Drivers of Government Activity in European Countries: Do Partisan Politics Still Divide East and West?

This article takes a novel look at the relationship between government activity, partisan preferences and varieties of capitalism. Evidence from panel regressions for 25 EU countries from 1990 to 2014 suggests that there are major divides among European countries in terms of the drivers of government activity, that is, government spending and government regulation. The European divide appears to be even more pronounced between liberal and coordinated economic systems than between the classical geographical divide of east and west, which is typically used in most contributions. While both divides apply to the determinants of government activity in general, a reversal of the classical partisan effect for the east is to be found only in specific cases and, is most likely in government spending in liberal eastern countries.


Introduction
The drivers of government activity have long been discussed in the academic literature. In particular, partisan preferences have attracted attention as one crucial determinant of government activity (see Imbeau et al., 2001 andPotrafke, 2017 for a meta-analysis of this literature). Classically, left-leaning governments are assumed to prefer an increasing role of the state in the economy, while right-leaning governments seek to minimalize such state intervention.
However, Tavits and Letki (2009) claimed that the traditional partisan effect on governmental activity has been reversed in post-communist central and eastern European countries (CEECs). According to their results, eastern left-wing parties reduced government consumption in an attempt to demonstrate their firm commitment to market economic reforms and to dissociate themselves from the communist past. However, these reversed partisan preferences in CEECs have been found for the pre-accession phase only, without including western EU countries as a benchmark in the sample and without considering regulations as a substitute for government spending. Therefore, whether or not an east/west divide in the effect of partisan preferences on government activity still exists remains an empirical question.
In addition, Tavits and Letki neglected an additional divide along economic systems. The literature on varieties of capitalism (Hall and Soskice, 2001) shows that Europe has varieties of both prototypes: market-based, Anglo-Saxon liberal market economies (LMEs) and consensus-based, continental European coordinated market economies (CMEs). Integrating welfare policies and, therefore, government activity (see Amable, 2003;Schröder, 2013) confirms this major divide, which also includes LME and CME varieties in eastern Europe (see Ahlborn et al., 2016). So far, only a few studies have hinted at the determining role of partisan politics in the formation of economic systems (Iversen and Stephens, 2008) and the different role played by partisan politics in existing economic systems (Pierre, 2015;Shelton, 2012). Anecdotal evidence suggests that left-wing parties in LMEs are less in favour of government activity than those in CMEs, as witnessed by reactions to the financial crisis.
The purpose of this article is to analyse the possible heterogeneity that may arise among EU economies due to the eastern enlargement or due to differences in capitalist systems. The article proceeds as follows. In Section I, we discuss the relevance of the traditional left-right preferences, given parties' reactions to the negative effects of globalization by using a model of an open versus closed society cleavage and develop our argument for examining both divides in the context of an analysis of partisan preferences. In Section II we explain our econometric model to consider east/west and LME/CME subsamples of European countries. Section III gives the empirical results based on panel data for 25 EU countries from 1990 to 2014.

