienced the expansion of access to education, health care, and public services, the elimination of unemployment and the artificial suppression of income differentials (Orenstein Reference Orenstein2008). As a result, Eastern Europeans are less tolerant of economic inequality and generally expect the state to alleviate income differences (Haggard and Kaufman Reference Haggard and Kaufman2008; Pop-Eleches and Tucker Reference Pop-Eleches and Tucker2017). This expectation allows me to adjudicate between the demand-based argument formulated by existing scholarship and the supply-based theory that I propose here.
The quality of government hypothesis has not yet been systematically tested. Although the scope of this project is limited to the post-communist world, the argument is broadly applicable to states beset by rampant corruption, lower bureaucratic quality and ineffective enforcement of the rule of law. These problems affect a wide range of developing countries in Asia, Africa and Latin America. Indeed, a rich literature on social policy in Latin America has demonstrated that the implementation of social policy reforms is not always smooth and effective (Aramburú Reference Aramburú2010). Similarly, research on Africa has detailed the region's struggles with public sector corruption and low state capacity (Levy and Kpundeh Reference Levy and Kpundeh2004). The quality of government might therefore have equally important implications for redistribution in these regions.
The next section presents a brief overview of the existing welfare state literature. I proceed to highlight the pernicious impact of low-level corruption, administrative incompetence, and weak law enforcement on the allocation and effectiveness of state resources. I test this argument with cross-sectional time-series analyses of twenty-one Central and Eastern European countries between 1996 and 2013. This is the first systematic quantitative study to examine the determinants of economic redistribution in this region. The results support my argument. The fifth section addresses endogeneity concerns. I conclude by discussing possible implications for future research.
Determinants of Redistribution
A rich literature explores the determinants of redistribution, or the reduction in income inequality attributable to government taxes and transfers (Iversen and Soskice Reference Iversen and Soskice2006; Lupu and Pontusson Reference Lupu and Pontusson2011). Scholars have focused on the specific factors that motivate voters’ demand for redistribution as well as on the conditions that shape political elites’ ability to deliver social policies. These demand- and supply-oriented factors often complement and reinforce each other (Garay Reference Garay2017; Huber and Stephens Reference Huber and Stephens2012).
Existing work has emphasized the role of regime type and partisanship; democracies and leftist governments have been found to pursue a more egalitarian agenda (Boix Reference Boix2003; Huber and Stephens Reference Huber and Stephens2001, Reference Huber and Stephens2012; Korpi Reference Korpi1983; Meltzer and Richard Reference Meltzer and Richard1981; Muller Reference Muller1988; Stephens Reference Stephens1979). Political competition makes ruling elites responsive to a larger constituency and facilitates demands for higher redistribution (de Mesquita and Smith Reference De Mesquita and Smith2005; Haggard and Kaufman Reference Haggard and Kaufman2008). Simultaneously, labor unions and left-wing parties represent the interests of low-income constituencies. While these perspectives have found empirical support in Latin America and the advanced industrialized world (Esping-Andersen Reference Esping-Andersen1990; Huber and Stephens Reference Huber and Stephens2012), the literature on Eastern Europe has been less conclusive. Although Lipsmeyer (Reference Lipsmeyer2002) and Rueschemeyer, Cook and Orenstein (Reference Rueschemeyer, Cook and Orenstein1999) indicate that communist successor parties in the region have resisted the tightening of financing and eligibility criteria, Tavits and Letki (Reference Tavits and Letki2009) have shown that leftist cabinets have not raised general government, health and education spending.
An extensive strand of research has instead highlighted the emergence of new forms of risk related to the processes of industrialization, deindustrialization and globalization (Iversen and Cusack Reference Iversen and Cusack2000; Van Kersbergen and Vis Reference Van Kersbergen and Vis2013; Wilensky Reference Wilensky1974). These transformations alter the traditional employment structure and generate demands for greater social protection (Adsera and Boix Reference Adsera and Boix2002; Garrett Reference Garrett1998). Faced with the political incentive to compensate losers, national governments are expected to increase redistribution (Mosley Reference Mosley2003; Rodrik Reference Rodrik1996). Such increases, however, might be limited or impossible if a more generous welfare state undermines economic competitiveness (Boix Reference Boix2003; Muller Reference Muller1988; Rudra Reference Rudra2008). Indeed, research on Central and Eastern Europe has revealed that higher levels of foreign direct investment (FDI) and trade integration have led to cuts in social spending, while competition for capital has kept taxes low (Appel and Orenstein Reference Appel and Orenstein2018; Cerami and Vanhuysse Reference Cerami and Vanhuysse2009; Mahutga and Bandelj Reference Mahutga and Bandelj2008).
