decentralization as a condition of party system ... – defined as the extent to which political...

Decentralization as a Condition of Party System Nationalization: Evidence from Latin America and Central and Eastern Europe

Imke Harbers Leiden University

[email protected]

Draft Version: Please do not cite

Paper prepared for presentation at the 2008 ECPR Joint Sessions of Workshops, Rennes, April 11th-16th, 2008

Workshop 1: The Nationalization of Party Systems in Central and Eastern Europe

1

Introduction Politics at the beginning of the 20th century has often been understood as the development of a national political arena (e.g. Caramani 2004; Collier and Collier 1991 for Latin America; Diaz-Cayeros 2006 for Mexico). In Western Europe, in response to the increasing centralization of government activities, the focus of political actors – particularly political parties – had shifted from the subnational to the national level. At the end of the 20th century, however, a broad and persistent movement away from centralized patterns of governance has become one of the defining trends in politics. In third wave democracies the return to electoral politics has often coincided with institutional reforms to promote decentralization. Rather than an increasing centralization of government activities, these countries experience the transfer of resources and responsibilities to the subnational level at the time of rising electoral competitiveness. Often, decentralization has been promoted by international organizations, such as the EU, the World Bank and the Inter-American Development Bank. Particularly in Latin America and Central and Eastern Europe (CEE), where authoritarian regimes have generally been characterized by extensive centralism, decentralization constitutes a departure from previous patterns of governance. Many scholars have embraced decentralization because of potentially beneficial policy outcomes (e.g. Tibout 1956, Weingast 1995). More recently, however, concerns about how these reforms affect party systems have been voiced. Sabatini (2003), for example, argues that parties in Latin America are poorly equipped to deal with the erosion of centralized governance and he links party decline and failure in many of the new democracies to decentralization. Lalander (2003) and Ryan (2004) contend that decentralization increases party system fragmentation. Moreover, while the nationalization of politics in Western Europe has been associated with the weakening of local elites resulting in increased accountability and responsiveness, in countries democratizing as part of the third wave decentralization often seems to have strengthened local elites (Ryan 2004, Bizberg 2001). Overall, however, we have very little empirical evidence about the effect of decentralization on party system characteristics (van Biezen & Hopkin 2006: 14). Particularly the effect of decentralization on party system nationalization in new democracies remains poorly understood. This is lamentable as party system nationalization – defined as the extent to which political parties obtain similar vote shares throughout the national territory – plays an important role in shaping the terms of political competition. The extent to which vote patterns are nationalized or regionalized is likely to affect parties’ political and electoral strategies (Jones & Mainwaring 2003: 140), with important consequences for the nature of political representation. This paper explores how decentralization has affected party system nationalization in Latin America and Central and Eastern Europe. It develops the argument that in new democracies, where homogenizing functional cleavages are relatively weak, a high level of decentralization can inhibit the formation of nationalized party systems. Political decentralization, i.e. the presence of a subnational electoral arena, creates opportunities for the formation of distinct subnational party systems that structure competition for subnational offices. Fiscal decentralization provides subnational political actors with access to resources and increases their fiscal impact.

2

The first section of the paper reviews the relevant literature and discusses the mechanisms through which decentralization affects party system nationalization. Based on electoral data for 108 elections in six Eastern European and fifteen Latin American democracies, the following section explores the effect of both fiscal and political decentralization on party system nationalization empirically. The negative effect of political and fiscal decentralization on nationalization remains robust throughout different model specifications, which suggests that a trade-off exists between decentralized governance and nationalized party systems. Decentralization as a Condition of Party System Nationalization Which factors promote the development of nationalized party systems? How does decentralization affect nationalization? Previous theoretical work suggests that the centralization of government activities and party system nationalization are linked. In his historical study of Western Europe Caramani (2004) attributes party system nationalization to the declining relevance of pre-industrial territorially-based cleavages vis-à-vis socioeconomic functional cleavages. Macro-processes of political and social modernization promoted the integration of peripheries into the nation-state, while dislocating important political issues from the local to the national level contributed to the formation of a political center. The focus of political actors – particularly political parties – then shifted from the subnational to the national level. Parties abandoned territorially-focused strategies in favor of broader national strategies. What ensued was the “territorial homogenization of electoral behavior” (Caramani 2004: 1) and the creation of a national political arena where major parties are competitive throughout the territory. In Western Europe, the process of progressive nationalization was completed before the Second World War. After the war, West European countries sustained highly nationalized politics. According to Caramani, the predominance of the homogenizing left-right cleavage has created a situation in which even decentralization and the creation of federal structures have not led to a relapse to de-nationalized politics and regionalized vote patterns (2004: 291-292). Caramani’s argument emphasizes the process of nation-building through the creation of external boundaries, the formation of a political center and the resulting development of functional cleavages. This argument might not travel well to new democracies, which have experienced different processes of state formation and political modernization. One key characteristic of new democracies in Latin America as well as in Central and Eastern Europe is the relative weakness of functional homogenizing cleavages. In particular the left-right dimension has proven to be less powerful in structuring party competition (e.g. Roberts 2002, Karasimeonov 1996) and other mechanisms, such as clientelist exchanges and personalized politics, play an important role in linking voters to parties (Conaghan 1995, Levitsky 2003). In other words, while overall West European democracies have been remarkably resistant to regionalizing trends once a high level of nationalization was achieved, in new democracies the relative weakness of functionally-based cleavages might make electoral behavior more vulnerable to de-nationalization.

3

Like Caramani, Chhibber and Kollman (2004) examine party system nationalization in historical perspective. Their main interest lies with over-time differences within countries. Focusing on Canada, Great Britain, India and the United States they seek to explain why and when national parties emerged and why during certain time periods regional parties have been able to draw significant vote shares. While Caramani emphasizes a progressive development toward nationalized politics, Chhibber and Kollman identify cyclical patterns of nationalization and de-nationalization within party systems.1 Building on an actor-centered approach, the authors attribute changes in party aggregation across districts to changes in the relationship between the national and subnational governments. The mechanism encouraging party system nationalization is the strategic incentives provided to voters and candidates by the distribution of authority across levels of government. According to Chhibber and Kollman (2004: 21) “party systems become more national as governments centralize authority; in contrast, there are more opportunities for regional, state, or provincial parties to thrive as provincial or state governments gain more authority relative to the national level.” Even though their study focuses on countries with single-member districts, in principle the same logic should be at work in other electoral systems. A larger role for the central government, the authors contend, should make the party system more national regardless of the electoral system (233-234). Brancati (2008) explains the origins and strength of regional parties from an actor-centered perspective. Like Chhibber and Kollman, she claims that cleavage-based approaches are insufficient to account for the variation of regional party strength across countries and over time. Her analysis stresses the importance of institutional incentives provided to voters and candidates by political decentralization. Decentralized political systems encourage the formation and growth of regional political parties, she argues, through the presence of regional legislatures. Regional legislatures create opportunities for regional parties to govern. The realistic prospect of sharing power at the local level creates incentives for politicians to form such parties and encourages voters to support them. The argument presented in this paper builds on all three of these analyses. While it shares the focus on decentralization as one of the conditions that shape party system nationalization, it aims to extend the scope of previous work to democracies in Latin America and in Central and Eastern Europe. Decentralization and Party System Nationalization in New Democracies Decentralization disperses resources and responsibilities over multiple levels of government. It constitutes “a process of state reform…that transfers responsibilities, resources, and authority from higher to lower levels of government in the context of a specific type of state” (Falleti 2005). I argue that in the absence of strong homogenizing functional cleavages decentralization is likely to lead to a

1 The only country examined by both studies is the UK. While Chhibber and Kollman (2004: 221) point out that the UK exhibits the lowest degree of dynamic development over time, among Western European countries Caramani (2004: 291) finds the UK somewhat atypical due to its pattern of regionalization in the 1970s.

