Electoral Systems and Party Systems: Which is the Cause and Which is the Consequence? Mette Bakken August

Electoral Systems and Party Systems: Which is the Cause and Which is the Consequence? Mette Bakken August 2008 1 Abstract: Electoral systems research can be divided into two main branches: the study of electoral system consequences and electoral systems as consequences. Whilst aiming to explain different causal processes with the former treating electoral systems as an independent and the latter as the dependent variable, the two areas are intimately linked by the way in which electoral system choices are political constructs maintained or changed by democratically elected politicians. This paper explores the relationship between electoral systems and party systems: does electoral system determine party system or is it the other way around? More specifically, it centers on the impact of change and persistence. Two hypotheses can be formulated: changes in the party system provoke electoral system reform (H1) and reforms subsequently trigger party system change (H2). Where party configurations are stable, electoral reforms are not expected and where no reforms are introduced, existing party systems are anticipated to endure. The paper is constructed in four parts. The first section reviews the Duverger tradition and the new agenda literature with particular focus on their views as regards the electoral system party system relationship. Part two and three presents the theoretical framework of analysis and provides an overview over data and methodology used. Part four analyzes the impact of party system developments on electoral system reform and vice versa. The consistency between theoretical expectation and empirical evidence is discussed and the relevance of the two theoretical perspectives is compared in the conclusion. (Work in progress do not quote) 1 Mette Bakken is a PhD candidate at the European University Institute (contact: mette.bakken@eui.eu). This paper was prepared for the ECPR Graduate Conference in Barcelona 25-27 August 2008. 1

1. Previous research: the Duverger tradition and the new agenda Duverger s seminal work represents the starting point of modern electoral system research. At the core of his work are the famous propositions on the relationship between electoral systems and the number of parties: (1) Proportional representation tends to lead to the formation of many independent parties (2) the two ballot majority system tends to lead to the formulation of many parties that are allied with each other, (3) the plurality rule tends to produce a two-party system (quoted in Duverger 1986, 70). According to Duverger, two factors account for the reductive and multiplying effects of plurality and proportional electoral systems, respectively. First, the mechanical effect of the plurality formula and single-member districts directly affects the number of parties by producing highly disproportional results in favour of the bigger parties and, on the contrary, disfavouring the smaller ones. Second, the electoral system influences party system development indirectly through the psychological effect operating on two levels: within the political elite and the electorate. Under single-member district plurality rules, strategically minded voters will refrain from wasting their votes on parties and candidates that are not likely to make it to the top and thereby to reduce their numbers. Moreover, parties are encouraged to consolidate in two blocks : parties considered to be out of the race are persuaded to withdraw their candidates and to support any of the two main parties to prevent the least favourable candidate or party from winning. Under more proportional electoral systems, the mechanical effect is reduced and, consequently, also the psychological reductive effect. Hence, such systems tend to increase the number of competitors. Being among the most well-known and cited assertions in comparative politics, Duverger s propositions have been the subject of considerable theoretical and empirical examination and revision. In the most pure Duvergian tradition, Lijphart (1994), Rae (1976), Sartori (1997) and Taagepera and Shugart (1989) have studied the relationship between electoral system and the number of parties from a comparative empirical perspective. In the attempt to discover what properties within the electoral systems that contribute to the reductive or multiplying effects of various systems, special attention has been given to the effects of district magnitude, electoral formula and thresholds. Alongside the third wave of democratization, the geographical scope of similar studies has expanded to the new democracies in Eastern Europe (Moser 1999; Birch 2003), Latin America (Jones 1993) and 2

Africa (Lindberg 2005). Others have focused more exclusively on the anticipated psychological aspect of various electoral system designs such as electoral system effects on internal party development (Bogdanor and Butler 1983; Katz 1980), voter turnout (Blais 2000; Downs 1957; Dhillon and Peralta 2002) and strategic voting (Fisher 2004). Finally, attention has been given to the interactive effects of the party and voter level strategic behaviour (Blais and Carty 1991). A subfield worth mentioning in this respect is studies of electoral system consequences on women representation (Ballington and Karam 2005; Matland and Tylor 1997; Zimmerman and Rule 1994) and the representation of minority groups (Lijphart 1991a, 1991b; Horowitz 1991; Reilly 2002) in the legislatures. Methodologically, the Duverger propositions have been attacked from two main quarters. First, Wildavsky (1959) has pointed towards the logical fallacy of Duverger s argument in that, while acknowledging peripheral factors as to explain for the failure of reality, he nevertheless refrains from recognizing the effects of such factors, thereby rejecting a multicausal approach and firmly defending the exclusive explanatory power of electoral system on party system development. 2 Second, and more relevant here, it has been suggested that Duverger mistook the direction of causality and that it is party systems determine electoral systems rather then vice versa (Farrell 2001; Grumm 1958; Sartori 1997). Empirical investigations have challenged the conventional causal relationship between the two factors. Rokkan (1970) suggested that the shift from majoritarian to proportional electoral systems in continental Europe in the beginning of the 20th century came about as a strategic response on part of the bourgeois parties threatened by the introduction of the universal suffrage, the emergence of new political parties, and, in length, the prospects of becoming in a minority position. Boix (1999) has developed and formalized the Rokkan hypothesis. In Setting the rules of the game, he concludes that [t]he selection (or preservation) of different electoral rules can be traced to the strategic decisions made by the current ruling parties, foreseeing the coordinating consequences of different electoral systems, to maximize their representation in parliament (Boix 1999, 621). Further investigations carried out by Colomer (2005) suggest that party system expansion usually have preceded shifts to more proportional electoral rules. Based on these findings, Benoit (2007: x) asserts that [t]he 2 Hence, Duverger reduces the importance of societal cleavages on party system development, as emphasized by Lipset and Rokkan (1967), to a residual error, something that might perturb a party system away from its central tendency defined by the electoral law (Cox 1997). According to Wildavsky (1959: 303, methodological inadequacies [therefore] prevent Duverger from fulfilling ( ) his stated objective of presenting a general theory of parties. 3