A different economic explanation of redistribution revolves around inequality. Building on Meltzer and Richard (Reference Meltzer and Richard1981), some have predicted that more unequal societies will alleviate income differentials more than relatively egalitarian states (Jaeger Reference Jaeger2013; Kenworthy and Pontusson Reference Kenworthy and Pontusson2005). This is because the difference between the average and the median voter's income is greater when inequality is high. Empirical work, however, has yielded mixed results, leading scholars to shift their focus to labor market segmentation and the specific structure of inequality (Alesina and Glaeser Reference Alesina and Glaeser2004; Dallinger Reference Dallinger2010).
Most traditional theoretical frameworks thus focus on the conditions that put redistribution on the political agenda, assuming that states can effortlessly implement inequality-alleviating policies. This assumption, however, is not always correct. Partly acknowledging this, Rothstein, Samanni and Teorell (Reference Rothstein, Samanni and Teorell2012) predict a positive relationship between social spending and the quality of government. Transparent, competent and impartial public institutions induce citizens to embrace a more active role for the state in redistributive affairs. Svallfors (Reference Svallfors2013) provides further evidence for this microfoundational argument, linking trust in public institutions to support for higher taxes and generous social policies. In contrast, when the masses distrust the state apparatus, they resist the expansion of the welfare state (Gilley Reference Gilley2006; Seligson Reference Seligson2002; Skocpol Reference Skocpol1992) even though they believe economic inequality is unacceptable (Loveless and Whitefield Reference Loveless and Whitefield2011). Political mobilization for redistribution is therefore connected to the efficiency of public institutions.
Although this argument incorporates the state apparatus, the hypothesized causal link neglects the direct effect that government has on redistribution. This omission raises questions about the applicability of this theory to environments where deeply entrenched policy legacies motivate people to expect the state to actively participate in social provision. Where public support for the welfare state is extensive, the relationship between redistribution and the quality of government is more likely to be driven by supply-side factors. Although poor government might weaken citizens’ support for redistribution, this effect might take time to crystallize. In contrast, the inefficiency and corruption of the institutional framework does not allow inequality-reducing social programs to reach their potential in the short run. This direct effect remains unexplored and under-theorized in the welfare state literature.
The perspective developed here contributes to a large scholarship on the impact of the quality of government on different political and socioeconomic outcomes. Better institutional quality and law enforcement are associated with lower levels of poverty, mortality and inequality (Chong and Calderon Reference Chong and Calderon2000; Holmberg et al. Reference Holmberg, Rothstein and Nasiritousi2011; Holmberg and Rothstein Reference Rothstein2011) and higher economic growth, per capita income, life expectancy, well-being and social trust (Gilley Reference Gilley2006; Helliwell and Huang Reference Helliwell and Huang2008; Holmberg et al. Reference Holmberg, Rothstein and Nasiritousi2011; Kaufmann Reference Kaufmann2004; Kaufmann, Kraay, and Mastruzzi Reference Kaufmann, Kraay and Mastruzzi2009; Rothstein and Teorell Reference Rothstein and Teorell2008). Furthermore, citizens living in better-governed states and regions are more politically engaged (Stockemer and Sundström Reference Stockemer, Sundström, Debiel and Gawrich2014) and supportive of democracy (Dahlberg, Linde and Holmberg Reference Dahlberg, Linde and Holmberg2015; Magalhães Reference Magalhães2014). I build on this work by exploring the direct effect of government quality on economic redistribution.
Redistribution and the Quality of Government
Drawing on the work of Rothstein, Samanni and Teorell (Reference Rothstein, Samanni and Teorell2012), the premise of this article is that the quality of government, broadly defined in terms of the corruptness of the public administration, the enforcement of the rule of law and the effectiveness of the state apparatus, shapes states’ capacity to successfully implement redistribution. Effectively decreasing inequality requires well-functioning institutions. Although existing work on the welfare state often takes them for granted, such institutions are frequently absent. In fact, weak administrative capacity and limited enforcement are the norm in many developing countries (Holland Reference Holland2016). Thus, even when they exist, institutions do not always work the way they are supposed to (Helmke and Levitsky Reference Helmke and Levitsky2004). The quality of government, which reflects how smoothly institutions function, therefore has the potential to meaningfully affect redistribution.
The absence of transparent institutions, competent civil servants and a rigorously enforced legal framework facilitates redistributive leakage. Redistribution can thus remain low even when governments adopt generous social policies and increase social spending. The presence of a redistributive framework does not guarantee that resources will go toward social programs. As the story of the Bulgarian accountant shows, low-level corruption and embezzlement might divert funds away from their designated use. Instead of helping low-income households, social spending might thus go to self-interested bureaucrats or people who have privileged access to the state. In these circumstances, it loses its potential to reduce income differentials.