4

denationalization of the party system. Both political decentralization – the presence of a subnational electoral arena – and fiscal decentralization – the extent to which subnational governments manage resources – encourage a regionalization of the vote.2 Political decentralization creates opportunities for the formation of distinct subnational party systems that structure competition for subnational offices. Previous accounts have often conceived of subnational elections in Western Europe as instances of ‘second order elections’ (Reif & Schmitt 1980), which are conducted largely in terms of national political considerations. Yet, this congruence between the national and subnational levels is linked to the presence of the left-right dimension as “super issue” that shapes evaluations of political performance and preferences at all levels of government (Jérôme & Lewis-Beck 1999). In other words, the hierarchy of elections depends on the presence of functional cleavages. The literature on party systems outside Europe contains a wealth of information about distinct subnational party systems and recognizes such systems as widespread empirical phenomena.3 Within one country, we might thus find multiple party systems at different levels of government. Distinct systems – understood as “patterned interactions” – can develop at the subnational level because cleavage structures and the salience of particular issues at this level differ from the national level (Bardi & Mair 2008). Therefore, the introduction of subnational elections can encourage the formation of regional parties (Harmel & Robertson 1985, Brancati 2008) and forces statewide parties to deliberate consciously about the formulation of political strategies appropriate for different subnational arenas (van Houten 2003). The dynamics of coalition formation in these subnational arenas may diverge quite substantially from those at the national level. Hence, it is possible that patterns of party competition develop at the subnational level possessing all the properties of “systems” (Sartori 1976).4 If subnational party systems emerge, they are likely to affect electoral competition for national offices within the region. ‘Spill-over’ from subnational to national elections occurs, for example, because regional parties start contending national elections (Brancati 2008). Once regional parties are formed and established, the costs of competing in national elections are relatively low. Even though regional parties are unlikely to share power at the national level, contending national elections might improve their electoral prospects in the next regional election. Moreover, coalitions and alignments formed by state-wide parties for the purpose for subnational elections 2 In addition to these two dimensions, some authors distinguish an administrative dimension of decentralization (e.g. Schneider 2003). Administrative decentralization then refers to the extent to which subnational governments carry out policy responsibilities. 3 Studies of state-level party systems in the US have been particularly influential in this regard (e.g. Key 1949, Ranney & Kendal 1956, Schlesinger 1991). Recent scholarship has emphasized the emergence of “subnational authoritarianisms” and the persistence of local party systems with limited levels of competitiveness in new democracies (e.g. Fox 1994, Gibson 2005, McMann 2006). Gibson and Suarez-Cao (2007) argue that this wealth of empirical evidence about subnational party systems has remained under-theorized and propose the concept of ‘federalized party systems’ to analyze systems “in which more than one territorially delimited party system operates” (p. 6). 4 Sartori (1976) remains somewhat ambiguous about whether or not subnational party systems are indeed systems in their own right. On the one hand, he is emphasizes that party systems at different levels of government are interdependent and that “Florida or Louisiana are not states in the sense in which Mexico and Tanzania are such” (p. 83). Nevertheless, he them moves on to analyze “state party systems” in the US according to the same criteria used to evaluate national party systems (p.85).

5

are likely to constrain these parties in subsequent national electoral campaigns. Political decentralization therefore encourages the regionalization of vote patterns. If subnational elections grant political authority and legitimacy to local politicians, fiscal decentralization provides access to resources and some autonomy from the center. In an environment where ideology plays a limited role in structuring party competition, public resources, which can be turned into patronage goods, have a substantial impact on the terms of competition (e.g. Greene 2007). The availability of resources matters particularly in many new democracies, where contingent and fluid ties between voters and parties rather than programmatic linkages predominate (Roberts 2002). Empirical studies show that in a number of third wave democracies, fiscal decentralization has strengthened local elites and created an additional layer of particularistic exchanges (Ryan 2004, Bizberg 2001). Thus, while in established democracies party system denationalization has been inhibited by the predominance of homogenizing functional cleavages, in new democracies both nationalization and denationalization can occur. The extent to which party systems are nationalized is likely to depend on the degree of fiscal and political decentralization. In decentralized polities, party systems are likely to be less nationalized than in centralized polities. Data and Measurement This constitutes, to my knowledge, the first attempt to understand how political and fiscal decentralization affect party system nationalization in Latin America and Central and Eastern Europe from a cross-regional perspective. So far, literature treating party system nationalization as dependent variable has generally focused on established democracies in Western Europe or North America and to a lesser extent on India. Previous work on party system nationalization in Latin America has tended to emphasize the measurement of nationalization, rather than its explanation (e.g. Jones & Mainwaring 2003, Alemán & Kellam 2008).5 Pioneering work exploring the causes of party system nationalization in CEE has been undertaken by Bochsler (2006) and Tiemann (2006). Both studies focus on fiscal aspects of decentralization and reject the hypothesis that it has a significant effect on party system nationalization.6 The authors disagree on alternative explanations. Whereas Bochsler identifies ethnicity as key determinant of party system nationalization, in Tiemann’s analysis the effect of ethnicity is less clear-cut. Tiemann emphasizes institutional variables, particularly the

5 Even though Jones and Mainwaring’s (2003) main ambition is to put party system nationalization on the map as a dimension of party competition, they conclude by introducing potential explanatory factors for the variation in party system nationalization they observe. Firstly, federal countries have lower nationalization scores than unitary systems as such systems. Secondly, there is an inverse relationship between party system fragmentation and nationalization. Thirdly, low nationalization tends to be associated with high electoral volatility. 6 While both studies initially discover a statistical relationship between fiscal decentralization and party system nationalization for their countries, the authors conclude that the relationship is driven by statistical outliers. Bochsler attributed the relationship to the case of Bosnia whereas in Tiemann’s data the relationship appears to be driven by three Russian observations. In both studies, the initial correlation disappears once these countries are excluded.

