Electoral Systems and Party System Stability in Post-Communist Europe

Sarah Birch Department of Government University of Essex Wivenhoe Park Colchester CO4 3SQ United Kingdom [email protected]

Paper prepared for presentation at the 97th annual meeting of the American Political Science Association, San Francisco, 30 August 2 September, 2001.

ABSTRACT: This paper analyses the impact of electoral institutions on party system stability and change in twelve countries of post-communist Central Europe and the former Soviet Union. It argues that in the post-communist context it is necessary to break party system change down into two components: volatility and replacement. Volatility, a concept already well developed in the literature, refers to changes over successive elections in the balance of party support. Replacement is understood as the extent to which new political formations are successful in drawing support. Empirical analysis of these two variables demonstrates that they exhibit different patterns and that they are influenced by separate sets of factors. Electoral institution variables are found to have a strong impact on volatility by not replacement, which is influenced instead by sub-regional location and rates of electoral participation.

Electoral Systems and Party System Stability in Post-Communist Europe Stability and change in party systems has long fascinated political scientists, but the nature of the research questions they have asked has been conditioned by the cases they have taken as their objects of study. When we turn our attention to areas of the world that have not been subject to such intense scholarly investigation as the established democracies of the West, we may need to re-think the conceptual tools we use in interpreting and explaining changes in the configuration of electoral contenders. The specific aim of this paper is to formulate new ways of understanding party system change and to assess how electoral institutions have affected recent party system developments in post-communist Europe.1 The focus of the analysis will be on two distinct aspects of party system dynamics: volatility and replacement. Volatility, a concept already well developed in the literature, refers to changes over successive elections in the balance of party support. But volatility understood in this sense fails adequately to capture the numerous party foundations, splits, mergers, and name changes characteristic of post-communist politics. In this context it is also desirable to deploy the complementary concept of party replacement the extent to which new political formations are successful in drawing electoral support.2 Most scholars agree that the development of stable, coherent representative parties which can shape and channel popular preferences is crucial to successful democratisation in the wake of political transition. In the early days of democratisation in Central and Eastern Europe, politicians and citizens alike were often sceptical of the efficacy of parties in contributing to representation, given the experience of one-party rule. But in all cases the need for parties soon became apparent to a large majority of the political class, and subsequent efforts by politicians have been directed at attracting votes to their fledgling political organisations. At the same time, parties have begun to establish working relations and patterns of alliance within national legislatures. These dual processes have been influenced by numerous contextual factors, including the pre-communist experience, the legacy of the old regime, the pace of socioeconomic change, institutional choices, and the attributes of specific countries.3 Mobility in the party system has both advantages and disadvantages. As Bartolini and Mair (1990) argue, variations in electoral support are the main basis on which party leaders plan Research for this paper was supported by a generous grant from the Economic and Social Research Council (No. L213252021). I would like to thank Marina Popescu and Robertas Pogorelis for their research assistance and for sharing with me their expertise on Romania and Lithuania respectively. Thanks are also due to Frances Millard for help interpreting the Polish party system. I take responsibility for any errors of fact or judgement in this paper. 1 In the absence of clear criteria for determining the systematicity of party configurations, the term party system will be used in this paper to refer to the sum total of parties that are electorally active in a given country at a given time. 2 Similar distinctions have been made or implied by several authors, including Mair, 1997; Toka,1997; Kitschelt et al, 1999; Colton, 2000; and Markowski, 2001, though the concepts are measured differently in the present analysis (see below). 3 The literature on these processes is abundant, but special mention should be made of several comparative volumes, including Pridham and Lewis, 1996; Olson and Norton, 1996; Kitschelt et al, 1999; Lewis, 2000; and 2001.