Partisan Preferences and Government Activity: Recent Insights
Classic partisan theories offer different explanations for the formation of the right-left division. Partisan preferences are the shared preferences of individual politicians and legislators seeking to overcome collective action problems in governance by building coalitions in the form of parties and formulating party and government ideologies as a common denominator (Aldrich, 1995). Alternatively, the right-left division is said to mirror the preferences of the electorate and social cleavages as party members seek votes and office (Bartolini and Mair, 1990). 1 Hence, the right-left division can be formed either in a top-down or bottom-up way. Economic interests constitute the primary dimension of party competition in modern democracies (Benoit and Laver, 2006). Left leaning governments are assumed to prefer an increasing role of the state in the economy, while right-leaning governments seek to minimalize state intervention (Iversen and Stephens, 2008).
Yet, empirical evidence for this supposed 'fact' remains ambiguous, as demonstrated by survey articles by Imbeau et al. (2001) and Potrafke (2017). 2 Herwartz and Theilen (2014) find that partisan influence on social expenditure has decreased since the 1990s, which is in line with the findings of other scholars (Iversen, 2001;Kittel and Obinger, 2003;Potrafke, 2009), but they also find that, over time, the electoral cycle has become a more important over time driver of social expenditure. In another study, Herwartz and Theilen (2017) find that, in times of unfavourable economic conditions, left-wing parties increase social expenditure more than their counterparts. Hence, whether government ideology still constitutes a major driver of government expenditure remains unclear.
The empirical evidence for the effects of government ideology on regulationthe second most important aspect of government activityis somewhat less ambiguous. Jäger (2017) for example, finds that, while the influence of partisan preferences on social spending may have decreased, the preferences of left and right-wing parties are still shaping differences in economic policy-making. Belloc and Nicita (2012), Belke et al. (2007) and Obinger et al. (2014) draw similar conclusions, as their analyses suggest that right-wing parties more actively engage in privatization. Furthermore, Potrafke (2010) as well as Smith and Urpelainen (2016) find that right-wing parties also favour product market deregulation compared with leftist parties. However, Belloc and Nicita (2011) find that ideological cleavages play no important role in the regulation of network industries, as governments under control of both left and right-wing parties pursued deregulation of such industries (compared with 'centre' governments). Overall, however, empirical evidence suggests that government ideology shapes regulation: right-wing parties more actively pursue deregulation and privatization, as opposed to left-wing parties that tend to limit economic freedom.
All in all, evidence one the role of government ideology is rather mixed. This is supported by the fact that party competition has changed due to globalization. Recent research demonstrates that party politics have become less driven by ideological programmatic differences (De Simone and Mourao, 2016;Mair et al., 2014). A new political cleavage has opened, running between the support for open versus closed societies (Enyedi, 2008). As a reaction to the negative effects of globalization (for example, increased inequality) parties have sought to compensate and protect their electoral constituencies but their reactions have not been uniform (Adams et al., 2009;Haupt, 2009). Leftist parties have either followed traditional lines and became closed economic leftists (Burgoon, 2012;Tavits and Potter, 2015) or they have instead adopted a liberal reform agenda, for example, the social democrats implementing labour market deregulation during the Agenda 2010 reforms in Germany (Weishaupt, 2010) or new Labour in the UK maintaining a liberal reform agenda (Weishaupt, 2018). The rightist parties rally voters on the non-economic values dimension of their programmatic ideologies (Tavits and Letki, 2014;Ward et al., 2015), which currently figures prominently in the closed societies approach of new conservativism in eastern Europe (Bluhm and Varga, 2018).
Hence, the classical assumption that left governments prefer a higher level of government activity seems to be supported for government regulation rather than for government spending. At the same time, established relationships are challenged by new party cleavages, implying that ideology is decreasing in importance, as expressed by indicators available in the literature.

First Source of Heterogeneity: Differences between East and West
The validity of the classical hypothesis about the effects of partisan preferences on government activity has been especially challenged by the path taken by the new member states of the EU in eastern Europe (CEECs). Most prominently, Tavits and Letki (2009) argue that left-wing parties in central and eastern Europe are willing to reduce government consumption in order to demonstrate their firm commitment to market economic reforms and dissociate themselves from the communist past. They suggest that leftist parties in the east are able to do so as they can draw on a committed, non-fragmented and loyal base of supporters to whom they can credibly sell policies that apparently reduce welfare as a means of last resort. Their argument rests on the assumption that left-leaning, post-communist governments need to establish themselves as credible democratic parties that are committed to market economic reform.
This transition-centred explanation concurs with the research on general effects of Europeanization on parties' preferences and policy strategies (see Poguntke et al., 2007, Ladrech, 2012. Some studies show that the tempered partisan politics in CEECs before accession to the EU were shaped by the EU accession process itself (Vachudova, 2008;Vachudova and Hooghe, 2009). Among other factors, European integration has limited domestic governments' choices on economic policies (including regulation, tax and welfare) and has constrained their ability to credibly differentiate themselves along economic issues (Steiner and Martin, 2012;Ward et al., 2015).
However, the historical experiment of large-scale privatization during the process of transition shows a different picture. As Bjørnskov and Potrafke (2011) note, right-wing parties in the east tend to show a bias towards the privatization of small entities, but the same pronounced differences between left and right-wing positions found in OECD countries could also be observed in the east: privatization is stronger in the case of right-wing government and this is even more pronounced in the early years of transition. As Bjørnskov and Potrafke also point out, the determination of party position during transition is blurred by a natural flux in party organizations and positions as well as an additional cleavage opened by national positions.
All in all, the evidence discussed so far is mainly based on pre-accession data strongly influenced by both transition and Europeanization. In addition, Tavits and Letki's article, which most prominently argues for a reversed partisan effect, concentrates on government spending and, hence, neglects traditional positions with respect to government regulation.
Recent research suggests that new party positions have emerged in the post-accession era. With respect to leftist parties, there is a tendency to combine a culturally liberal agenda with more traditional left-leaning positions on economic and social issues (Vachudova, 2008; but see Haughton, 2009). With respect to rightist parties, new conservatism promoted economic nationalism and closed society views, especially in Poland and Hungary. The Law and Justice (PiS) government in Poland increased both its level of government spending and regulation. The 500 plus programme provided a clearly identifiable financial boost to many low-income households, while the strategy for responsible development promoted a recentralization of state power and the partial reversal of privatization (Krzysztof, 2018). The Fidesz government in Hungary instead followed traditional lines with a lean government approach and a massive retrenchment of welfare during financial crisis years but, similar to PiS in Poland, it strongly increased government involvement in nearly all policy areas including the pension system and public work programmes (Szikra, 2014).
Hence, while the return to traditional values in the case of leftist parties should have reduced a potential pre-accession reversal of partisan preferences, the new conservatism development should support a reversed partisan effect, with right-wing parties spending more, or regulating more, than traditional partisan preferences would suggest (Jarosz and Kozak, 2016).