6

electoral system, as key to understanding the formation of national party system nationalization. To explore the effect of decentralization on nationalization, I collected constituency-level electoral data from democracies in Latin America and Central and Eastern Europe. In each country, the analysis begins with the first democratic election after the beginning of the third wave of democratization for which data is available. Only countries with a score of four or higher on the Polity Index were considered democratic. Overall, party system nationalization measures were computed for a total of 108 elections in six Eastern European and fifteen Latin American countries. Dependent Variable: Nationalization, my dependent variable, refers to the extent to which subnational territorial units mirror the overall national pattern of party competition. If patterns of competition overlap, the system is considered nationalized. If patterns of competition diverge, the system is considered regionalized.7 In this analysis, vote shares have been aggregated at the level of primary administrative divisions. In federations, these are the states (Brazil, Mexico and Venezuela) or provinces (Argentina). For unitary countries, units considered are departments, regions or counties. Table 1 provides an overview of the names of administrative divisions and the included elections for each country. To measure nationalization, I use the party system nationalization score (PSNS) proposed by Jones and Mainwaring (2003). It is based on the Gini coefficient, a widely used measure of inequality, which ranges from zero in cases of perfect equality to one in cases of perfect inequality. Calculation of PSNS proceeds in two steps. First, a Gini coefficient is computed that reflects the vote distribution of each party. For this party nationalization score (PNS), a Gini coefficient of 0 signifies that a party received the same percentage of votes in every subnational unit. In a second step, the obtained Gini coefficients are multiplied by the respective party’s share of the national valid vote and then added up. Hence, the contribution of each party to the party system measure is proportionate to its share of the national vote. The overall Gini coefficient is then subtracted from 1 so that high scores indicate a high level of party system nationalization. As the calculation of the initial PNS is based on vote shares (rather than total votes), each district is weighted equally, regardless of differences in population.8

7 In addition to this conceptualization, nationalization has also been understood as the uniformity of electoral change across districts (e.g. Stokes 1967, Morgenstern & Potthoff 2005, Alemán & Kellam 2008). 8 I used the STATA command ‘ineqdec0’ (Jenkins 2001) to calculate Gini-coefficients. For a detailed account of how to calculate PNS and PSNS, please refer to Jones and Mainwaring (2003). In general, indicators of nationalization can be either party-based or district-based (e.g. Kasuya & Moenius 2007). In both cases, nationalization is calculated on the basis of lower house elections to national parliaments. Note that so far nationalization measures have focused on competition for national offices only and do not take into account patterns of party competition for subnational offices. When Chhibber and Kollman (2004) discuss district-level ‘party systems’, for example, this refers to party competition for national offices. For the purpose of this study, however, the use of constituency-level data for national elections is appropriate as I am interested in how decentralization affects national party systems.

7

Even though some concerns about this indicator have been raised (Bochsler 2006), confidence in the measure is strengthened by the fact that it compares well with other nationalization measures. It correlates highly with Chhibber and Kollman’s party aggregation measure, which suggests that both indicators measure similar processes and outcomes (2004: 175-178). Kasuya and Moenius’ district-focused measure of nationalization and PSNS also yield very similar substantive results when applied to the evolution of the US party system (2007: 8). Independent Variables: Decentralization refers to the division of labor between levels of government. Systems are decentralized to the extent that the central government possesses a smaller share of fiscal resources and grants more autonomy to subnational levels of government (Schneider 2003). Since decentralization is a relative concept, measurement is not entirely straightforward and substantial debate exists in the literature about the best way to capture the relationship between levels of government. Some scholars reject quantification of the concept altogether. Chhibber and Kollman, for example, define decentralization or provincialization, as they call it, as including perceptions of voters about the relative distribution of power between levels of government as well as de facto changes in power distribution. The concept comprises political and economic aspects such as changes in constitutional or legal authority, alterations in the relative economic role of the national or provincial governments, and fluctuations in the size of national and provincial governments. Hence, provincialization is “too multifaceted a process to peg … to a few statistics” (2004: 104). Despite these reservations, measuring levels of decentralization across countries and over time is important and in my opinion possible. To reduce the complexity of decentralization, it is helpful to distinguish between the fiscal and political dimension of decentralization. A country is considered politically decentralized if both subnational legislatures and executives are directly elected. Data for this indicator are available through the Worldbank’s Database of Political Institutions (Beck et al 2001). To measure fiscal decentralization, I have constructed a scale composed of four items: subnational spending as percentage of GDP, subnational expenditure as a percentage of GDP, subnational spending as percentage of total government spending and subnational expenditure as percentage of total government expenditure.9 These measures do not necessarily reflect how much autonomy subnational governments have with regard to the use of these resources as transfers from the central government might be earmarked for specific purposes. Nevertheless, they reflect the overall level of fiscal resources managed at the subnational level and indicate the fiscal impact of local governments (Schneider 2003: 37). The size of subnational and national governments also yields information about their relative importance. Chhibber and Kollman (2004) argue that the number of public employees at different levels of government provides a good indication of

9 The scale reliability coefficient for these four items is .86. Data have been obtained from the Government Finance Statistics CD-Rom, the Historical Government Finance Statistics CD-Rom and the Government Finance Statistics Yearbook (various years) published by the International Monetary Fund.

8

provincialization. In addition to the fiscal decentralization scale, I therefore use government employment data – obtained from the IMF’s Government Finance Statistics – to calculate the percentage of total government wages attributed to subnational governments. Table 2 summarizes the country means of the interval-level variables in the analysis. Control Variables: Party system characteristics are closely related to the social and political heterogeneity of society (Coppedge 1997, Bochsler 2006). Particularly ethnicity affects the relative ease or difficulty with which nationalized parties can be formed. The ethno-linguistic fractionalization measure used in this analysis ranges from 0 to 1, with higher values indicating a higher level of heterogeneity.10 Incentives created by the political and electoral system also affect nationalization. A substantial body of literature in comparative politics argues that concurrent presidential elections reduce the number of effective parliamentary parties. As in presidential systems the grand prize is generally awarded in a single national district, these systems encourage the formation of national alliances between parties. If presidential elections are held concurrently with lower house elections, the resulting coattails effects should promote higher nationalization (Shugart & Carey 1992, Jones 1995, Mainwaring & Shugart 1997, Cox 1997).11 In addition to presidentialism, district magnitude has been identified as affecting party system characteristics (Cox 1997). Single-member districts have been shown to reduce nationalization, while PR systems with larger districts encourage nationalization (Morgenstern & Vázquez-D’Elía 2007: 162-163). The effect of single-member districts is attributed to parties’ strategic use of resources, as they may decide not to compete in districts where their chance of winning is small, and to heterogeneity between districts as similar voters tend to be concentrated geographically. Tiemann (2006) – following Carey & Shugart 1995 and Kitschelt & Smith 2002 – shows the localizing and personalizing effect of single-member districts in CEE. In this analysis, district magnitude is measured as the log of the mean district magnitude for lower house elections.12 The Analysis Trends in Party System Nationalization A first glance at the data shows that countries in the analysis differ quite substantially both with regard to absolute levels of party system nationalization and in the development of nationalization during the time period considered. Figures 1 and 2 display trends for 23 elections in six Eastern European democracies and 85 elections in fifteen Latin American democracies. Nationalization scores range from .98 (Chile

10 For a discussion of the measure see Taylor and Hudson (1976). Data were obtained from Philip G. Roeder’s website. 11 See Golder (2006) for a nuanced account of the interaction between presidential elections, legislative fragmentation and social heterogeneity. 12 Data have been obtained from the Database of Political Institutions.