strategies. Likewise, rises and falls in partisan support may be taken to indicate that the party system is responsive to the changing preferences of voters (Budge, 1982). In the context of rapid socio-economic change such as that experienced during the period of post-communist economic restructuring, such adaptability may be a good thing. But turbulence in the party system cannot necessarily be attributed to variations in electoral preferences alone. On the contrary, there is much research indicating that the basic political proclivities of post-communist electorates have remained relatively stable over the past ten years and that most of the change in post-communist party systems has been driven instead by lite-level fluidity (e.g. Markowski, 1997; Miller et al, 1998; Birch, 2000; King, 2000; Rose et al, 2001). A consistently high level of party system instability has four main consequences that are detrimental to democratic consolidation. 1. It reduces accountability voters cannot throw the rascals out if the rascals no longer exist as a unified group. 2. It impedes party institutionalisation by decreasing the level of long-term commitment which politicians, activists, and voters have to their party. 3. It significantly increases uncertainty, hampering the ability of politicians and voters to engage in strategically-driven co-ordination 4. It raises the stakes of the electoral game. This may have the consequence of weakening the democratic commitment of politicians who may seek other ways of feathering their beds so as to insure themselves against possible political loss at the next election. The successful consolidation of democracy requires enough uncertainty to keep losers in the political game, but not so much that elections become a lottery. Despite the potential benefits of flexibility in the party system, excessive instability undermines the basis for political representation. It is likely to be associated with general political and social instability which has the potential to generate an authoritarian backlash.4 The conceptualisation and measurement of party system change is thus key to an adequate understanding of post-communist political transformation. This analysis will necessarily build on valuable previous work that has been conducted on post-communist party system development. Yet it will also challenge some of the generalisations that have been made about party systems in the region on the basis of a limited range of cases. The cluster of states on the Western border of Eastern Europe - Poland, the Czech Republic, and Hungary have, together with Russia, received a disproportionate share of the scholarly attention that has been devoted to the study of post-communist parties, despite the fact that they together constitute only a fraction of the 22 states that make up the region. One of the objectives of this study is to examine a wider range of states in order to evaluate whether conclusions that have been drawn from smaller N studies hold more generally. Unfortunately it has only been possible to include 12 countries in the dataset examined here, for reasons both of data availability and depth of knowledge on my part. But the cases are divided evenly between This may be what we are currently observing in Russia, where president Putin enjoys extremely high levels of support across much of the political spectrum for his alleged success in bringing order to the violence-plagued country.
4

Central Europe and the former Soviet Union, enabling systematic comparisons across two key sub-regions. I Conceptualising and Measuring Party System Instability The concept of electoral volatility was long understood to explain change in electoral preferences over time as a result of secular changes in values and social structure.5 But the way in which the concept is measured confounds mass level variation in preferences and lite-driven variation in the objects of choice on offer. Measured in terms of the sum of change in support for individual parties between one election and the next, electoral volatility is a multi-layered and multi-dimensional concept (Crewe, 1985:8). The overall or aggregate volatility between election t and election t + 1 is the result of a combination of three factors: (1) changes in party support on the part of individual voters who vote at both election t and election t + 1; (2) change in the composition of the electorate as a result of exits, entries, and variations in abstentionism; (3) changes in the range of parties on offer as a result of party entries, exits, alterations in the composition of coalitions, and non-contestation. Most studies of volatility in established democracies have been primarily concerned with voter behaviour, so they have been mainly interested in the first and possibly the second of these factors. They have generally assumed that third the turnover of electoral contenders - is small enough that the standard measures of volatility are a good proxy for changing voter behaviour. In other words, the assumption in much of the literature has been that lite-level change in the supply of parties reflects and is prompted by changes in the demands made by the electorate.6 Whether or not this market model is an accurate reflection of party system development in existing democracies, it is not an apt characterisation of party system change in many democratising states. Whereas in most established democracies there is a high degree of continuity in the range of political options on offer from one election to the next, emergent party systems in post-transition countries are often considerably less stable. Most countries in the Eastern European region have followed a far more command economy style development in which the supply of parties has been based primarily on the whim of lites with little reference to the demands, desires, or preferences of the electorate. Most of the emerging party systems of Central and Eastern Europe have been characterised by fluidity, splits, mergers and weak links with their corresponding electorates. In this context the turnover of parties - or party replacement - is of substantive interest in its own right, especially when it is thought to form a large component of overall party system instability. A high level of churn in the range of parties on offer at election time can be expected to generate voter disorientation and hamper party institutionalisation. Thus when dealing with party systems that experience much inter-electoral change in the identities of parties, it is desirable to disentangle volatility within the existing party system from the emergence of new parties. Not only will this give greater conceptual clarity to the measure of volatility, but it will also make it easier to trace the impact of institutional design on electoral outcomes. Thus Mogens Pedersen defines volatility as the net change within the electoral party system resulting from individual vote transfers (1979:3). 6 Several writers have, however, noted that party system instability can coexist with stable electoral preferences (Crewe, 1985; Mair, 1997).
5