influence of electoral system on party systems cannot be established independently ( ) because the causal relationship is bi-directional, or at worst, reversed entirely. Notwithstanding early calls for further investigations of the causal direction of Duverger s propositions (1976; Rokkan 1970), research taking electoral system as the dependent rather than the independent variable remained largely disregarded. In the beginning of the 1990s, however, the major changes undertaken in established democracies such as New Zealand, Italy and Japan refuted the idea that electoral systems tend to be very stable and resist change (Lijphart 1994: 52) and that fundamental changes are rare and arise only in extraordinary historical situation (Nohlen 1984: 218). Moreover, the fall of the authoritarian and communist regimes in Latin America, Africa and Eastern Europe generated extensive activity in the field of electoral system design as these countries embarked on writing new democratic constitutions. Together with a general trend in comparative politics towards explaining institutions, this contributed to the vitalization of the new agenda within electoral system research focusing on how electoral systems themselves come into being, are maintained and changed. Whilst a theoretical model of electoral system choice, persistence and change has not yet been developed, researchers have covered substantial ground over the past fifteen years. The first distinction can be made between interest-based and none-interest-based choice. Interestbased explanations take the expected consequences of electoral system as the point of departure and are therefore intimately linked with the Duverger research agenda. Owing to the distinct character of electoral systems in particular with reference to its reflexive and redistributive characteristics whereby i) it is the politicians, who are in the end most directly affected by the system in place, who determine the rules by which they themselves will be elected in the future and ii) the game of choosing among rules is played out in a sum-zero way which means that some can only win on the expense of others self-interest based explanations for change have attracted considerable attention among researchers. Selfinterests can be based on the particular interests of the politicians, the parties and/or the existing party system (see e.g. Benoit 2004). It should be noted that self-interests in terms of a politician s career-opportunities, the party s seat-or policy-maximizing objectives and party systems cartel-protection may create coordination problems when it comes to identifying self-interest based outcomes of electoral system choice. 4

The more credible interest-based explanation takes into account the wider implications of electoral system choice on democratic development. By indirectly affecting and structuring numerous aspects of political life, such as representation and accountability, types of government and government stability, party system development, conflict management and the like, electoral systems have been described as the cogs that keeps the wheels of democracy properly functioning (Farrell 2001: 2). General-interest based (or democracymaximizing) explanations assume that politicians act on the basis of the logic of appropriateness and genuinely strive to put in place the most truly democratic electoral system (see e.g. Blais and Massicotte 1997). With its reliance upon the altruist virtues of the politicians, the general-interest perspective represents the polar opposite to the models based on egoistic self-interests of the decision-makers. None-interest-based explanations, on the other hand, emphasize factors that affect electoral system choice beyond intentional design based on expected outcomes. These explanations refer to institutional constraints, historical, economic and political aspects and dynamics, regional and temporal diffusion, as well as idiosyncratic factors and accidents. Thus, the key feature distinguishing interests-based explanations from none-interest-based explanations refers to the extent to which the alternatives are systematically evaluated on the basis of their consequences or rather products of convergent contextual forces (Benoit 2004). Figure 1 provides a typology over the mentioned explanations for electoral system choice. Empirical research in the area of electoral system choice has expanded significantly over the past fifteen year period. Numerous single-country studies have accounted for modes of change and the factors that determine outcomes in established democracies as well as democratizing and transitional states (see i.e. Bawn 1993; Benoit and Hayden 2001; Benoit and Schiemann 2001; Brady and Mo 1992; Denemark 2001; Díaz-Cayeros and Magaloni 2001; Donovan 1995; Hooghe, Maddens, and Noppe 2006; Kaminski 2002; Nagel 1994; Navarra and Sobbrio 2001; Remington and Smith 1996; Vowles 1995). Larger editorials such as Shugart and Wattenberg s Mixed Member Electoral Systems: The Best of Both Worlds (2001), Colomer s Handbook of Electoral System Choice (2004d) and Gallagher and Mitchell s The Politics of Electoral Systems (2005) have contributed with additional and updated case studies of electoral system change and persistence. The influx of case-studies in the area has, with thick descriptions of socio-political context within which the process of change take place, 5