A legal case from Romania illustrates this point. In 2014, the Bucharest police apprehended four people accused of using their access to public servants to obtain social benefits for citizens who did not qualify for economic assistance. Abuse of office thus redirected public funds to higher-income individuals, further exacerbating inequality.
Low government effectiveness can have a similar effect. If the civil service is comprised of officials who lack the necessary training and qualifications, it might fail to accurately assess the economic circumstances of potential beneficiaries and to properly apply social programs’ eligibility criteria. Citizens entitled to benefits and services can therefore remain outside the reach of schemes designed to help them, while individuals who do not qualify for social assistance might receive it. Consequently, social policies will not improve the socioeconomic condition of underprivileged communities and redistribution will remain low.
Several recent events from Eastern Europe exemplify these dynamics. In 2018, the leader of Romania's ruling party, Liviu Dragnea, was found guilty of intervening to keep two female party workers on the payroll of a child protection agency. Political appointments of this nature raise questions about the competence of the civil servants responsible for the implementation of social policy. In 2012, the Bratislava police established that the mother of a young murder victim had continued to collect child benefits years after her daughter's death despite her failure to file a legally required school attendance certificate. Slovakia's labor minister at the time, Jan Richter, explained that this oversight was likely due to the increasing shortage of qualified social workers (and the accompanying excessively high caseloads) in the country. In 2007, a BBC investigation found that some Polish workers in Britain were illegally claiming child benefits in both the United Kingdom and Poland. Officials from several Polish regions confirmed that up to half of the applications they process are fraudulent. Many of these cases remain undetected. This inability to properly apply eligibility criteria puts a financial strain on the Polish social protection system and might widen income differentials instead of reducing them.
Funds are also likely to be misallocated if the civil service is not effectively insulated from political pressures. Politically dependent bureaucrats can alter the eligibility criteria of social programs to include people who do not qualify for assistance in order to promote the political goals of the parties or public figures they work for. Local and national-level political agents might demand the inclusion of certain groups for electoral purposes. Such clientelistic forms of behavior can result in the diversion of funds away from those with real needs to beneficiaries who do not necessarily fall into this group. Even if it is legally designed to be progressive, social spending thus becomes ineffective at alleviating income inequality.
In the absence of adequate monitoring mechanisms and a properly enforced legal framework, such malpractices persist, creating a vicious cycle of abuse of power. If criminal acts are not punished, perpetrators have few incentives to properly implement redistributive legislation. Furthermore, citizens have few channels through which to demand accountability. The failure to enforce the rule of law will therefore exacerbate the problem of fund misallocation.
This conclusion is in line with the expectation that enforcement has distributional consequences but differs from recent work showing that noncompliance can alleviate inequality. According to Holland (Reference Holland2017), politicians refuse to implement laws that can hurt low-income citizens when they depend on poor voters to win office. In contrast, because I focus on redistribution rather than market income inequality, I expect non-compliance to have a regressive impact. I am primarily interested in the implementation of redistributive legislation once policies have been adopted. Furthermore, unlike Holland, who examines the strategic behavior of elected politicians, my theory centers on the behavior of the state bureaucracy.
The quality of government therefore affects the extent to which the state is capable of implementing redistribution. Poorly governed societies struggle to allocate resources to inequality-alleviating programs and reach those in need. This implies that countries with low-quality governments have lower levels of redistribution than better-governed states.
Data, Measurement and Models
To test this argument, this article employs cross-sectional time series analysis of twenty-one Central and Eastern European countries. Armenia, Azerbaijan, Belarus, Bulgaria, the Czech Republic, Estonia, Georgia, Hungary, Kazakhstan, Kyrgyzstan, Latvia, Lithuania, Moldova, Poland, Romania, Russia, Slovakia, Slovenia, Tajikistan, Turkmenistan, Ukraine and Uzbekistan were all members of the former communist bloc. As a result, they witnessed the establishment of comprehensive social protection systems during the post-war period. Combined with the communist economic model, these welfare states kept income inequality low (at least among non-party members). Because of this legacy, support for state-sponsored redistribution has historically been high in the region (Loveless and Whitefield Reference Loveless and Whitefield2011; Mason, Kluegel and Khakhulina Reference Mason, Kluegel and Khakhulina2000; Rose and Makkai Reference Rose and Makkai1995).Footnote 1 Furthermore, former socialist countries typically exhibit higher levels of corruption and lower compliance with the rule of law than their Western European neighbors (Earle Reference Earle2000; Grzymala-Busse Reference Grzymala-Busse2007; Vachudova Reference Vachudova2009). Focusing on this region thus allows me to assess the impact of the quality of government on economic redistribution in a context where the demobilizational mechanism proposed by Rothstein, Samanni and Teorell (Reference Rothstein, Samanni and Teorell2012) is less likely.