Second Source of Heterogeneity: Differences between Economic Systems in LME and CME Countries
Economic systems can be expected to determine or moderate the playing field for partisan preferences. The traditional varieties of capitalism approach (Hall and Soskice, 2001) assumes there are two prototypes of capitalist systems in advanced market economies that differ in terms of institutions and regulations for the production system: LMEs achieve 'radical innovation' by formal interaction and coordination on markets; CMEs achieve 'incremental' innovation by (partly informal) non-market interaction and coordination.
The traditional varieties of capitalism approach has recently been extended by considering the literature on the worlds of welfare states, inspired by the seminal work of Esping-Andersen (1990). Schröder (2013) and Amable (2003), among others, have shown that CME countries differ according to government spending butgiven the underlying preferencesin consistent economic systems spending behaviour ought to mirror regulation behaviour. Hence, differences in the level of government activity, both in terms of spending and regulation, are relatively clear: CMEs are expected to regulate and spend more than LMEs. In addition, Pierre (2015) reveals that the relationship between regulation and spending differs between economic systems because CMEs tend to compensate potential losers for deregulation by increasing spending.
Several studies have classified countries along this distinction between LME and CME, a divide that also runs across European countries. The (European) LME type, characterized by small states and minimal regulation of the economy, is exemplified by the UK. The CME type, characterized by stronger welfare states and more regulation, is typically associated with Germany (Hall and Soskice, 2001). Around these two prototypes, LME (mainly Anglo-Saxon countries) and CME country groups (Scandinavian, continental and Mediterranean European countries) have been identified among developed economies, mainly via cluster analysis (see Ahlborn et al., 2016;Danforth, 2014;Schneider and Paunescu, 2012). In their empirical study, Ahlborn et al. (2016) confirm the hypothesis by Greskovits (2007, 2012) about varieties of east European systems reflecting western differences: these CEECs split up into two country groups that resemble the LME/CME prototypes, respectively Table (S1). 3 Hence, considering differences in economic systems for a sample of European countries is an important control variable when testing for partisan preferences. Different economic systems imply different structural constraints and opportunities for government parties to increase or decrease government activity (see Cusack et al., 2007;Iversen and Soskice, 2006;Korpi, 2006). In contrast with LMEs, CMEs are typically associated with proportional representative systems that bias government partisanship in a centre-left direction. Voters in CMEs favour government activity and higher levels of regulation and redistribution. In addition, unions and employers' organizations are traditionally stronger and have a greater impact on political outcomes. Hence, in LME countries, governments can be assumed to be less constrained by coalition governments, which are more likely to be the result of proportional voting, and less likely to be constrained by influential stakeholders beyond the government (for example, trade unions or business associations). Hence, governments are less bound to a consensual decision-making process.
Certainly, in majoritarian systems, governments also need to cater to median voters who prefer less redistribution (Iversen and Soskice, 2009). Nevertheless, the marginal effect of a more than average left-leaning government should be higher than in CMEs, where consensus and compromise between different groups within and outside the government are in-built constraints to the realization of more extreme partisan programmes. This conclusion, however, depends on the political constraints built into the political system. These constraints are generally higher in CME countries but differ strongly within LME and CME groups. For example, in terms of bicameralism and federalism, the USA and Germany face similar constraints, which are significantly higher than in the UK and Sweden (see Lijphart, 1999, Appendix A). Hence, leftist politics should be more effective in LME countries than in CME countries, although this is conditional on political constraints and the need to form coalition governments.
Apart from differences in effectiveness, Shelton (2012, p. 214) argues that [g]iven the importance of institutions in framing policy issues and suggesting solutions with particular distributional and ideological consequences, one would expect economies with such broadly different institutional bases to be characterized by different political cleavages and, as a result, by different partisan political business cycles.
According to Shelton's results, the left in CME countries is associated with lower output growth, lower interest rates, higher unemployment and higher inflation, while the left in LME countries is associated with higher output growth, higher short and long-term interest rates but not higher unemployment and inflation. This is a clear indication of different perceptions of trade-offs involved in government activity between CME and LME counties. While, to our knowledge, there has been no comprehensive treatment of the positions of left-wing parties' positions in different economic systems, some support for our hypothesis can be based on anecdotal evidence from western and eastern prototype systems.
UK (LME) versus Germany (CME). Nachtweih (2009) shows that the transformation of social democratic parties in the UK (new Labour) and Germany (Social Democratic party) is embedded in the respective liberal or coordinated background. New Labour focuses on poverty protection in a liberal welfare state whereas the Social Democratic party focuses more traditionally on the protection of living standards in an extended welfare system. In line with this distinction, Bermeo and Pontusson note that the reaction to the financial crisis was influenced by economic systems. In the UK, the Labour government, as in other LME countries, avoided a large fiscal stimulus. In Germany, the Social Democrats in the coalition government agreed to a large fiscal stimulus, as was the case for other export-led CME countries (Bermeo and Pontusson, 2012, p. 18).
Estonia (LME) versus Slovenia (CME). Leftist parties in Slovenia have their organizational roots in the old socialist parties, following traditional leftist policies. For example, the social democrats sought to protect as many of the gains of socialism as possible and declared the necessity of state interventions to stop radical economic liberalism (Cabada, 2005). In Estonia, successor parties have been unable to survive after independence and left-wing parties are clearly pro-market (Adam et al., 2009;Vogt, 2003). As a consequence, it is difficult to detect any specific leftist policies in Estonia. For example, the country had to increase its level of regulation in order to converge towards the EU level in the pre-accession phase and, as other LME countries, it followed a strict austerity programme during the financial crisis period. 4,5 However, whether or not these differences in left-wing parties' positions towards government activity in different economic systems translate into different patterns of government activity remains an empirical question.