provided an ample ground for comparative investigations (Birch 2005; Birch et al. 2002; Colomer 2004b; Colomer 2005; Croissant, Bruns, and John 2002; Gallagher 2005, Katz 2005), although still remaining few in numbers. Figure 1: Explaining electoral system choice: a typology. Explanations for electoral system choice Interest-based explanations Non-interest-based explanations Self-interests General interests History Seat-maximization Fair representation Socio-political and economic context Policy maximization Government effectiveness Randomness/ accidents Career opportunities Accountability Diffusion Cartel preservation Prevent conflict Institutional constraints 2. Theoretical framework of analysis... clearly the study of origins and of reciprocal effects between the correlates of electoral-system selection and subsequent effects of the chosen system is a research frontier for the twenty-first century (Shugart 2005, 51). Building on previous research, particularly the study on electoral reforms in Latin America carried out by Karen Remmer and published in Party Politics earlier this year, and following up on Shugart s quest above, this study aims to look into the dynamics between party 6

systems and electoral systems in the European context. The two main theoretical statements that will be scrutinized are: H1: Electoral system affects party system development H2: Party system affects electoral system choice The research question can be formulated as follows: How does change in the electoral rules affect party system development and how does change in the party system affect electoral system choice? Analyses will be based upon the three key variables of electoral system choice, party system fragmentation and electoral volatility, all of which are recognized as central in electoral system research. Following in the footsteps of Duverger and his followers, it is expected that when the electoral system is changed, the party system will change accordingly. Thus, ex-ante shifts towards more proportional systems are anticipated to be followed by party system fragmentation whereas majoritarianizing electoral system reforms are likely to produce reductions in the number of parties. The central logic adheres to the psychological and mechanical effects of the electoral rules. By lowering the barriers for new entry, proportionalizing reforms are likely to trigger new party participation, to lower strategic voting in the electorate and in extension increase party fragmentation in the legislature. Majoritarianzing reforms works the other way around: higher thresholds of representation will keep newcomers away from the election, increase tactical voting and thereby produce fewer parties in the parliament. The first hypothesis can thereby be re-formulated: H1: The probability of party system fragmentation increases with the introduction of proportionalizing electoral system reforms and, conversely, the likelihood of party system consolidation increases with the introduction of majoritarianizing reforms. In accordance with the new agenda within electoral system research, on the other hand, it is expected that ex-ante shifts in the party system will be followed by electoral system change that seeks to reinforce the current trends in the party system development (Remmer 2008). When the party system becomes more fragmented, the electoral system will become more proportional as a consequence of the fact that a) smaller parties, which will benefit from more proportional rules, are increasingly empowered and in position of pushing for such reforms and b) larger parties usually advantaged by more majoritarian (or stronger) electoral systems faces increasing uncertainty about their future fortunes. Conversely, when the 7

number of parties is decreasing, the gains enjoyed by the winning parties are likely to translate into majoritarian reforms to further consolidate their position. Thus, the second hypothesis can be re-written: H2: The likelihood of proportionalizing reform increases when the number of parties represented in the legislature rise and, conversely, the probability of majoritarianizing reforms increase when the number of parties decline. 3. Data and methodology The analysis explores the relationship between party system and electoral system choice in 29 European countries over 60 years covering the period 1945 to 2005. The theoretical perspective studied guides the further case selection: only democratic countries where democratically elected politicians have the chance to make deliberate choices upon which electoral rules to use, are included. Thus, for western European the whole period is covered whilst for the southern and eastern parts of the region, party system change and electoral system choices are analyzed from the second democratic election (1970s and 1990s, respectively). The unit of analysis is legislative periods during which electoral system choices are made (before an election) and during which the electoral system proves to have had an effect (after an election). The two primary variables are i) electoral system choice, ii) party system fragmentation. Electoral system choice is operationalized according to persistence versus change which is subsequently operationalized into majoritarianizing versus proportionalizing electoral reforms thus comprising three categories (persistence, proportional reforms and majoritarianizing reforms). Persistence or change as well as the direction of reforms are based upon three variables well-known in conventional electoral system research: assembly size, district magnitude and electoral formula with one additional element which has become increasingly relevant in particular with the introduction of mixed member electoral system but also within the proportional representation family: tiers. A fifth variable threshold will be included at a later stage but is currently missing in the data set. The conventional 20 percent criterion for change is applied for assembly size and district magnitude latter variables. Party system fragmentation is measured according Laakso and Taagepera s effective number of parliamentary parties (ENPP). 8