A subsequent set of analyses focuses more carefully on the ten countries that joined the European Union (EU) during the 2004 and the 2007 accession waves.Footnote 2 Limiting my analysis allows me to explore dynamics related to public support for redistribution, partisan ideology, revenues and social spending, data for which are only available for these ten states. It also addresses a potential source of endogeneity – that EU membership shaped the quality of government in Central and Eastern Europe. All ten countries in the smaller sample initiated the accession process in the 1990s, which largely determined the broader type of economic, political and judicial reforms that they pursued during the transition (Vachudova Reference Vachudova2005). This reform path could have affected the quality of government by requiring measures that penalized corruption, strengthened bureaucratic capacity and improved efficiency. In this sense, this subset of cases enables me to assess the relationship between redistribution and the quality of government in a more homogenous context.
The dependent variable, the level of relative redistribution, constitutes the difference between the pre- and the post-tax-and-transfer Gini coefficients expressed as a percent of the former. It reveals the generosity of a country's welfare state and the progressivity of its tax system. The variable is measured annually for the period between 1996 and 2012.Footnote 3 Data are available through the Standardized World Income Inequality Database (Solt Reference Solt2016), which standardizes data from various sources to facilitate cross-country comparability.Footnote 4
As illustrated by Figure 1, although redistribution changes little from year to year, it fluctuates over time. Furthermore, the extent to which different states engage in redistribution varies across the region. While some countries, such as the Czech Republic, Hungary and Slovakia, attain relatively high levels of redistribution, others, like Bulgaria, Romania and Tajikistan, tend to redistribute significantly less. The dependent variable thus presents interesting patterns across time and space.
Figure 1. Relative redistribution in Central and Eastern Europe (1990–2013)
The main explanatory variable is the Quality of Government, which is an index incorporating three of the World Bank's Worldwide Governance Indicators (WGI) (Kaufmann, Kraay and Mastruzzi Reference Kaufmann, Kraay and Mastruzzi2009). The WGIs pool data from different surveys to create indicators that seek to provide a comprehensive evaluation of each country's political system. The Control of Corruption measure reflects ‘the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as “capture” of the state by elites and private interests’ (Kaufmann, Kraay and Mastruzzi Reference Kaufmann, Kraay and Mastruzzi2009). The Government Effectiveness indicator combines responses about the quality of the bureaucracy, the competence of the public administration, the independence of the civil service from political pressures, and the quality of policy formulation and implementation. Lastly, the Rule of Law indicator captures perceptions of the incidence of crime, the effectiveness and predictability of the judiciary, and the enforceability of contracts. The composite index sums and averages these three measures. I also run models with each of these indicators to assess their separate impacts on redistribution. The variables have been measured annually since 1996.Footnote 5 Higher values indicate better government.Footnote 6 To test the robustness of my findings, I rerun my analysis with the International Country Risk Guide's quality of government index, which closely resembles the WGI index. The results are discussed in the appendix.
Preliminary evidence from correlation analysis offers initial support for my argument. Figure 2 plots relative redistribution against the quality of government index for all 21 countries in my sample in 2000. Better government is associated with higher levels of redistribution.
Figure 2. Redistribution and the quality of government in Central and Eastern Europe (2000)
To illustrate my proposed causal mechanisms, I run a set of models with data from the Varieties of Democracy Project (V-DEM, Coppedge et al. Reference Coppedge2017).Footnote 7 Public Sector Embezzlement captures ‘how often [……] public sector employees steal, embezzle, or misappropriate public funds or other state resources for personal or family use’ (V-DEM 2016, p. 172). It thus reflects the existence of public servants like the Bulgarian accountant mentioned in the introduction. Higher values indicate less pervasive malpractices. The Public Sector Corruption Index is calculated by adding a measure of ‘how often […] public sector employees grant favors in exchange for bribes, kickbacks, or other material inducements’ (V-DEM 2016, p. 67) to the embezzlement indicator. In this case, the directionality of the index goes from lower to higher corruption. Finally, a separate indicator captures the rigor and impartiality of the public administration. Higher values indicate higher impartiality.