II. Data and Methodology
In order to answer these questions on the determinants of government activity, we employed panel data for 25 European countries including the EU-15 and the ten CEECs that entered the EU in 2004 and 2007 respectively (see Table S1). Our data set is restricted by the start of the transition process in 1990 and the limited availability of some of the data series beyond 2014. We start this section with a brief description of our empirical approach before we turn to the data description and our methodological approach.

Empirical Model
Our regression model is based on Shelton (2007) and assumes that the determinants of the level of government activity differ between categories of government activity (govactivity) and can be broadly differentiated into ideology, other political determinants (politics) and economic determinants (economic). We further assumed that the effect of ideology may be different for eastern European countries (east) and that, in addition, this difference may be modified by EU membership, that is, after accession for the CEECs (eu). Because we were interested in explaining different levels of government activity rather than its dynamics, we considered changes in the endogenous variable as a robustness check only. Our basic econometric model reads govactivity ¼ a þ bÃideology þ cÃpolitics þ dÃeconomics þ gÃeu þ hÃeastþu: (1) Separate regressions were run for govactivity, measured in terms of spending and regulation, and economics and politics constitute vectors of economic and political control variables.
We analysed specific effects for CEECs before and after accession, assuming that Hence, the final regression model reads as follows: Due to the limitations of the data set, it is not possible to acknowledge overlapping cluster effects using additional dummies in one regression. Instead, we estimated equation 3 for five subsamples (ALL, EAST, WEST, LME, CME). The sample splits follow the allocation of EU countries according to Table S1.