Both bivariat and multivariate techniques are utilized to explore the relationship between the two core variables: party system and electoral system change. The latter analysis utilize Remmer s set-up and thereby includes, in addition to electoral system reforms, party system change and volatility, the following controls: the number of parties, GDP, democratic age and year (see Remmer 2008 for details). The analysis first explores the bivariat relationships between party system and electoral system change. Second, multivariate techniques are utilized to further analyze the relationship between the two variable so of interest. The two equations can thereby be presented as follows: Party system = election system + lagged ENPP + lagged GDP + democratic age (+ year) Electoral system = lagged ENPP + lagged ENPP + lagged EVS + lagged GDP + democratic age (+ year) Data on electoral system reforms has been based on Golder s dataset on electoral systems in the world for the period up to 2000 3 whilst the last five-year period comes from different sources ranging from journal articles to official websites of electoral management bodies, government websites, electoral laws and the like. Statistics related to ENPP and volatility comes from M. Gallagher s and P. Mair s collections 4 whilst GDP and democratic age come from the Polity IV project. 5 4. Analysis Table 1 below provides a summary over electoral system reforms in democratic Europe in the period 1945-2005. There have been six major changes, defined as reforms introducing large-scale change from one system to another, and 28 minor changes or modifications of the existing system in use. In total, eleven reforms were of the majoritarianizing kind whereas 23 went in the proportionalizing direction. Looking at changes in the lower or only tier more in detail, it has been six changes in assembly size, 19 changes in district magnitude and 22 changes in the electoral formula. Finally, there are six modifications of the number of tiers. Table 1 summarizes the type of reforms that were undertaken. 3 http://homepages.nyu.edu/~mrg217/elections.html 4 http://www.tcd.ie/political_science/staff/michael_gallagher/elsystems/ 5 http://www.systemicpeace.org/polity/polity4.htm 9

Table 1: Electoral system reform in democratic Europe 1945-2005. Country Year Reform Direction Magnitude Austria 1971 L: increased district magnitude and formula shift from Droop + Min. to Hare H: decreased number of districts (for rem. seat allocation) 1994 Increased number of tiers from 2 to 3 + Min. L 1: lower district magnitude H 1: resembles previous L (rem. seat allocation) H 2: increased district magnitude (number of districts from 2 to 1) Belgium 1995 Decreased assembly size (+) Min. 2003 Increased district magnitude and abolishing upper tier + Min. Denmark 1953 L: Increased district magnitude and formula shift from d Hondt + Min. to modified St. Laguë 1971 L: Increased district magnitude + Min. France 1951 From PR to PR-conditional - Min. 1958 From PR conditional to TRS - Maj. 1986 From TRS to PR + Maj. 1988 From PR to TRS - Maj. Germany 1953 Increased assembly size and increased upper tier seat share + Min. H: increased district magnitude 1957 H: Increased district magnitude + Min. 1990 H: Formula shift from d Hondt to Hare, size etc + Min. Greece............ Iceland 1959 From MM to PR + Maj. 1987 L: Formula shift from d Hondt to Hare + Min. 2003 L: Increased district magnitude and formula shift from Hare to (+) Min. d Hondt Italy 1953 Reinforced PR and formula shift from reinforced Imperiali to - Min. Imperial 1958 None-reinforced PR + Min. 1994 From PR to MMP - Maj. Malta 1987 Introduction of the majority-vote guarantee + Min. Netherlands 1956 Increased assembly size + Min. Norway 1953 Increased district magnitude and formula shift form d Hondt to + Min. modified St. Laguë 1989 Introduction of an upper tier + Min. 2005 Increased upper tier seats + Min. Sweden 1952 Formula shift from d Hondt to modified St. Laguë + Min. 1970 Increased assembly size and introduction of an upper tier + Min. L: increased district magnitude Bulgaria 1991 From MMM to PR + decreased assembly size + Maj. Czech Rep. 2002 Decreased district magnitude and from Droop to d Hondt - Min. Lithuania 2000 Formula shift from TRS to plurality in first tier - Min. 2004 Formula shift from plurality to TRS in first tier + Min. Poland 1993 L: Decreased district magnitude and formula shift from Hare to - Min. d Hondt H: Formula shift from modified St. Laguë to d Hondt 2001 Increased district magnitude in lowertier, formula shift from d Hondt to modified St.Laguë and upper tier abolished + Min. 2005 Formula shift from St. Laguë to d Hondt - Min. Slovak Rep. 1992 Formula shift from Hare to Droop - Min. 1998 Increased district magnitude in lower tier and abolishing of upper tier (+) Min. Note: Change direction in brackets shows the preliminary coding of these reforms which have elements of both proportionalizing and majoritarianizing reforms. 10

Looking into the first hypotheses, treating the party system as the dependent variable, it is expected that the number of parties will increase after a proportionalizing reform and decrease after a majoritarianizing reform, whilst the changes will be insignificant when the status quo system is kept in place. The mean comparison analysis lend support to the traditional Duvergian thinking: in the first elections after a majoritarianizing change, the number of parties was reduced by on average 0.67 whereas the number increased slightly by 0.13 for proportionalizing reforms with the mean party system change is smallest where no electoral system reform was introduced Table 2). Notably, mean differences are modest and not statistically significant. The means associated with the second hypothesis and new agenda theoretical framework suggests that minor changes in party system fragmentation is associated with electoral system persistence with the mean being only slightly higher prior to the proportionalizing reforms (Table 3). The most surprising results is found in the group of cases that introduced majoritarianizing reforms which on average had experienced an increase of 0.67 parties in the last election before the reform was introduced. Table 2: he effect of electoral system choice on party system change ENPP Table 3: The effect of party system change on electoral system choice ENPP lagged Mean St.d (N) Mean St.d (N) Majoritarianizing reform -0.665 2.45 (N=11) Majoritarianizing reform 0.676 0.777 (N=7) Electoral system persistence 0.044 0.84 (N=271) Electoral system persistence 0.015 0.965 (N=250) Proportionalizing reform 0.129 0.845 (N=23) Proportionalizing reform 0.092 0.655 (N=19) Looking more detailed into the interaction between the number of parties and party system change on the one hand and electoral system choice on the other, the results shows significant ambiguity. Figure 2a: Party system and electoral system choice in Scandinavia, 1945-2005. 11