A set of variables accounts for the effect of alternative explanations.Footnote 8 Economic growth, gross domestic product (GDP) per capita, inflation, debt and budget deficit control for unfavorable economic conditions that might limit governments’ capacity to expand social spending. The unemployment rate, the dependency ratio and the industry's value added to national GDP reflect the level of demand for benefits. FDI inflows, openness to trade, and capital account openness (Chinn and Ito Reference Chinn and Ito2008) capture the compensation and the race-to-the-bottom hypotheses.Footnote 9 Pre-tax-and-transfer inequality levels account for the possibility that more unequal societies redistribute more. Lastly, I include a measure of democracy (each country's Polity score (Marshall et al. Reference Marshall, Gurr and Jaggers2016)), ethnic fractionalization (Dahlberg et al. Reference Dahlberg2016), voter turnout, female labor force participation, checks and balances, and EU membership.
The reduced sample models allow me to assess the partisan ideology hypothesis by controlling for the percentage of parliamentary seats occupied by left-wing parties in government, weighted by the number of days in office in a given year.Footnote 10 These models also enable me to test the main alternative mechanism proposed by Rothstein, Samanni and Teorell (Reference Rothstein, Samanni and Teorell2012) by adding a measure of public support for redistribution, which is the proportion of European Social Survey (2016) respondents who agree that ‘[t]he government should take measures to reduce differences in income levels’.Footnote 11
Pooling data for cross-sectional time-series analysis presents several challenges that make the standard application of ordinary least squares (OLS) regression inappropriate. Crucially, they produce temporally autoregressive and cross-sectionally correlated error terms, which result in biased and inconsistent parameter estimates (Hicks Reference Hicks, Hicks and Janoski1994). To address this concern, I estimate Prais-Winsten models, which combine panel-corrected standard errors with ar(1) corrections. This strategy corrects for first-order autoregressiveness (Beck and Katz Reference Beck and Katz1995, Reference Beck and Katz2004). I refrain from including a lagged value of the dependent variable because this inappropriately suppresses the explanatory power of other covariates and eliminates cross-sectional variation (Aachen Reference Aachen2000). Prais-Winsten regressions, which accomplish the same effect without these limitations, are therefore more suitable for the purposes of this analysis. Lastly, my modeling technique assumes that disturbances are heteroskedastic across panels.Footnote 12 This empirical strategy is therefore quite conservative. As a robustness check, I run error correction models, which yield similar results (results reported in the appendix).
Results
Table 1 presents the results from five Prais-Winsten regressions examining the determinants of redistribution in twenty-one post-communist countries. Model 1 regresses redistribution on the WGI index. Model 2 shows the full specification. Models 3, 4 and 5 test for the individual impact of the three dimensions of the WGI index – control of corruption, government effectiveness and the rule of law. The R2 values associated with the models indicate that the quality of government indicators explain a meaningful proportion of the variation in redistribution.Footnote 13
Table 1. Redistribution and the quality of government: extended sample
***p < 0.001; **p < 0.01; *p < 0.05
The coefficients of all WGI indicators suggest that the quality of government has a positive and statistically significant impact on redistribution. On average, and holding the effect of all other variables constant, better-governed societies tend to redistribute income more than poorly governed states. This impact is substantively meaningful: a two-standard-deviation change in the composite WGI index results in an 18.7-percentage-point change in redistribution. For the control of corruption, rule of law and government effectiveness measures, this change is 9.9, 20.7 and 16.6 percentage points, respectively. Thus, countries with stronger bureaucratic apparatuses, lower corruption and more effective law enforcement attain higher levels of redistribution. The jackknife models in the appendix confirm that this result is not driven by any specific country.
Several other findings stand out. Contrary to studies on advanced capitalist democracies and Latin America (Huber and Stephens Reference Huber and Stephens2001, Reference Huber and Stephens2012), the results fit the Meltzer-Richards model: redistribution and inequality are positively correlated, suggesting that more inegalitarian societies alleviate inequality more than relatively more equal countries, or that rising market income inequality leads to more concentrated efforts to redistribute over time. This might be because public pressure for better social safety nets is indeed heeded by politicians in Central and Eastern Europe, or because political elites in the region are sensitive to growing inequality and attempt to address it in anticipation of a public backlash. This strategic choice might be due to the perceived costs of not undertaking measures to limit rising income differentials in an environment marked by a legacy of (at least apparent) egalitarianism. The relationship between inequality and redistribution might also be positive because the countries that implemented comprehensive economic reforms and experienced the largest increases in inequality invested more in the development of effective welfare states. Appendix Figures 10 and 11 show that the market Gini coefficient rose steadily in states that adopted market-based reforms the fastest. Meanwhile, Bulgaria, Romania and the Commonwealth of Independent States (CIS) countries experienced slower and smaller increases in income differentials. Consequently, they might have faced a weaker urgency to redistribute.