9

2005) to .38 (Argentina 2005). Argentina (standard deviation, σ = .12) is the country with the highest level of change over time, while Honduras (σ = .01) is the most stable. Overall, voting behavior in Eastern European elections appears to be slightly more nationalized than in Latin America. The average nationalization score for CEE is .83, compared to .76 in Latin America. Also, nationalization scores for elections in CEE elections appear fairly clustered (σ = .08) when compared to Latin America (σ = .14). This finding might have to be attributed to the smaller number of elections in CEE in the dataset, however. Moreover, as some Latin American countries democratized a decade before CEE, the longer experience with elections creates more room for dynamic developments over time.13 The data show that in new democracies, nationalization as well as regionalization of vote patterns can be observed. While a number of countries, such as Argentina and Colombia, move towards regionalized vote patters, party systems in Chile, Estonia and Romania nationalize. In another group of countries (e.g. Uruguay, Costa Rica and Slovenia) no trend in either direction is observable and vote patterns appear relatively stable from one election to the next. This supports the hypothesis that in the absence of strong homogenizing functional cleavages party system nationalization is fairly dynamic and over-time changes are possible. If in new democracies both nationalization and regionalization can be observed, the question remains how we can account for these trends. The following section tests the hypotheses raised above and shows that a substantive portion of the variance in PSNS – both between and within countries – can be explained by fiscal and political decentralization. Explaining Party System Nationalization As observations on the dependent variable are clustered within countries and the number of observations per country differs quite substantially, the data pose a number of econometric challenges. The analysis below is therefore conducted in several steps. I begin with basic ordinary least squares (OLS) models. To account for the hierarchical nature of the data, I also estimate the relationship using mixed-effects multi-level models.14 A first set of models ignores within country variation and addresses the clustering of data by using country means. Here, countries, rather than elections, are the relevant unit of observation. As some countries change their institutional setup during the period considered (e.g. introduce subnational elections or reform their electoral system), institutional variables cannot be included in these models. The independent

13 At present, my data is more extensive for Latin America than for Central and Eastern Europe. Therefore, conclusions drawn for CEE have to remain somewhat more cautious. In the regression analyses, I have carefully checked for outliers and examined residuals to see whether CEE departs from the overall trend. This does not appear to be the case. 14 Stata 10 is the statistical package used to estimate models. See Cohen et al. (2003), chapters 14 and 15, for a discussion of different methods for analyzing clustered data.

10

variables are the country means of subnational wage share and the fiscal decentralization scale. Figure 3 provides a first illustration of the negative relationship between subnational wage share and party system nationalization. Model 1 and 3 confirm this finding for both fiscal variables. The scatterplot also underscores the relevance of ethno-linguistic fractionalization. Ecuador, a relatively heterogeneous country, appears as an outlier below the line while relatively homogenous countries such as Slovenia and the Czech Republic come out well above the fitted line. Models 2 and 4 therefore control for ethnicity. The negative effect of fiscal decentralization on party system nationalization is consistent across all four models.15 The fiscal decentralization scale has a range of 22.8 (min: 1.2; max: 24), the predicted difference in party nationalization between the lowest and the highest level of decentralization is thus .19 (Model 4). Considering that the country mean of party system nationalization ranges from .57 to .94, this is a substantial effect. Pooling all party system nationalization scores makes it possible to include institutional variables in the analysis (see Table 4). The effect of fiscal decentralization in Model 1 is similar to the results of the country mean model, both in terms of significance and size. Model 2 explores the effect of political decentralization on party system nationalization. Again, we find the hypothesized negative effect. In politically decentralized systems party system nationalization is .12 lower than in polities without elected subnational executives. The size of the effect is somewhat reduced in Model 3, which also includes fiscal decentralization and ethno-linguistic fractionalization, but its sign and statistical significant remain. In Model 4 another control variable – concurrent presidential elections – is added. While this variable shows the expected positive sign, it fails to reach significance. The effects of political and fiscal decentralization, however, remain robust. Model 5 replicates the analysis of the previous model substituting the fiscal decentralization scale for subnational wage share, and reaches similar substantive conclusions. Model 6 includes district magnitude as final control variable. Here the results are somewhat surprising. While fiscal and political decentralization continue to show similar substantive effects, the variable controlling for concurrent presidential elections now becomes significant. The sign is positive, supporting the initial hypothesis that concurrent presidential elections have a nationalizing effect. District magnitude, however, has a negative sign, thus suggesting that party systems are more regionalized in large districts. This runs counter to theoretical expectations as well as to empirical results obtained in previous studies. To account for this finding, it is necessary to take a closer look at the dataset and, more specifically, the kind of electoral systems included in the analysis. Previous studies of district magnitude have emphasized the localizing effect of single-member districts (Tiemann 2006, Morgenstern & Vázquez-D’Elía 2007). As no Latin American country employs this formula, however, and all CEE countries included in the analysis use variants of proportional representation, the smallest value of district magnitude in the dataset is 2 (Chile).

15 Excluding federal countries, which are among the most decentralized and least nationalized in the analysis, from models 2 and 4 does not substantially alter the results.

11

In the absence of single-member districts, another effect appears to play out. Jones and Mainwaring (2003) observe an inverse relationship between party system fragmentation and nationalization. The mechanism underlying this association is based on the relative importance of smaller parties in such systems. Even though party system fragmentation and nationalization are distinct phenomena, empirically smaller parties tend to be less nationalized than larger parties. As in fragmented party systems these parties account for a substantial share of the vote, aggregate party system nationalization is lower. PR systems with large districts are associated with relatively larger party systems (e.g. Birch 2007). The negative relationship between party system nationalization and district magnitude might therefore be a consequence of the party-based nature of the party system nationalization score. Moreover, these findings suggest that the relationship between district magnitude and nationalization might in fact be curvilinear, with both very small districts and very large districts encouraging regionalization. The final set of models (Table 5) takes into account the hierarchical structure of the data and clusters observations by country. Again, the results suggest a consistently negative effect of political and fiscal decentralization on party system nationalization. In Model 5, the fiscal decentralization scale, political decentralization and ethno-linguistic fractionalization show a significant negative effect on party system nationalization, while concurrent presidential elections display a positive effect. In Model 6, the fiscal decentralization scale is replaced with subnational wage shares. While in this model the indicator does not reach common levels of significance, the sign remains negative. The negative effect of political decentralization is consistent throughout all models. Caveats and Alternative Explanations The analysis conducted above raises a number of issues that merit critical discussion. Firstly, it tests what is essentially a dynamic temporal argument cross-sectionally. Hence, the emphasis is on between country variation rather than on within country variation. Given the small number of elections, particularly in Central and Eastern Europe, however, there is insufficient cross-temporal variation to observe substantial changes over time. Moreover, party system (de-)nationalization is likely to be a gradual process where drastic changes from one election to the next are unlikely, even if decentralization has taken place. Nevertheless, the current analysis does provide evidence that party system nationalization is likely to be inhibited by high levels of decentralization. This suggests that a trade-off exists between high party system nationalization and decentralization. As both nationalized party competition and decentralization are expected to generate benefits for democratic governance, this is an interesting finding. Another criticism that might be leveled against the current analysis is that it suffers from endogeneity problems. Does decentralization really cause party system denationalization or is it the other way around? Concerns about endogeneity are inherent to all cross-temporal arguments and have also featured prominently in previous studies of party system nationalization. For the case of Bosnia, for example, Bochsler (2006) concludes that low party system nationalization and the presence of