Unfortunately a number of difficulties arise in attempting to develop two separate measures for volatility and replacement. Some of these difficulties are inherent in any measure based on aggregate electoral results alone. The impossibility of ever measuring individual-level volatility on the basis of aggregate results is well known, and though aggregate and individuallevel volatility have been found to co-vary in established Western democracies (Crewe and Denver, 1985; Bartolini and Mair, 1990), we cannot assume that this will be the case in other contexts. Furthermore, a voter who votes for party A at election t may not have had any change in party preference at election t + 1, but s/he will nevertheless be required to change his or her expressed preference and vote for party B (or C, or some other party) if party A is no longer on the ballot. But the voter might equally have switched his or her preference and would have voted for party B even if party A were still available. It is therefore impossible ever to capture the true amount of switching forced or voluntary without recourse to specially-designed comparative surveys (which do not exist for most of the states under consideration here), nor is it ever possible to know how much aggregate change is due to true variations in party preference or changes in the choice of parties on offer. But we can develop approximations of volatility among existing parties and party replacement by examining trends in the electoral support for different types of party. Volatility among existing parties can be estimated by calculating the amount of change observed within the set of parties (or other groupings that present themselves to voters) that contest two consecutive elections. For this purpose we exclude all those parties that did not contest both elections and sum the differences in the vote shares of the remaining parties. To obtain the true volatility score, it is necessary to divide not by 2 as with the Pedersen Index, but by the sum of the fractional shares of the total vote at each election of the parties which are included in the calculations.7 More formally, volatility is defined as: |ci,t+1 ci,t| V = ci,t+1 + ci,t where V is volatility, ci,t is the vote share of continuous party i at the first election (t) and ci,t+1 is the vote share of continuous party i at the second election (t + 1). If volatility thus defined is a measure of changes in the electoral fortunes of existing players in the political game, party replacement can be understood as the degree of penetration of new players into the party system. It can be measured as the proportion of the electorate attracted to new political formations, or the sum of the vote shares won by electoral contenders at election t + 1 that had not contested election t. It is worth noting that though volatility and replacement both vary from 0 to 100, they are not directly comparable, given that they are measured in different ways and from different vote totals. In order to enhance the comparability of the two figures, it is useful to supplement the volatility score as defined above with the conventional score using 2 as a denominator. This

Prior to summing, any votes won by independents are excluded from the total vote and party vote shares are then recalculated on the basis of the revised denominator.