The positive coefficient of the turnout variable offers partial support for the demand-related interpretation. Higher levels of electoral participation are associated with higher relative redistribution, ceteris paribus. This is consistent with a situation in which governing elites fear being penalized for not taking action against inequality. Similarly, a larger industrial sector translates into higher levels of redistribution. This might be because industrial workers are better organized and therefore more capable of mobilizing in demand for redistribution. In contrast, higher ethnic fragmentation and female labor force participation inhibit redistribution. The other economic controls fail to reach statistical significance.
Table 2 re-runs Models 1–5 for the smaller sample of countries, adding controls for public support for redistribution, partisanship and electoral disproportionality. The results remain substantively unchanged. All WGI indicators are positively signed and statistically significant. As in the larger sample, their impact is not negligible. A 2-standard-deviation increase in the composite index, control for corruption, government effectiveness and rule of law translates into a 34.9, 21.8, 26.5 and 40.6 percentage-point rise in redistribution, respectively. Indeed, the quality of government remains an important predictor of redistribution even in these more conservative models.
Table 2. Redistribution and the quality of government: reduced sample
***p < 0.001; **p < 0.01; *p < 0.05
Like before, higher market inequality and a larger industrial sector are related to higher levels of redistribution. Interestingly, ethnic fragmentation and voter turnout lose statistical significance. By contrast, the coefficients for government debt and the dependency ratio are statistically significant. Surprisingly, higher government debt is positively correlated with inequality-reducing efforts, which suggests that spending more on redistribution might put a strain on government budgets. While the negative coefficient carried by the dependency ratio seems counterintuitive, it might be explained by the introduction of private pension schemes and the cuts in education and pension transfers during the transition period. Lastly, EU membership is negatively related to redistribution, perhaps due to the fiscal constraints imposed by the EU.
Surprisingly, public support for redistribution yields inconclusive results. Its positive coefficient is consistent with Rothstein et al.'s argument, indicating that a public expecting the state to alleviate inequality is conducive to higher levels of redistribution. Nevertheless, the variable's lack of statistical significance in two out of four models suggests that the quality of government does not affect redistribution solely by decreasing public support for the welfare state (if this were the case, the WGI index would not be significant). This points to an alternative mechanism directly linking redistribution and the quality of government.
Causal Mechanism Tests
This mechanism might be better illustrated by looking at the interaction between the quality of government and social spending or state revenues. Scholarship on the welfare state has established that these variables typically play an important role in redistribution as they determine the amount a state can (and does) spend on social transfers and benefits (Huber and Stephens Reference Huber and Stephens2001). The argument developed here, however, posits that social spending and state revenues are not effective at reducing market income inequality in poorly governed countries. The quality of government thus moderates the redistributive effect of social spending.
To test this proposition, Table 3 interacts the WGI Index with social spending and state revenues as a percent of national GDP. The models include all of the controls featured in Table 2, but I only present the statistically significant coefficients. The full specification is in the appendix. The interaction terms are positively signed and statistically significant, lending support to the idea that the quality of government mediates the effect of social spending or state revenues on redistribution.
Table 3. Social spending and revenues models (reduced sample)
***p < 0.001; **p < 0.01; *p < 0.05
Figures 3 and 4 show the marginal effect of social protection spending and state revenues over the range of the WGI index. The plots reveal that the impact of spending/revenues on redistribution consistently increases as the quality of government improves. In fact, this effect switches from negative to positive in the case of revenues. State revenues have a regressive effect on income inequality in poorly governed countries. However, where corruption is limited, the bureaucracy is competent and the rule of law is upheld, social protection spending and state revenues have a positive impact on redistribution. A dollar raised and spent on social assistance schemes is much more effective at reducing inequality in such environments.
Figure 3. Average marginal effects of social spending on redistribution
Figure 4. Average marginal effects of tax revenues on redistribution
Data from V-DEM yield additional support for this argument. The regressions below test the direct impact of public sector embezzlement, public sector corruption, and the rigor and impartiality of the public administration on economic redistribution. The first three models are run on the extended sample, and the second three on the reduced sample. To save space, I only show the relevant coefficients although the regressions include all of the controls from Tables 1 and 2. The full specification, as well as models that interact these variables with social spending and state revenues and introduce fixed effects, are included in the appendix.