12

ethno-regional parties caused decentralization reforms, rather than the other way around. Chhibber and Kollman (2004) acknowledge concerns about endogeneity in their analysis but argue that trends toward centralization and decentralization over time are actually the consequences of larger forces that work mostly independently of the party system. While critical and self-conscious reflections about endogeneity are in order, I agree with Chhibber and Kollman in emphasizing decentralization as a cause rather than an effect. There are two principal reasons for this. The first reason is grounded in previous studies of decentralization and subnational politics. While analyses of decentralization in Western Europe have emphasized the importance of pressures from regional parties (de Winter 1998), in new democracies decentralization is more adequately understood from a top-down perspective. Decentralization is seen as a consequence of large-scale political changes, particularly democratization (e.g. Blair 2000, Grindle 2000, Crook & Manor 1998, Nickson 1995) and the switch to a new economic model (e.g. Litvak et al 1998). Given the widespread trend toward decentralized governance, there appears to be a ‘diffusion effect’ and country-specific approaches are unable to account sufficiently for the global “wave of decentralization” (Manor 1999; Eaton & Dickovick 2004: 90). Case studies of parties and party systems in Latin America show that after decentralization local politics reinvigorated. The recent emergence of indigenous parties in Latin America has been explained in terms of opportunities created by decentralization for the formation of such parties (Van Cott 2003, Rice & Van Cott 2006). In Mexico regional cleavages, which had been muted through the extensive centralization of politics under the hegemonic party, reappeared after the introduction of decentralization reforms (Magaloni 1999, Poiré 1999, Alemán & Kellam 2008). Distinct subnational party systems emerged and vote patterns in national elections regionalized. Overall, there is substantial evidence that local political actors benefited from the opportunity structure created by decentralization. The second reason why in my opinion it is more appropriate to treat decentralization as a cause, rather than an effect, derives from a closer examination of the data at hand. Colombia and Venezuela allow us to explore the effect of political decentralization, as both countries embarked on a process of political decentralization in the late 1980s (Lalander 2003).16 In both countries, the party system nationalization scores show a steep decline in the period following these reforms (Figure 1). Even though for most countries in the analysis cross-temporal variation is limited due to the small number of time points, a close look at the case of Argentina, the country with the largest number of time points in the dataset, allows us explore the direction of causality. Figure 4 shows that trends in party system nationalization and fiscal decentralization are closely related.17

16 Colombia introduced direct elections for governors in 1991, Venezuela began the process of decentralization in 1989. 17 Please note that in Figure 4 the coding of the decentralization scale used in the statistical analyses has been reversed, so that higher values of the scale reflect a higher level of centralization. Bivariate time-series analyses of the fiscal centralization scale (fisc) and party system nationalization (psns) indicate that fisc granger-causes psns. Hence, fisc precedes psns in the causal chain. Yet, given the small number of observations, inferences drawn from the time-series analyses have to remain tentative.

13

In a first exploration of possible causes of the striking denationalization of Argentine electoral competition, Leiras (2007a, 2007b) attributes this pattern to the consolidation of political machines at the provincial level. As programmatic differences between parties and party factions declined nationally, parties became more regionally based. Because provincial politicians – especially governors – are endowed with generous resources, they do not depend on presidential spoils. Moreover, federal institutions provide them with ample opportunities to influence national policy irrespective of their relationship with the president. Under these conditions, provincial party branches experience feeble incentives to commit to their parties’ federal electoral campaigns. This has resulted in diverging patterns of party competition across provinces. Concluding Remarks In Latin America and CEE, regionalization as well as nationalization of vote patterns can be observed. While in established democracies party system nationalization has been associated with the development of functional cleavages and the centralization of government activities, so far we know relatively little about how this relationship plays out in new democracies. This paper has explored political and fiscal decentralization as explanations of different trends in the formation of nationalized party systems and finds that both political and fiscal decentralization are associated with regionalized vote patterns. Political decentralization creates opportunities for the formation of distinct subnational party systems that structure competition for subnational offices. If such systems emerge, their dynamics are likely to affect vote patterns in national competitions. Fiscal decentralization provides subnational political actors with access to resources and creates opportunities for the cultivation of subnational clientelist networks and particularistic exchanges. As both political and fiscal decentralization encourage the regionalization of vote patterns, a trade-off appears to exist between party system nationalization and decentralized governance. Considering that nationalized party competition as well as decentralization are expected to generate benefits for democratic governance, this is an interesting result.

14

References Alemán, Eduardo and Marisa Kellam (2008, in press). “The Nationalization of

Electoral Change in the Americas”, Electoral Studies. Bardi, Luciano and Peter Mair (2008). “The Parameters of Party Systems” in: Party

Politics, Vol. 14(2), pp. 147-166. Beck, Thorsten, George Clarke, Alberto Groff, Philip Keefer, and Patrick Walsh,

(2001). "New tools in comparative political economy: The Database of Political Institutions." in: World Bank Economic Review, Vol. 15(1), pp. 165-176.

Biezen, Ingrid van and Jonathan Hopkin (2006). “Party organisation in multi-level contexts”, in: Dan Hough and Charlie Jeffery, eds. Devolution and Electoral Politics, Manchester: Manchester University Press, pp. 14-36.

Birch, Sarah (2007). “Electoral Systems” in: Stephen White, Judy Batt and Paul G. Lewis, eds. Developments in Central and East European Politics 4. Basingstroke: Palgrave.

Bizberg, Ilan (2001). “Behind the Scenes – Clientele and Citizens in the Mexican Political Transition” in: Iberoamericana, Vol. 1(2), pp. 87-105.

Blair, Harry (2000). “Participation and Accountability at the Periphery: Democratic Local Governance in Six Countries” in: World Development, Vol. 28(1), pp. 21-39.

Bochsler, Daniel (2006) “The Nationalization of Political Parties: A Triangle Model, applied on the Eastern and Central European Countries”, CEU Political Science Journal, Vol. 1(4).