second measure provides an approximation of the proportion of the total electorate that switched votes among existing parties (see Table 3 below). These measures are straightforward in theory, but in practice it is necessary to decide what constitutes the same electoral contender. The formation of entirely new parties presents no problems, nor does the electoral extinction of an existing party. The problems arise when parties split, merge, form or leave coalitions, and/or change their names from one election to the next. As noted above, such changes have been very common in post-communist Europe, so it makes sense to deal with this problem systematically rather than making ad hoc assumptions about the continuity of party identities. From the point of view of the research questions addressed in this paper, the crucial criterion is that of institutional continuity, which involves such factors as continuity of organisational resources from one election to the next as well as aspects of party identity, most importantly the degree to which voters are aware of the changes of identity that their party has undergone and are willing to follow their party under its new guise. Again this is something that we can not know for certain without detailed (and nonexistent) individual-level data. But we can make some educated assumptions as to which parties and coalitions satisfy criteria of institutional continuity and which do not. The details of the criteria used to classify parties and other political groupings into continuous and non-continuous are laid out in Appendix 1, together with assumptions made about data from individual countries. II Explaining Party System Change in Post-Communist Europe The majority of Eastern European states have what Peter Mair describes as open party systems in which patterns of competition are not well established and parties do not have stable support bases (Mair, 1997). A number of scholars have noted the high volatility levels in post-communist Europe in comparison with average rates in Western European countries (Cotta, 1996; Rose et al, 1998:118-19; Olson, 1998; Mair, 1997; Lewis, 2000). In a survey of party system development in Western Europe between 1885 and 1985, Bartolini and Mair find that the average volatility over this period was 9.1 (Bartolini and Mair, 1990:148). The highest scores were 32.1 in Germany between 1919 and 1920, followed by 31.1 in France 1902-6 (1990: Appendix 2). In post-communist Europe the average volatility in the decade following the communist collapse was 18.3 measured in terms of the total vote. Levels have been as high as 43.8, and there have been three cases of scores over 30 (Romania 1990-1992, Moldova 1994-1998, and Lithuania 1996-2000). In only three cases has volatility as a proportion of the total vote been lower than the Western European average the Czech Republic 1990-1992 and 1996-1998, and Slovakia 19901992. There have also been extremely high levels of electoral replacement in some countries. In Moldova not one of the parties elected in 1994 was re-elected in 1998. And in Latvia the two largest parties in 1995 between them polled only 3.35 per cent of the vote four years later (see Tables 3 and 4 below). These patterns indicate that party system change in post-communist Europe has more in common with patterns observed in other democratising regions such as Latin America than they do with their Western neighbours.8 These high levels of instability can be explained by aspects of the post-communist transition process. Firstly, the lack of a usable pre-existing party systems in most countries in the region has meant that politicians have had effectively to form party systems from scratch, at
8

In Latin America electoral volatility has reached levels similar to those found in Eastern Europe (Mainwaring and Scully, 1995:8). See also Coppedge, 1998:559; Roberts and Wibbels, 1999.

the same time civil society has been severely under-developed and voter identification with parties has generally been low (Wyman et al, 1995:546; Miller et al, 1998:170). Secondly, rapid socio-structural change has accompanied economic change, resulting in significant shifts in the objective interests of large sections of the electorate over the course of the past decade. Institutional choices can also be expected to affect levels of electoral instability, though the precise relationship may be complex. There is a long debate as to whether the shape of the party system causes given electoral systems to be adopted or whether those electoral systems cause the party system to take the shape it does (see Cox, 1997 for an overview). Most would agree that there is an element of both in many cases, but it is worth noting that if a given configuration of parties brings about a change in the electoral system, it is precisely because those parties anticipate the effects which electoral institutions will have on future developments in electoral system design. There is therefore no question that electoral systems do have effects though they are not always those intended by their authors; the question is the degree to which the causal relationship is reciprocal (Cox, 1997; Moser, 2001; Birch, forthcoming). When we examine the impact of electoral institutions on variations in votes won, we are concerned with what Duverger termed the psychological effect of electoral systems as opposed to the mechanical variations induced by the aggregation of votes into seats under different conversion formulae (Duverger, 1959). Electoral systems have implicit in them psychological or strategic incentives both for voters and politicians which are linked to the likelihood of success of electoral options under different rules. If voters want to avoid wasting their votes, they will opt for parties or coalitions that have a reasonable chance of winning seat. Office-seeking politicians, for their part, have an incentive to group together into electoral formations that have the best chance of promoting them to positions of power. Duvergers main thesis which has been validated empirically on numerous occasions since was that a high threshold of success would generate consolidation in the party system, as only relatively large parties would be electorally viable. Majority or plurality systems in single-member constituencies generally entail a high threshold of success, because only the top contender in each constituency wins the one and only seat. Likewise, high formal thresholds at constituency or, more commonly, national level impose restrictions which should also decrease the vote share of small parties. We would expect to see volatility dampened under such systems because voters will be wary of opting for electoral alternatives that have little chance of winning and with fewer potential objects of choice, there ought to be less switching.9 There are also reasons for believing that electoral systems should influence volatility indirectly due to the effect they have on party system size The number of parties that compete for elections has been found to be positively associated with electoral volatility (Pedersen, 1979; Crewe, 1985; Bartolini and Mair, 1990), and it is well known that the electoral system is one of the main determinants of the number of parties that compete in elections (e.g. Rae, 1971; Taagepera and Shugart, 1989; Cox, 1997). Fewer parties in the system mean fewer alternatives that might appeal to voters at given points in the ideological spectrum, whereas a greater number of parties means more parties that are proximal to voters in ideological space, so more options that each individual voter might be attracted to (Pedersen, 1979; Crewe, 1985). We would thus
9