Consistent with my argument, public sector corruption is found to be negative and statistically significant (Table 4). This suggests that bureaucratic malpractices suppress the state's ability to reduce inequality. If public servants regularly steal from the state and/or abuse their office, they act against the goal of social policies. Thus, building on previous models, which indicate that an incompetent bureaucracy hinders redistribution, these regressions show that resource embezzlement and public sector corruption also obstruct the capacity of the state to reduce economic inequality.Footnote 14
Table 4. Public sector corruption and embezzlement models
***p < 0.001; **p < 0.01; *p < 0.05
Endogeneity
A cause for concern might be the presence of endogeneity. The primary endogeneity problem that could undermine the findings is the possibility that some background factor directly and simultaneously shapes both redistribution and the quality of government. This section discusses four potential confounders. I examine the theoretical underpinnings of these background conditions and explain why they do not challenge my argument. Table 5 presents the results from models that control for them. To save space, I only present the coefficients for the WGI index and the additional controls. The full specification is in the appendix.
Table 5. Endogeneity checks
***p < 0.001; **p < 0.01; *p < 0.05
Economic structure
The structure of the economy might simultaneously affect redistribution and the quality of government. Rural countries are likely to have insufficiently developed bureaucratic and law enforcement systems. Simultaneously, research on the welfare state has shown that industrialization is associated with greater redistribution (Wilensky Reference Wilensky1974). To account for this possibility, I add a control for the proportion of the population that lives in rural areas. The variable is insignificant in the extended sample models but statistically significant in the reduced sample. Contrary to expectations, its effect is positive, suggesting that more rural countries tend to redistribute more than urban economies. This might be because the EU emphasizes regional development.Footnote 15
Political competition
Secondly, the level of political competition might also be a confounder. Noncompetitive party systems can reduce representation and accountability, promote state capture and facilitate exploitation of the state apparatus (Grzymala-Busse Reference Grzymala-Busse2007; Hellman Reference Hellman1998; O'Dwyer Reference O'Dwyer2006; Vachudova Reference Vachudova2005). They can also suppress redistribution, as political elites with undisputed power refuse to increase social spending or raise taxes. To control for political competition, I add a variable capturing the effective number of parties in the national legislature.Footnote 16 Its coefficient fails to reach statistical significance, implying that a political system characterized by the presence of more parties does not induce higher levels of redistribution. As an additional robustness test, I use V-Dem's index of political competition, which captures the institutionalization of competition and government restrictions on political competition. The variable is positively signed and statistically significant in the reduced sample, lending support to the idea that political elites might fear the electoral consequences of rising inequality in countries with a more egalitarian historical legacy.
Type of communist regime
Thirdly, Kitschelt (Reference Kitschelt1999) has argued that communism took different shapes in Central and Eastern Europe. Different countries experienced different levels of repression, clientelism, bureaucratic capacity and reform. These legacies might have contributed to subsequent levels of redistribution and government quality. To test for this possibility, I add a variable that assumes a value of 1 for former members of the Soviet Union, 2 for bureaucratic authoritarian countries (the Czech Republic and Slovakia), 3 for national accommodative regimes (Hungary, Poland, Slovenia and the Baltic republics), and 4 for patrimonial states (Bulgaria and Romania). Communist legacy is significant and negatively signed in the reduced-sample models: a transition from a bureaucratic authoritarian to a national accommodative or patrimonial order is associated with a decrease in the level of redistribution. Nonetheless, the WGI indices remain statistically significant.
EU funds
Lastly, the amount of EU funds that Eastern European countries receive might alleviate budget constraints, facilitating higher redistribution and investment in bureaucratic institutions. To evaluate this possibility, I include a control for EU funds as a percent of GDP. The variable fails to reach statistical significance.Footnote 17 This suggests that the economic assistance the region receives from the EU does not allow it to engage in higher redistribution.
The analysis thus demonstrates that the quality of government is an important determinant of redistribution. I do not argue that this variable is entirely exogenous to historical legacies, economic development or political competition. Existing research has shown that state capacity may be influenced by these factors. Rather, my argument is that regardless of its own determinants, the quality of government has a strong and independent effect on redistribution.
Conclusion
This article argues that there is a meaningful relationship between a country's level of redistribution and its quality of government. This association remains understudied in the welfare state literature. Only recently have scholars begun to explore the impact of corruption, bureaucratic incompetence, and the ineffective enforcement of the rule of law on the capacity of a state to reduce income inequality through taxes and transfers (Rothstein, Samanni and Teorell Reference Rothstein, Samanni and Teorell2012). Most existing work has so far been limited to the developed world, where poor government supposedly delegitimizes the state.
This analysis strives to expand the scope of these studies beyond advanced democracies, where the quality of government is high enough to prevent substantial redistributive leakage. Less developed countries often suffer from higher levels of corruption and lower government effectiveness. These problems directly affect the supply of redistribution by diverting resources from their designated uses. Public funds thus end up in the pockets of public officials or recipients who do not qualify for them. Legitimate beneficiaries, by contrast, either do not receive the benefits for which they are eligible or get a much smaller fraction of overall social spending than they are entitled to by law. An incompetent and politically dependent bureaucracy further exacerbates this problem as civil servants prove incapable of reaching those most in need. In a context where the primacy of the law is not respected, malpractices persist.