Brancati, Dawn (2006). “Decentralization: Fueling the Fire or Dampening the Flames of Ethnic Conflict and Secessionism?” in: International Organization, Vol. 60, pp. 651-685.

Brancati, Dawn (2008, in press). “The Origins and Strength of Regional Parties” in: British Journal of Political Science, Vol.38, pp. 135-159.

Caramani, Daniele (2004). The Nationalization of Politics: The Formation of National Electorates and Party Systems in Western Europe. Cambridge: Cambridge University Press.

Carey, John M. and Matthew S. Shugart (1995). “Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas”, in: Electoral Studies, Vol. 14(4), pp. 417-439.

Cohen, Jacob, Patricia Cohen, Stephen G. West and Leona S. Aiken (2003). Applied Multiple Regression/Behavioral Analysis for the Social Sciences. Mahwah, N.J. : L. Erlbaum Associates.

Collier, Ruth Berins and David Collier (1991). Shaping the Political Arena: Critical Junctures, the Labor Movement, and Regime Dynamics in Latin America. Princeton: Princeton University Press.

Conaghan, Catherine (1995). “Politicians Against Parties: Discord and Disconnection in Ecuador's Party System” in: Scott Mainwaring and Timothy Scully, eds. Building Democratic Institutions: Party Systems in Latin America. Stanford: Stanford University Press, p.434-458.

Cox, Gary (1997). Making Votes Count. New York: Cambridge University Press. Crook, Richard C. and James Manor (1998). Democracy and Decentralisation in

South Asia and West Africa: Participation, Accountability and Performance, Cambridge: Cambridge University Press.

15

Diaz-Cayeros, Alberto (2006). Federalism, Fiscal Authority, and Centralization in Latin America. Cambridge: Cambridge University Press.

Eaton, Kent and J. Tyler Dickovick (2004) “The Politics of Re-Centralization in Argentina and Brazil” in: Latin American Research Review, Vol. 39(1), pp. 90-122.

Falleti, Tulia G. (2005) “A Sequential Theory of Decentralization: Latin American Cases in Comparative Perspective” in: American Political Science Review, Vol.99(3), pp.327-346.

Fox, Jonathan (1994) “Latin America’s Emerging Local Politics” in: Journal of Democracy, Vol. 5, pp. 105-116.

Gibson, Edward L. (2005). “Boundary Control: Subnational Authoritarianism in Democratic Countries”, in: World Politics, Vol. 58(1), pp. 101-132.

Gibson, Edward L. and Julieta Suarez-Cao (2007) “Competition and Power in Federalized Party Systems”, Paper Presented at the Annual Meeting of the American Political Science Association, August 29-September 1st, Chicago, IL

Golder, Matt (2006). “Presidential Coattails and Legislative Fragmentation”, in: American Journal of Political Science, Vol.50(1), pp. 34-48.

Grindle, Merilee S. (2000) Audacious Reforms: Institutional Invention and Democracy in Latin America, Baltimore: The Johns Hopkins University Press.

Harmel, Robert, and John D. Robertson (1985). “Formation and Success of New Parties: A Cross-National Analysis” in: International Political Science Review, Vol.6(4), pp. 501-23.

Houten, Peter van (2003) “Territorial Aspects of West European Party Politics: National Parties in Regional Party Systems”, Working Paper.

Jenkins, Stephen P. (2001) ‘ineqdec0.ado’, STATA Command. University of Essex. Jérome, Bruno and Michael S. Lewis-Beck (1999). “Is local politics local? French

Evidence” in: European Journal of Political Research, Vol. 35(2), pp. 181-197.

Jones, Mark (1995). Electoral Laws and the Survival of Presidential Democracies. Notre Dame: University of Notre Dame Press.

Jones, Mark and Scott Mainwaring (2003). “The Nationalization of Parties and Party Systems: An Empirical Measure and an Application to the Americas” in: Party Politics, Vol. 9, pp. 139-166.

Karasimeonov, Georgi (1996). “Bulgaria’s New Party System” in: Geoffrey Pridham and Paul G. Lewis, eds. Stabilizing Fragile Democracies. London and New York: Routledge.

Kasuya, Yuko and Johannes Moenius (2007, in press). “The Nationalization of Party Systems: Conceptual issues and alternative district-focused measures”, Electoral Studies.

Key, V.O. (1949). Southern Politics in State and Nation. Knoxville: University of Tennessee Press.

Kitschelt, Herbert and Regina Smyth (2002). “Programmatic Party Cohesion in Postcommunist Democracies. Russia in Comparative Context” Comparative Political Studies, Vol.35(10), p. 1228-1256.

Lalander, Richard O. (2003). “Decentralization and the Party System in Venezuela” in: Iberoamericana. Nordic Journal of Latin American and Caribbean Studies, Vol. 33(1), pp. 97-121.

16

Leiras, Marcelo (2007a). Todos los caballos del rey: La integración de los partidos políticos y el gobierno democrático de la Argentina, 1995-2003. Buenos Aires, Prometeo.

Leiras, Marcelo (2007b). “Intra-Party Exchanges and the Dynamics of Presidential Coalitions in Argentina”, Paper presented at the XXVII International Congress of the Latin American Studies Association, September 5-8, 2007, Montreal, Canada.

Levitsky, Steven (2003). Transforming Labor-Based Parties in Latin America: Argentine Peronism in Comparative Perspective. New York: Cambridge University Press.

Litvack, Jennie, Junaid Ahmad, Richard Bird (1998). Rethinking Decentralization in Developing Countries, Sector Studies Series, Washington, D.C.: The World Bank.

Mainwaring, Scott and Matthew Soberg Shugart, eds (1997). Presidentialism and Democracy in Latin America. New York: Cambridge University Press.

Manor, James (1999). The Political Economy of Democratic Decentralization. Washington, D.C.: The World Bank.

Marshall, Monty G. and Keith Jaggers (2002). Polity IV Data Set. [Computer file; version p4v2002] College Park, MD: Center for International Development and Conflict Management, University of Maryland.

Morgenstern, Scott and Richard F. Potthoff (2005). “The components of elections: district heterogeneity, district-time effects and volatility” in: Electoral Studies, Vol. 24, pp. 17-40.

Morgenstern, Scott and Javier Vázquez-D’Elía (2007). “Electoral Laws, Parties, and Party Systems in Latin America” in: Annual Review of Political Science, Vol. 10, pp. 143-168.

Nickson, R. Andrew (1995). Local Government in Latin America. Boulder: Lynne Rienner.

Ranny, Austin and Willmoore Kendall (1956). Democracy and the American Party System. New York: Harcourt.

Reif, Karlheinz and Hermann Schmitt (1980). „Nine second-order elections: A conceptual framework for the analysis of European election results”, in: European Journal of Political Research, Vol. 8, pp.3-44.