Bartolini and Mair (1990) find that the opposite has been true throughout large portions of the history of Western European party competition, a result they explain with reference to the magnification in vote switching which results from tactical voting under systems with small constituencies and high thresholds. But the differences they find are small enough and inconsistent enough over time to warrant caution in interpreting them.

expect the electoral system to influence volatility indirectly via the number of parties it encourages to compete; restrictive systems with small constituencies or high formal thresholds as the national level can be expected to discourage parties from competing and thereby reduce volatility. We should also anticipate party volatility in response to changes in the electoral system, which will either open up new avenues of possible success or restrict the chances for victory; in their study of Western European party systems, Bartolini and Mair (1990) find that electoral system change is one of the most powerful determinants of volatility, and there is little reason to believe that it should not have an equally great effect on the post-communist context. It is also worth considering the role of presidentialism in affecting the outcome of parliamentary elections. If parliamentary office is a lesser goal for politicians than the presidency which it will surely be in any system with a directly-elected executive presidency then popular politicians may use political parties primarily as vehicles to launch a bid for the presidency. Previous work has found that two-round presidential electoral systems which are universal among the presidential democracies of Central and Eastern Europe have the effect of fragmenting the party system by encouraging popular politicians to form separate political organisations in their efforts to jockey for a place in the second round (Shugart and Carey, 1992; Jones, 1995). A final electorally-related factor which may be related to volatility is electoral participation, which Bartolini and Mair (1990) find to be positively associated with volatility in Western Europe. Previous research gives us less material with which to make predictions about what influences party replacement. But we would also expect the electoral systems to have direct effects on replacement under conditions of high uncertainty such as obtain in post-communist countries. Supporters of major parties (which will in most cases be in a majority) have an incentive to stick with parties whose seat-winning ability has been proven rather than gravitating to new or small political organisations, and the number of trustworthy parties will be smaller in systems with high electoral formal or natural thresholds. With high thresholds we would anticipate a more pronounced learning curve as parties come to appreciate their electoral strength (or lack thereof) and react accordingly by merging with other parties or dropping out of the game. The electoral prize of the presidency may also provide an incentive for hopefuls to form new parties to increase their chance of success in the presidential race. Finally, electoral participation may be of relevance: in as much as new parties succeed in mobilising previously inactive sectors of the electorate, we would expect to find an association between high turnout and high levels of party replacement. III Analysis Twelve countries are included in the present study: the six Central European states of Bulgaria, the Czech Republic, Hungary, Poland, Romania, and Slovakia, and the six former Soviet states of Estonia, Latvia, Lithuania, Moldova, Russia, and Ukraine. The dataset on which the analysis is based comprises vote totals from elections to lower or only chambers of national parliaments. The transitional elections held in Poland in 1989 and in the Soviet republics in 1990 are not included as they were not fully competitive and/or not conducted on a clear multi-party basis.10
10

The dataset is also restricted to national-level results, which is a further reason for excluding the 1990 Soviet republican elections. Though the Baltic states had by this time all made moves