With the exception of Finland, the Scandinavian countries have introduced a substantial number of reforms, all of the proportionalizing kind (marked with red circles by the parliament that introduced the new laws) (Figure 2a). In three out of ten cases of electoral system change, proportionalization was against the expectations followed by declining number of parties in the parliament. Thus, the expectations were met six of the cases, with one reform rendering no change. Looking into the matter of the reversed relationship, party system fragmentation preceded proportionalizing reforms in three whereas the opposite is true for four cases. Taking the overall picture into account, the modest changes in the party system and electoral system persistence in Finland seem to have reinforced each other thus creating conditions for institutional equilibrium. Also in Austria and Germany, the two reforms undertaken in each country have been of the proportionalizing nature (Figure 2b). In Austria, the reform adopted in 1970 was neither preceded nor succeeded by party system change thus rendering support to none of the hypotheses outlined. The second reform, however, goes hand in hand with both theories as it was introduced by a parliament that had experienced substantial fragmentation and was followed by additional fragmentation. As far as Germany is concerned, the initial proportionalizing reform resulted in, contrary to the expectations, a substantial decrease in the number of parties. The relationship between party and electoral system as regards the second reform is harder to interpret since this indicates the electoral system put in place for the first election after Germany once again became a unified state. Notwithstanding substantial fluctuation in the party system configurations since the mid-1980s, no reforms have been ensued. Figure 2b: Party system and electoral system choice in the Austria, Germany and Switzerland, 1945-2005. Figure 2c: Party system and electoral system choice in the Benelux countries, 1945-2005. 12

In the Benelux countries, the numbers of reforms are fewer (Figure 2c). The Netherlands increased the assembly size from 100 to 150 in the beginning of the 1950, but the reform was undertaken by a parliament that had not expected any changes as far as party system change goes. Moreover, the reform did not lead to any increase but rather a slight decrease in the number of parties in the subsequent election. In Belgium, two reforms were undertaken in the 1990s. Against all expectations, as fragmentation was increasing and reaching top levels, the parliament adopted a majoritarianizing reform (blue circle). After having been in effect in two elections, the system was once again revisited and proportionalizing measures were introduced with increased district magnitudes. This change goes hand in hand with the theoretical expectations as regards party system effects on electoral system whilst countering that of electoral system influence on party system development. Luxembourg has the most modest party system changes of the three countries thus supporting the equilibrium thesis about the relationship between party and electoral system. Figure 2d: Party system and electoral system choice in the United Kingdom, Ireland and Malta, 1945-2005. The established two-party system in the UK has for long been taken to account for the lack of any changes in the electoral rules, notwithstanding several attempts over the years (Figure 2d). As the party system is moving in waves towards two-and-a-half parties, the issue of proportionalizing reforms is likely to arise on the agenda also in the future. Looking at the more truly stable two-party system of Malta, it is rather surprising to find a proportionalizing reform which goes again both the first hypothesis regarding party system as a product of electoral system choice and the second hypothesis regarding the electoral system choice as a product of party system developments. Notably, 13

this reform was particular in its kind, only introducing a majority guarantee for parties that are able to secure a majority of the votes in the backdrop of the experience where a party twice had secured most seats in the parliament notwithstanding less votes then the runnerup party. Ireland s STV system has remained without change throughout the period despite fluctuating more, by far, as compared to its neighbour and its STV companion. Notably, fluctuations remains within the one-party range (between 2.5 and 3.5 parties). France and Italy are known for being among the most electoral reform friendly countries of the western European countries (Figure 2e), particularly so because they have tended to introduce major changes to the systems Italy 1992 and France 1958, 1986 and 1988. Majoritarianizing reforms have on three out of five cases increased fragmentations in parliament thus undermining the empirical evidence of Duverger. At the other side of the coin, they have not either been preceded by any significant reductions. Quite on the contrary, the two reforms of 1986 (return from PR to TRS) and 1992 (from PR to MM in Italy) were both introduced by parliaments that have immediate experience of increased Figure 2e: Party system and electoral system choice in France and Italy, 1945-2005. fragmentation. Looking closer at the 1986 reform in France, whereby it has been claimed that the introduction of PR was solely a strategic choice by the Socialist party facing severe electoral losses, visualize the difficulty of basing self-interests of political parties on a priori changes rather than subsequent electoral expectations and underlines the need for thicker analyses of the relationship between party and electoral systems not being captured by large-scale statistical analysis. Figure 2f: Party system and electoral system choice in Greece, Portugal and Spain, 1970s-2005. 14