Discussion of Political and Economic Variables Govactivity
The conventional measure for government spending is government consumption (GOVCON), taken from the World Development Indicators. 6 This variable includes all government consumption, investment and transfer payments and is measured in percentage of GDP. It is most widely used in the literature because of its availability on an annual basis. As argued in Section 1, regulation not only determines the variety of capitalism but it may also act as a substitute for, or a complement to government spending. An indicator for measuring government regulation on an internationally comparable basis is provided by the economic freedom of the world index taken from the Fraser Institute. GOVREG consists of the subcategories of credit market, labour market and business regulation. We inverted the original indicator so that higher values (max = 10) indicate a high level of government regulation. The EFW variables restrict our data set because EFW indices are only available on an annual basis from 2000 on.

Politics
We implemented three political control variables, which may also impact on government activity, and therefore bias our results if omitted. election accounts for the (expected) remaining years of a legislature and, hence, is zero for election years. govtype measures the size of government coalitions on a scale of one to six (single-party to caretaker government). polcon provides a commonly used complex measure of checks and balances implemented in order to limit the power of a central government. We also considered the quality of political institutions (pol) approximated by the polity IV index measuring the quality of democracy, as a control for the quality of institutions.
As a measure of ideology, that is, the ideological position of governments, we used the database of political institutions from the World Bank, which measures the ideology of the chief executive's party orientation with respect to economic policy (EXECRLC). These data are provided on a yearly basis by the World Bank and also allow for a broader coverage. However, they are based on a less disaggregated basis only (right -1, centre -2, left -3). The data operationalize party ideology according to nominal partisan labels rather than programmatic preferences of economic policies, they do not differentiate between party's and executive's orientation, and they also use indicators produced from other sources (see the criticism in Zoco, 2004, Rydland et al., 2008. However, they have been shown to be more robust to changes in the sample size and in the economic model than the alternative from the party government dataset (PGDS). 7

Economics
While the election variable takes political cycles into account, unemployment (unemp) takes economic cycles into account. We considered a number of standard controls according to the literature on public economics analysing the level of government spending, as in Ahrens et al. (2015); for a literature review, see Shelton, (2007) such as initial level of GDP (initialgdp), population growth (pop), openness (open), the ratio of elderly population (elderly) and the public debt ratio (debt). Additionally, we accounted for effects stemming from crisis development throughout the sample by adopting the indicator for financial crises based on the Laeven and Valencia (2012) database on banking crises (crises).
We used the same empirical model for our regressions to explain government regulation. This is in line with political economy models of regulation (see Feld, 2007 and, for the inclusion of partisan preferences Vaillancourt, 2007 andHeinemann, 2007). The effects were, of course, expected to differ from spending effects with respect to the standard economic variables. For example, income was expected to have a positive impact on spending because richer countries tend to spend more because of their ability to tax but a negative impact on regulation because of a substitution effect. It is also plausible that a higher share of elderly people in the population implies a higher demand for spending but has uncertain effects for regulation, and that deregulation affects the younger population in the first place.

Empirical Methodology
We adopted a time and country fixed-effect estimation in order to account for omitted variables explaining common trends over time and other country characteristics. While this may interfere with time-invariant and stable exogenous variables, especially with our country group dummies reflecting geography and economic systems, significant results for the country group dummies could be expected to be much more robust and, therefore, reliable. 8 Table 1 shows the regression results based on the database on political institutions coding for partisan ideology as well as the results for both the determinants of government spending (columns 1-5) and government regulation (columns 6-10). As a first indication of European heterogeneity with respect to government activity, the sample splits reveal significantly higher R 2 for the subsamples. In line with results from the cluster analysis on economic systems (see Ahlborn et al., 2016), the LME and EAST sub-groups show the highest level of homogeneity in terms of the explanatory power of the econometric model. In line with the literature surveys on partisan ideology (see Potrafke, 2017), the explanatory power of the model is higher for regulation than for spending, although it was adopted from the literature on government spending.