Roberts, Kenneth M. (2002) “Party-society linkages and democratic representation in Latin America”, in: Canadian Journal of Latin American and Caribbean Studies, Vol.27(53), Spring: pp.9-34.

Roeder, Philip G. (2001) "Ethnolinguistic Fractionalization (ELF) Indices, 1961 and 1985." http//:weber.ucsd.edu\~proeder\elf.htm.

Rice, Roberta and Donna Lee van Cott (2006) “The Emergence and Performance of Indigenous Peoples’ Parties in South America: A Subnational Statistical Analysis”, in: Comparative Political Studies, Vol.39(6), pp. 709-732.

Ryan, Jeffrey J. (2004). “Decentralization and Democratic Instability: The Case of Costa Rica” in: Public Administration Review, Vol. 64(1), pp. 81-91.

Sabatini, Christopher (2003). “Latin America’s Lost Illusions: Decentralization and Political Parties” in: Journal of Democracy, Vol. 14(2), pp. 138-150.

Sartori, Giovanni (1976). Parties and Party Systems: A Framework for Analysis. New York: Cambridge University Press.

Schattschneider, E.E. (1960). The Semi-Sovereign People: A Realist’s View of Democracy in America. New York: Holt, Rinehart and Winston.

17

Schlesinger, Joseph A. (1991). Political Parties and the Winning of Office. Ann Arbor: University of Michigan Press.

Schneider, Aaron (2003). “Decentralization: Conceptualization and Measurement” in: Studies in Comparative International Development. Vol. 38(3), pp. 32-56.

Shugart, Matthew Soberg and John M. Carey (1992). Presidents and Assemblies: Constitutional Design and Electoral Dynamics. Cambridge: Cambridge University Press.

StataCorp (2007). Stata Statistical Software: Release 10. College Station: StataCorp LP.

Stokes, Donald (1967). “Parties and the Nationalization of Electoral Forces” in: William Nisbet Chambers and Walter Dean Burnham (eds). The American Party Systems: Stages of Political Development. New York: Oxford University Press, pp. 182-202.

Sundquist, James L. (1973). Dynamic of the Party System: Alignment and Realignment of Political Parties in the United States. Washington, D.D.: Brookings Institution.

Taylor, Charles Lewis and Michael C. Hudson (1976). World Handbook of Political and Social Indicators. New Haven: Yale University Press.

Tiebout, Charles M. (1956). “A Pure Theory of Local Expenditures” The Journal of Political Economy, Vol. 64(5), pp. 416-424.

Tiemann, Guido (2006). Wahlsysteme, Parteiensysteme und politische Repräsentation in Osteuropa. Wiesbaden: Verlag für Sozialwissenschaften.

Van Cott, Donna Lee (2003). “Institutional Change and Ethnic Parties in South America”, in: Latin American Politics and Society, Vol 45(2), pp. 1-39.

Weingast, Barry R. (1995). “The Economic Role of Political Institutions: Market-Preserving Federalism and Economic Development” in: Journal of Law, Economics, & Organization, Vol. (1), pp. 1-31.

Winter, Lieven de and Huri Tursan, eds (1998). Regionalist Parties in Western Europe, New York: Routledge.

18

Electoral Databases: Brancati, Dawn. (2007). Constituency-Level Elections (CLE) Dataset. New York:

New York. (http://www.cle.wustl.edu/index.php). Carr, Adam. Psephos- Adam Carr's Election Archive (http://psephos.adam-carr.net/). Corte Nacional Electoral de Bolivia. (http://www.cne.org.bo/proces%5Felectoral). Election Resources. (www.electionresources.org). Estonian National Electoral Committee. (http://www.vvk.ee/engindex.html). Gobierno de Chile, Miniterio Interior (http://www.elecciones.gov.cl). Instituto Federal Electoral de Mexico. (www.ife.org.mx). Lijphart Election Archive. (http://dodgson.ucsd.edu/lij). Marconi Nicolau, Jairo Cesar. Dados Eleitorais do Brasil (1982-2006).

(http://jaironicolau.iuperj.br/banco2004.html). Popescu, Marina and Martin Hannavy (2001). Project on Political Transformation

and the Electoral Process in Post-Communist Europe. University of Essex. (http://www.essex.ac.uk/elections).

Singer, Matthew M. (2004). District Level Electoral Data on the Internet. Duke

University. (http://www.duke.edu/~mms16/DistrictLeveldatasources.htm). Statistical Office of the Republic of Slovenia (www.stat.si). Tribunal Supremo de Elecciones. República de Costa Rica.

(http://www.tse.go.cr/escrutinio_f2006/Diputados/0.htm). Tribunal Supremo Electoral Honduras. (http://www.tse.hn/).

19

Tables and Figures Table 1: Administrative Divisions and Included Elections Country Territorial

Unit Number of Units

Elections Included

Elections not Included

Argentina Province 24 1983-2005 2003, 2007 Bolivia Department 9 1985-2005

Brazil State 27 1986-2006 Bulgaria Regions 28 1991-2001 2005 Chile Region 13 1989-2005 Colombia Departments 33 (26) 1978-2006 1991, 1994 Costa Rica Province 7 1982-2006 Czech Republic Region 14 (8) 1996-2006 Ecuador Province 22 (20) 1979-2006 1998, 2002 El Salvador Department 14 1994-2003 1985, 1988,

1991, 2006 Estonia Counties 19 (12) 1991-2007 Guatemala Department 22 1990-1999 1994, 2003,

2007 Honduras Department 18 1985-2005 Mexico States 32 1994-2006 Nicaragua Department 17 1990-2001 2006 Peru Region 26 1980-1990,

2001-2006

Poland Electoral Districts

52(41) 1993-2001 2005, 2007

Romania Counties 42 1992-2004 Slovenia Electoral

District 8 1996-2004 1992

Uruguay Department 19 1989-2004 Venezuela State 24 1983-1998 2000, 2005 Note: If new regions were created during the period considered, the initial number of districts is indicated in parentheses. For Estonia, it is not possible to disaggregate electoral data completely for the 15 counties, as some smaller counties vote together in national elections. During the 1990s, Peru experienced a period of non-democratic governance. Even though elections were conducted during this period, they did not meet democratic criteria and are therefore excluded from the analysis. In Poland, the number of provinces (voivodeship) was reduced from 49 to 16 in 1999. As I was unable to obtain electoral data disaggregated at the level of provinces, calculations are based on electoral districts. Slovenian data is also based on electoral districts.