Looking at the southern European countries establishing democratic rules finally in the 1970s, a relatively stable party system Greece after the first election, Portugal from the mid-1980s and Spain throughout the period may have justified the persistence of the electoral rules in place (Figure 2f). The major fluctuations in Portugal in the earlier period, however, remain unexplained. According to Geddes (1996), institutions are in flux during periods of democratic change but this idea seems to be refuted as far as the electoral system is concerned in the southern Europe. Now moving to the most recent democratic countries in the Central and Eastern Europe, Geddes suggestion seems to become true: the region has experienced an extraordinary amount of party and electoral system changes (Figure 2g). The four Visegrad countries altogether six reforms have been undertaken in the short 15 year period: three in Figure 2g: Party system and electoral system choice in Hungary, Poland and the Czech and Slovak Republics, 1990-2005. Poland, two in the Slovak and one in the Czech Republic with Hungary being the only country without any reforms. Starting with the latter, the Hungary is also the one of the four with the most stable party system thus supporting both hypotheses: no change in party system renders no change in the electoral system and conversely no change in electoral system produce no change in the party configurations. The Czech majoritarian reform was introduced on the backdrop of declining party system fragmentation but did not produce any significant change in the number of parties the first time it was put into use in 2002 thus lending support to H2. The two cases of electoral reform in the Slovak Republic support both hypotheses: the majoritarianizing reform undertaken by the first parliament led to a reduction in the number of parties whereas the second proportionalizing reform was adopted by a parliament experiencing increased fragmentation and was also succeeded by 15

further fragmentation. Finally, two out of three reforms supports the Duvergian line of thinking: the majoritarian change reduced the number of parties dramatically in the 1993 election, and the proportional reform in the mid-1990s was succeeded by slight fragmentation. Looking at the effect of party system change on electoral reform, the Polish reforms fails to be explained by the self-interests of the parties in parliament. Figure 2h: Party system and electoral system choice in Estonia, Latvia and Lithuania, 1990-2005. Figure 2i: Party system and electoral system choice in Bulgaria, Romania and Slovenia, 1990-2005. Reforms have also been undertaken in Bulgaria and Lithuania (Figures2h and 2i). In Bulgaria, the mixed system used for the first election was replaced by full-blown PR. Notwithstanding the change, the number of parties remained low, below two-and-a-half and three for the subsequent decade. Before the 2001 election in Lithuania, the TRS formula was replaced by the plurality rule in the first tier of the mixed system before being reversed altogether prior to the subsequent election (2004). The first reform supports neither of the hypotheses: it was neither preceded nor succeeded by declining fragmentation. The second reform supports both by being introduced against the backdrop of a slight increased in the number of parties and possibly contributing to furthering the tendency. Notwithstanding substantial party system change, the founding electoral system has remained largely in place in Estonia, Latvia, Romania and Slovenia. Looking at the regression analysis results in Table 3, neither proportionalizing nor majoritarianizing reforms have a significant impact upon legislative fragmentation. Taking into consideration that the analysis is carried out on the whole population of electoral system 16

choice in the period of interest, it can be pointed out that proportionalizing reforms seem, in accordance with Duverger s hypothesis, to result in increased fragmentation in the party system that follows the first election after reform has been introduced. The effect of majoritarianizing reforms is not only substantially smaller, but the direction is also reversed as compared to the expectations. The lagged number of parties which indicates the level of fragmentation prior to the reform is the one and only variable in the model that is Table 3: Impact of electoral system reforms on party system change Table 4: Impact of party system change on electoral system reform Majoritarianizing reform Proportionalizing reform Proportionalizing reform 0.229 (0.205) ENPP (lagged) 0.423 (0.389) 0.194 (0.201) Majoritarianizing reform 0.018 (0.457) ENPP (lagged) 0.305 (0.343) 0.013 (0.220) ENPP (lagged) -0.133 (0.040)*** GDP (lagged)(prc) 0.001 (0.001) Democratic age 0.000 (0.001) Year 0.002 (0.003) Constant -2.743 (5.398) Volatility (lagged) 0.070 (0.045) GDP (lagged)(prc) -0.013 (0.016) Democratic age -0.021 (0.019) Year 0.017 (0.048) Constant -38.646 (94.198) 0.014 (0.031) -0.002 (0.006) -0.018 (0.008)** 0.007 (0.019) -16.184 37.755 Observations 271 R 2 0.073 Observations 267 R 2 0.102 Note: Unstandardized coefficients with robust errors in parentheses. * Significant at 0.05 level, ** significant at 0.01 level, *** significant at the 0.001 level. For the multinominal regression analysis (Table 4), base category is electoral system persistence. statistically significant with the remaining control variables having no impact on party system change. 17