III. Empirical Results
With respect to ideology, results also differ between subsamples and between spending and regulation. Partisan ideology has no significant effect on government spending except for eastern countries. Here, partisan effects are positive in the EAST and in the CME    Table S1). For the definition of variables see text and Table S2. subsample (indicated by the positive cross-term ideologyeast). Hence, left-wing governments spend more, as is assumed by the traditional partisan theory. At the same time, more left-wing governments in eastern countries in the LME subsample spend less, as was assumed by Tavits and Letki (2009), but only after accession (indicated by the negative cross term ideologyeasteu). Hence, neglecting insignificant coefficients, a swing from centre to left-wing governments in eastern countries would imply that the government spending ratio was 1.35 per cent higher in eastern CME countries and 0.86 per cent lower in eastern LME countries after accession. Like Tavits and Letki, we could evaluate the economic significance of these ideology effects against the subsample averages of the year-on-year changes of government spending. By doing this, the traditional partisan effect for eastern CME countries accounts for 21 per cent and the reversed partisan effect for eastern LME countries after accession accounts for 70 per cent of the changes in average government spending. The latter result is consistent with the anecdotal evidence reported for Estonia that even the most left-wing parties support right-wing policies. 9 Results for the partisan effects on government regulation are most pronounced for the CME subsample. CME countries are split into a traditional partisan effect in western CME countries and a reversed partisan effect in eastern CME countries. For eastern CME countries, the cross-term (ideologyeast) more than compensates for the basic effect. The combined effect would account for 0.27 decline in the regulation index, which is 71 per cent of the average year-on-year changes of the index for the CME subsample. This result is consistent with anecdotal evidence reported for populist right-wing government behaviour in Poland and Hungary, which, against traditional behaviour of right-wing parties, regulates more. However, while this is economically significant, this reversed partisan effect in regulation only moderates the strong standard effect for government spending. For the full sample, we find that, in the case of regulation, the traditional partisan effect for the western CME countries obviously dominates in our sample of 25 European countries.
All in all, the discussion of the impact of ideology on government activity does not reveal a strong robust impact in one or the other direction. The traditionally assumed effect is revealed for spending for eastern countries, especially for the CME subsample, as well as for regulation for the full sample and the western countries in the CME subsample. Apart from this, a reversed partisan effect, as claimed by Tavits and Letki, is to be found in the LME and CME subsamples only. While the reversed partisan spending effect in LME countries is likely to imply reversed behaviour with respect to overall government activity, the reversed regulation effect in CME countries instead compensates for the traditional spending effect, with unclear consequences for overall government activity. As noted above, this result is consistent with anecdotal evidence reported for eastern countries in Section 1. 10 9 Our results are not strictly comparable to those of Tavits and Letki, who claim the effect to be twice the average change in total spending for a one standard deviation change in the ideology variable, which varies between À39 and + 36 (p. 563). Because our variable ranges between 1 (right wing) and 3 (left wing), we look directly at discrete changes from centre to left-wing governments. Changes are in absolute values and for subsamples. 10 Note that any significant and permanent reversed partisan effect refers to the level of government activity as suggested by the theory. It is not confirmed with respect to short-run effects on growth rates of government spending (see Table S3).
Overall, we conclude that partisan preferences matter less than suggested by earlier studies. This is in line with the emergence of new cleavages in party competition discussed in Section 1. In addition, the sample splits, which divide Europe into EAST/WEST and LME/CME, clearly matter beyond the impact of partisan preferences on government activity. It seems instead that the drivers of government activity differ in a more fundamental way. 11 This goes, in some way, beyond predictions on the basis of political and economic theories and, certainly, asks for future research to close these gaps.

Conclusion
This article has taken a novel look at the relationship between partisan preferences and government activity in EU countries considering the heterogeneity stemming from two major European divides into east/west and LME/CME respectively.
Concerning partisan effects, temporary or permanent reversal of the assumed normal partisan pattern resulting in lower spending and regulation by left-wing governments, are the exception rather than the rule. Even more importantly, the impact of ideology on government spending in eastern European countries also differs between LME and CME subsamples. For eastern CME countries, the reversed effects of partisan ideology on regulation, most likely stems from the non-traditional behaviour of populist right-wing parties in Poland and Hungary. This effect is compensated for by a traditional partisan effect on spending. Clear evidence for a permanently reversed partisan behaviour is revealed for LME countries such as Estonia only.
Hence, the result of Tavits and Letki is not broadly confirmed but remains a special case even among eastern EU countries. In addition, the drivers of government activity seem to differ in a more fundamental way along the east/west divide but even more pronounced also along the, so far largely neglected, LME/CME divide.

Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Table S1: LME/CME Sample Splits in an EU25 Sample Table S2: Variables, Definitions, Sources Table S3: Determinants of Government Spending and Regulation in EU25, Growth Rates (1990-2014) -Fixed Effects Estimation