Table 2a: Country Means for Dependent and Independent Variables in Central and Eastern Europe Party System

Nationalization Fiscal Decentralization Scale

Subnational Revenue as Percentage of GDP

Subnational Expenditure as Percentage of GDP

Subnational Expenditure as Percentage of Total Expenditure

Subnational Revenue as Percentage of Total Revenue

Subnational Wages as Proportion of Total Government Wages

Ethno-Linguistic Fractionalization

Bulgaria .81 (4)

11.61 6.19 (16)

8.08 (16)

16.96 (17)

14.36 (17)

.51 (17)

0.23

Czech Republic

.91 (4)

14.03 8.29 (14)

8.67 (14)

19.34 (14)

19.82 (14)

.23 (14)

0.11

Estonia .77 (5)

14.83 7.55 (16)

8.67 (16)

23.14 (16)

19.97 (16)

.43 (12)

0.53

Hungary - 14.46 6.07 (9)

14.68 (9)

22.76 (10)

12.65 (10)

.59 (13)

0.01

Latvia - 15.31 7.51 (6)

10.23 (6)

23.99 (6)

19.51 (6)

.45 (13)

0.61

Lithuania - 22.03 - - 24.09 (10)

20.86 (8)

.50 (11)

0.35

Poland .78 (3)

15.62 9.64 (10)

10.61 (10)

22.64 (10)

19.51 (9)

.50 (10)

0.04

Romania .78 (4)

8.73 4.14 (15)

4.72 (15)

13.60 (15)

12.44 (15)

.18 (15)

0.21

Slovakia - 7.77 4.90 (11)

4.13 (11)

9.97 (11)

12.08 (11)

.22 (11)

0.24

Slovenia .94 (3)

9.60 6.36 (14)

6.05 (14)

12.91 (15)

12.55 (15)

.22 (15)

0.18

Note: Number of observations indicated in parentheses.

21

Table 2b: Country Means for Dependent and Independent Variables in Latin America Party System








Argentina .69 (11)

23.97 8.24 (22)

10.09 (22)

40.54 (22)

37.01 (22)

.68 (20)

0.29

Bolivia .76 (6)

13.18 6.20 (25)

6.13 (22)

22.24 (21)

24.39 (22)

.29 (24)

0.74

Brazil .59 (6)

22.68 10.57 (12)

16.28 (12)

35.50 (12)

28.36 (12)

.67 (12)

0.58

Chile .89 (5)

5.39 2.06 (16)

2.05 (16)

9.13 (16)

8.31 (16)

.13 (15)

0.52

Colombia .69 (7)

13.81 5.17 (14)

6.21 (14)

27.10 (14)

18.51 (9)

.45 (13)

0.60

Costa Rica .88 (7)

2.08 .76 (26)

.78 (26)

3.30 (26)

3.47 (26)

.04 (14)

0.46

Ecuador .57 (9)

10.30 2.51 (2)

2.71 (2)

17.92 (2)

18.04 (2)

.18 (1)

0.66

El Salvador .85 (4)

2.05 1.17 (3)

.78 (3)

- 5.29 (1)

.06 (5)

0.16

Guatemala .76 (3)

3.52 .34 (4)

.98 (4)

9.09 (4)

3.68 (4)

.05 (4)

0.76

Honduras .92 (6)

- - - - - - 0.12

Mexico .82 (5)

17.64 5.90 (7)

6.49 (7)

29.82 (7)

28.37 (7)

- 0.22

22

Table 2b: Country Means for Dependent and Independent Variables in Latin America (continued) Party System








Nicaragua .86 (3)

5.49 2.32 (5)

2.66 (5)

7.98 (5)

9.02 (5)

.08 (4)

0.39

Peru .72 (5)

8.44 2.71 (16)

4.24 (16)

19.10 (16)

6.54 (10)

.32 (16)

0.51

Uruguay .87 (4)

7.69 - 3.19 (9)

10.27 (9)

10.55 (7)

- 0.38

Venezuela .76 (4)

1.19 .31 (2)

.64 (2)

2.54 (2)

1.24 (2)

- 0.52

Note: Number of observations indicated in parentheses.

Figure 1: Party System Nationalization in Latin America

Figure 2: Party System Nationalization in Central and Eastern Europe

Table 3: OLS Regression on Party System Nationalization (Country Mean) (1) (2) (3) (4)

Subnational Wage Share (Country Mean)

-.260** (.109)

-.289*** (.085)

Fiscal Decentralization Scale (Country Mean)

-.007** (.003)

-.008*** (.003)

Ethno-linguistic Fractionalization

-.262*** (.079)

-.273*** (.077)

Constant .857*** (.039)

.972*** (.046)

.858*** (.039)

.982*** (.047)

N Adjusted R2

17 0.224

17 0.529

20 0.155

20 0.487

Note: *** p<.01, ** p< .05, * p<.10 Data on subnational wage shares are not available for Mexico, Uruguay and Venezuela. Breusch-Pagan tests indicate that heteroscedasticity is not a problem in these models. Figure 3: Party System Nationalization and Subnational Wage Share

26

Table 4: Pooled OLS Regression on Party System Nationalization Scores

(1) (2) (3) (4) (5) (6) Fiscal Decentralization Scale

-.009*** (.002)

-.008*** (.002)

-.007*** (.002)

-.007*** (.002)

Subnational Wage Share

-.201** (.092)

Political Decentralization

-.116*** (.028)

-.068** (.029)

-.067** (.029)

-.086* (.047)

-.051* (.027)


-.264*** (.056)

-.298*** (.064)

-.286*** (.079)

-.513*** (.082)

Concurrent Presidential Elections

.034 (.256)

.014 (.034)

.089*** (.026)

District magnitude

-.056*** (.021)

Constant .890*** (.022)

.827*** (.013)

1.013*** (.033)

1.002*** (.036)

.987*** (.043)

1.16*** (.063)

N Adjusted R²

75 .292

87 .201

63 .504

63 .485

52 .458

51 .663

Note: *** p ≤.01, ** p≤ .05, * p≤.10 Robust standard errors were used in models 2 and 3. In these models, the fiscal decentralization scale is constructed as a three year moving average.

27

Table 5: Linear Mixed-Effects Model on Party System Nationalization Scores (1) (2) (3) (4) (5) (6) Fixed Part Fiscal Decentralization Scale

-.007*** (.002)

-.004* (.003)

-.006** (.002)

Subnational Wage Share

-.177** (.08)

-.109 (.089)

Political Decentralization

-.152*** (.033)

-.104** (.040)

-.086** (.035)

-.134*** (.048)


-.331*** (.092)

-.279*** (.102)

Concurrent Presidential Elections

.05* (.028)

.037 (.034)

Constant .867*** (.031)

.835*** (.033)

.835*** (.027)

.868*** (.036)

.987*** (.05)

.954*** (.054)

Random Part Random-intercept standard deviation

.067 (.017)

.077 (.023)

.101 (.021)

.085 (.024)

.058 (.021)

.056 (.025)

Residual standard deviation (√θ)

.079 (.008)

.084 (.009)

.074 (.006)

.076 (.008)

.075 (.008)

.080 (.01)

N Pseudo R²

75 .301

65 .281

87 .201

63 .302

63 .509

51 .52

Note: *** p ≤.01, ** p≤ .05, * p≤.10

28

Figure 4: Party System Nationalization and Fiscal Centralization in Argentina

decentralization as a condition of party system ... – defined as the extent to which political...

Documents