Finally, neither party system change nor volatility has significant impact upon electoral system change when controlling for the lagged number of parties and the contextual controls included (Table4). The coefficients of the variables of primary interest are all positive. As expected, volatility is associated with reforms of either kind. Contrary to the expectations, however, increased fragmentation seems to attract changes in both proportionalizing and majoritarianizing reforms. 5. Conclusion In her study of the relationship between party systems and electoral systems in Latin America, Remmer concludes that Reductions in party system fragmentation have led to the adoption of less permissive rules, indicating that as the effective number of parties decline, winners attempt to consolidate their gains. Although increased fragmentation does not yield symmetrical changes, the impact of electoral volatility on legislative electoral rules is consistent with theoretical expectations. The greater the level of volatility, the higher the probability of both strong and permissive electoral reform At the same time the analysis offer surprisingly little evidence that electoral reform alter party systems. Looking at the causes and consequences of electoral system reform in Latin America over the past two decades, the findings indicate that party system change generates institutional change more predictably than vice versa (Remmer 2008: 24) In Europe, however, the findings are less consistent. First, electoral system changes are not associated with any significant changes in the party structure. These results are thereby in harmony with Remmer s findings. Notably, and also as noted out by Remmer (2008: 22), when using regression techniques the changes are modelled to take effect in the first election after an electoral reform has been adopted which may be a false assumption. A later point in time could have been used thus to take into account that it takes time for both parties and voters to adapt to the new rules. Second, party system change is not significantly associated with electoral system change. Moreover, analysis shows that party system fragmentation precedes not only proportionalizing but also majoritarianizing reforms thus standing in sharp contrast to Remmer s findings and the theoretical expectations of seat-maximizing political parties and party systems. A crucial difference between the study on Europe and Latin America is the historical contexts. In this study, both old and new democracies have been included: western 18

Europe from the post-ww2 period, southern Europe from the 1970s and central and eastern Europe from the 1990s. Running separate analysis or introducing interaction terms identifying these different historical circumstances might alter the results. It can be hypothesized, for example, that in new democracies, where stakes are higher and where public awareness of democratic rules are lower, manipulation with the rules of the game in order to enhance one s own chances are more attractive and easy than in older democracies. List of References Ballington, Julie and Azza Karam. 2005. Women in Parliament: Beyond Numbers. International IDEA: Stockholm. Bawn, Kathleen. 1993. The Logic of Institutional Preferences: German Electoral Law as a Social Choice Outcome. American Journal of Political Science 37 (4):965-989. Benoit, Kenneth. 2004. Models of Electoral System Change. Electoral Studies 23:363-389. Benoit, Kenneth. 2007. Electoral Laws as Political Consequences: Explaining the Origins and Change of Electoral Institutions. Annual Review of Political Science 10:363-390. Benoit, Kenneth, and Jacqueline Hayden. 2001. Institutional Change and Persistence: The Evolution of Poland s Electoral System, 1989 2001. The Journal of Politics 66 (2):396-427. Benoit, Kenneth, and John W. Schiemann. 2001. Institutional Choice in New Democracies: Bargaining Over Hungary s 1989 Electoral Law. Journal of Theoretical Politics 13 (2):153-182. Birch, Sarah. 2003. Electoral Systems and Political Transformation in Post-Communist Europe. Basingstoke: Palgrave Macmillan. Birch, Sarah. 2005. Lessons from Eastern Europe: Electoral Reform Following the Collapse of Communism. Paper presented at the conference on Electoral Reform in Canada: Getting Past Debates about Electoral Systems at the Mount Allison University, Sackville, New Brunswick 10-12 May. Birch, Sarah, Frances Millard, Marina Popescu, and Kiaran Williams. 2002. Embodying Democracy: Electoral System Design in Post- Communist Europe. Basingstoke: Macmillan. Blais, André. 2000. To Vote or Not To Vote? The Merits and Limits of Rational Choice. Pittsburgh: University of Pittburgh Press. Blais, André, and R. K. Carty. 1991. The Psychological Impact of Electoral Laws: Measuring Duverger's Elusive Factor. British Journal of Political Science 21 (1):79-93. Blais, André, and Louis Massicotte. 1997. Electoral Formulas: A Macroscopic Perspective. European Journal of Political Research 32:107-129. Bogdanor, Vernon, and David Butler. 1983. Democracy and Elections. Cambridge: Cambridge University Press. Boix, Charles. 1999. Setting the Rules of the Game: The Choice of Electoral System in Advanced Democracies. American Political Science Review 93 (3):609-624. Brady, David, and Jongryn Mo. 1992. Electoral Systems and Institutional Choice: A Case Study of the 1988 Korean elections. Comparative Political Studies 24 (4):405-429. Colomer, Joseph M. 2004b. The Strategy and History of Electoral System Choice. In Handbook of Electoral System Choice, edited by J. M. Colomer. Basingstoke: Palgrave Macmillan. Colomer, Joseph M. 2005. It's Parties That Choose Electoral Systems (or, Duverger's Law Upside Down). Political Studies 53:1-21. Colomer, Joseph M. (ed.). 2004d. Handbook of Electoral System Choice. Basingstoke: Palgrave Macmillan. 19