Although this argument is applicable to all countries suffering from corruption, government ineffectiveness and the uneven application of the rule of law, I test it in the context of a region with a historical trajectory that allows me to adjudicate between my proposition and existing theories linking redistribution to the quality of government. The experience of communism in Central and Eastern Europe conditioned people’ perceptions of the role of the state in social provision. Post-communist citizens are thus generally more supportive of state-sponsored redistribution (Pop-Eleches and Tucker Reference Pop-Eleches and Tucker2017). As a result, the region provides an ideal setting in which to test the supply-based argument developed here.
To be clear, I do not argue that poor government cannot undermine trust in state institutions and decrease popular support for state-sponsored redistribution, as Rothstein, Samanni and Teorell (Reference Rothstein, Samanni and Teorell2012) posit. In fact, the demand- and supply-based effects are tightly linked: they can occur sequentially and reinforce each other. Noticing the corrupt behavior of civil servants and the inability of social provision schemes to reduce inequality, people might grow disillusioned with the state and may ultimately prefer a less generous system of social protection. Individual attitudes, however, are likely to respond to observations about the redistributive potential of the welfare state. Consequently, the mechanism I propose should precede the withdrawal of support for the welfare state that Rothstein, Samanni and Teorell (Reference Rothstein, Samanni and Teorell2012) predict.
This withdrawal might occur differently in different environments. Citizens in Central and Eastern Europe might take longer to change their views. If their tolerance of inequality is lower, individuals might be reluctant to cease to support state-sponsored redistribution and might need more time to adjust their attitudes. In other regions with different historical legacies, attitudes toward redistribution might be more easily and strongly swayed by information about the quality of government. Thus, citizens might quickly withdraw support for the welfare state after witnessing inefficiency, corruption or poor enforcement. The public opinion mechanism can thus manifest with different strength and speed in different contexts. In either case, however, the public opinion mechanism is likely to be activated in response to the state's inability to properly implement redistribution. In other words, redistributive leakage, which is the direct mechanism I focus on in this article, is likely to prompt people to react negatively to the welfare state.
Future research should extend the analysis to other regions. The quality of government might have equally important implications for redistribution and the design of the welfare state there. Indeed, recent work by Holland (Reference Holland2016, Reference Holland2017) has shown that forbearance, or the selective enforcement of the rule of law, plays a crucial role in shaping distributive outcomes in Latin America. While the focus of this article is on the ability of the existing social protection framework to alleviate inequality rather than on the ability of the state to affect market income through the selective application of its laws, my argument could also apply to that context given the generally higher levels of corruption, government inefficiency and failure to enact the rule of law in Latin America. In fact, as Aramburú (Reference Aramburú2010) points out in relation to the Juntos program in Peru, targeting potential beneficiaries in this part of the world also seems difficult. Díaz et al. (Reference Díaz2009) confirm this conclusion, highlighting problems related to identifying recipients that have prevented the state from reaching citizens living in poverty while granting benefits to relatively better-off households. Exploring whether forbearance perpetuates the persistence of ineffective social protection institutions is another research question worth pursuing.
Additional research is also required to further elucidate the ways in which individual preferences and structural characteristics interact to shape the redistributive capacity of the welfare state. Consistent with Rothstein, Samanni and Teorell (Reference Rothstein, Samanni and Teorell2012), we would expect to see corruption and inefficiency decrease support for redistribution. Is this relationship similar in different contexts? Or does the existing structure of the welfare state mediate the impact that the quality of government has on citizens’ support for state-sponsored redistribution? The answer to this question can shed light on the factors shaping the modern welfare state around the world.
Supplementary material
Data replication sets are available in Harvard Dataverse at: https://doi.org/10.7910/DVN/APVNPF and online appendices at:https://doi.org/10.1017/S0007123419000085.
Acknowledgements
I thank Cameron Ballard-Rosa, Tom Carsey, Liesbet Hooghe, Evelyne Huber, George Jiglau-Labunet, Gabriele Magni, Gary Marks, Lucy Martin, Elizabeth Menninga, Ana Petrova, Alina Postolache, Jenny Pribble, Graeme Robertson, John Stephens, Aleksandra Sznajder-Lee, Milada Vachudova, participants at the Comparative Working Group at the University of North Carolina at Chapel Hill, and anonymous reviewers for helpful comments. Replication data and code are available in Harvard Dataverse at https://doi.org/10.7910/DVN/APVNPF.