Croissant, Aurel, Gabriele Bruns, and Marei John. 2002. Electoral Politics in Southeast and East Asia. Singapore: South East Printing Pte Ltd. Denemark, David. 2001. Choosing MMP in New Zealand: Explaining the 1993 Electoral Reform. In Mixed-Member Electoral Systems: The Best of Both Worlds?, edited by M. S. Shugart and M. P. Wattenberg. Oxford: Oxford University Press. Dhillon, Amrita, and Susana Peralta. 2002. Economic Theories of Voter Turnout. The Economic Journal 112:F332- F352. Díaz- Cayeros, Alberto, and Beatriz Magaloni. 2001. Party Dominance and the Logic of Electoral Design in Mexico s Transition to Democracy. Journal of Theoretical Politics 13 (3):271-293. Donovan, Todd. 1995. The Politics of Electoral Reform in Italy. International Political Science Review 16 (1):47-64. Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper. Duverger, Maurice. 1986. Duverger's Law: Forty Years Later. In Electoral Laws and Their Political Consequences, edited by B. Grofman and A. Lijphart. New York: Agathon Press. Farrell, David M. 2001. Electoral Systems: A Comparative Introduction. Basingstoke: Palgrave. Fisher, Stephen D. 2004. Definition and Measurement of Tactical Voting: The Role of Rational Choice. British Journal of Political Science 34:152-166. Gallagher, Michael. 2005. Conclusion. In The Politics of Electoral Systems, edited by M. Gallagher and P. Mitchell. Oxford: Oxford University Press. Gallagher, Michael, and Paul Mitchell (eds). 2005. The Politics of Electoral Systems. Oxford: Oxford University Press. Grumm, John G. 1958. Theories of Electoral Systems. Midwest Journal of Political Science 2 (4):357-376. Hooghe, Marc, Bart Maddens, and Jo Noppe. 2006. Why Parties Adapt: Electoral Reform, Party Finance and Party Strategy in Belgium. Electoral Studies 25 (2):351-368. Horowitz, Donald L. 1991. A Democratic South Africa? Constitutional Engineering in a Divided Society. Berkley, Calif.: University of California Press. Jones, Mark P. 1993. The Political Consequences of Electoral Laws in Latin America and the Caribbean. Electoral Studies 12 (1):59-75. Kaminski, Marek. 2002. Do Parties Benefit from Electoral Manipulation? Electoral Laws and Heresthetics in Poland, 1989-1993. Journal of Theoretical Politics 14 (3):325-359. Katz, Richard. 1980. A Theory of Parties and Electoral Systems. Baltimore: John Hopkins University Press. Katz, Richard. 2005. Why Are There So Many (or So Few) Electoral Reforms? In The Politics of Electoral Systems, edited by M. Gallagher and P. Mitchell. Oxford: Oxford University Press. Lijphart, Arend. 1991a. Constitutional Choices for New Democracies. Journal of Democracy 2 (1):72-84. Lijphart, Arend. 1991b. Double- Checking the Evidence. Journal of Democracy 2 (3):42-48. Lijphart, Arend. 1994. Electoral Systems and Party Systems: A Study of Twenty-Seven Democracies 1945-1990. Oxford: Oxford University Press. Lindberg, Staffan I. 2005. Consequences of Electoral Systems in Africa: A Preliminary Inquiry. Electoral Studies 24 (1):41-64. Laakso, Markku, and Rein Taagepera. 1979. Effective" number of parties: a measure with application to West Europe. Comparative Political Studies 12 (1):3-27. Matland, Richard E., and Michelle E. Tylor. 1997. Electoral System Effects on Women's Representation. Comparative Political Studies 30 (2):186-210. Moser, Robert G. 1999. Electoral Systems and the Number of Parties in Postcommunist States. World Politics 51 (3):359-384. Nagel, Jack H. 1994. What Political Scientists Can Learn from the 1993 Electoral Reform in New Zealand. PS: Politican Science and Politics 27 (3):525-529. 20

Navarra, Pietro, and Giuseppe Sobbrio. 2001. Election Re- Running and the Nature of Constitutional Choices: Case of Italian Electoral Reform. Constitutional Political Economy 12 (1):31-50. Nohlen, Dieter. 1984. Changes and Choices in Electoral Systems. In Choosing an Electoral System: Issues and Alternatives, edited by A. Lijphart and B. Grofman. New York: Praeger. Rae, Douglas. 1976. The Political Consequences of Electoral Laws. New Heaven: Yale University Press. Remington, Thomas, and Steven S. Smith. 1996. Political Goals, Institutional Context and the Choice of an Electoral System: The Russian Parliamentary Election Law. American Journal of Political Science 40 (4):1253-1279. Remmer, Karen. 2008. The Politics of Institutional Change: Electoral Reform in Latin America, 1978-2002. Party Politics 14(1): 5-30. Reilly, Ben. 2002. Electoral Systems in Divided Societies: Electoral Engineering for Conflict Management. Cambridge: Cambridge University Press. Rokkan, Stein. 1970. Citizens, Elections, Parties. Oslo: Universitetsforlaget. Sartori, Giovanni. 1997. Comparative Constitutional Engineering: An Inquiry into Structures, Incentives and Outcomes. Basingstoke: Macmillan. Shugart, Matthew Søberg, and Martin P. Wattenberg (eds.). 2001. Mixed-Member Electoral Systems: The Best of Both Worlds? Oxford: Oxford University Press. Taagepera, Rein, and Matthew Shugart. 1989. Seats and Votes: The Effects and Determinants of Electoral Systems. New Haven: Yale University Press. Vowles, Jack. 1995. The Politics of Electoral Reform in New Zealand. International Political Science Review 16 (1):95-115. Wildavsky, Around B. 1959. A Methodological Critique of Duverger s Political Parties. The Journal of Politics 21 (2):303-318. Zimmerman, Joseph F., and Wilma Rule. 1994. Electoral Systems in Comparative Perspective: Their Impact on Women and Minorities. Westport, Connecticut: Greenwood Press. 21