The 1990 and 1992 Czechoslovak elections to the Chamber of the People in the Federal Assembly are divided into Czech and Slovak components and treated as separate elections (a move justified by the fact that the two parts of the Federation had distinct party systems with very little overlap in the identities of contenders). Finally, the Bulgarian Constituent Assembly elections of 1990 are included, given that the resultant assembly functioned as a parliament. Elections during the communist period were in most cases conducted in single-member constituencies according to an absolute majority rule. With the fall of communism, virtually all the states in the region changed either to proportional representation or mixed systems with a strong PR element. Ukraine was the only case among the 12 states analysed here to retain the single-member majority system into the post-communist period, before finally shedding it in favour of a mixed system for the 1994 elections (see Table 1). Mixed systems in the region are mostly of the parallel sort (as in Japan) rather than the compensatory type (as in Germany). The exception is the complex Hungarian system which is partially compensatory. Another relevant feature of electoral system design is the formal threshold of exclusion found in most PR systems. Thresholds tend to be relatively high in post-communist Europe; the average national threshold for single parties is 4.33 in this dataset of 39 PR elections. This most likely reflects a perceived need to consolidate the large party systems around a core of political groups (Birch, 2001). In theory it would be desirable to combine threshold and constituency size into a consolidated measure of the effective threshold (Taagepera and Shugart, 1989; Lijphart, 1994; Taagepera, 1998) rather than two separate variables for the formal threshold and the natural threshold induced by small constituency size. Unfortunately the latter is too difficult to estimate in the Eastern European context because the large numbers of candidates that contest elections. This, combined with regional limitation in the electoral reach of many parties means that elections in single-member constituencies can be won on miniscule percentages of the national vote. For example, the Common Cause movement in Russia won a seat in 1995 on the basis of only 0.24 per cent of the national vote, and this was not an isolated case; nine of the 21 parties that won single-member seats in this election won them on less than one per cent of the vote. Even when there is a formal threshold at the constituency level as in absolute majority elections - the lack of results from all constituencies in second-round races (due to first-rounds wins) makes it necessary to base the statistical analysis on first-round vote data to which the threshold does not apply. Under these circumstances is makes sense to consider constituency format and formal thresholds as separate variables (though, as we will see, they have similar effects). - Table 1 about here By mid-2001 almost all the countries in the dataset had held at least three elections and the only laggard, Ukraine, is due to hold its third post-Soviet parliamentary election in March 2002. Three countries the Czech republic, Romania, and Slovakia have held four elections, and Bulgaria five. The first elections in the Central European countries were all held in 1990 with the exception of Poland, and they were in most cases dominated by anti-communist front parties or movements. The elections themselves were used as a means of validating the rejection of communism. In the former Soviet countries, by contrast, first elections were held somewhat toward reasserting their sovereignty, they were still effectively under Soviet occupation at the time of these elections.

later after independence and the concomitant dismantling of the communist system had been achieved by other means. This meant that popular front organisations were generally less prominent in the first electoral events. But regardless of their role in the trajectories of transition, these founding elections gave most parties their first clear indication of the extent of the popular support they could command. The lessons learned from these inaugural polls were then used to craft new party strategies. The first feature of post-communist electoral competition that even the most casual observer will note is the large number of options available to voters. An average of 28.2 parties or coalitions have contested the 40 elections from 12 countries included in this dataset (see Table 2). This translates into a total of 454 partisan contenders in the first post-transition elections and 472 such organisations in the second elections. By comparison, it has been estimated that 200 parties contested elections in 13 Western European countries in the period between 1945 and 1977 (Pedersen, 1979:5). In other words, more than twice as many political organisations vied for votes in Central and Eastern Europe in the wake of the Cold War as in three decades of Western European electoral competition during the immediate post-Second World War period. Of interest is how this large number of parties is related to inter-electoral change. - Table 2 about here We would expect change in the size of the party system to be high between the first and the second elections as anti-communist front organisations split and the resultant splinters tested the electoral waters. We would also expect it to be high between the second and the third elections as the reality of the constraints of electoral system began to be felt and consolidation took place. This much is clear from the data presented in Table 2 on the absolute number of parties that contested the first three post-communist elections. In most cases there is a rise between the first and second elections followed by a decline in contenders between the second and third electoral events. The most noteworthy exception to this general pattern is Poland where Solidarity had already fragmented following the transitional elections of 1989 and a remarkable 111 parties and political groupings contested the 1991 elections.11 It is also of interest to note the difference in party activity between the Central European and former Soviet states; there have, on average, been half again as many electoral contestants in the former as in the latter. In some cases this is accounted for by country-specific anomalies. The case of the Polish 1991 elections has already been noted, and the consistently high figures in Romania can be explained with reference to the financial and political incentives for party formation embedded in the legal framework, especially as regards ethnic minority parties.12 But the Bulgarian and Hungarian figures are also above the FSU mean, as is the mean for Poland even when the 1991 elections are excluded. This is not the place to examine the determinants of the number of parties that enter an electoral contest, as this would require separate treatment, but the difference between the two sub-regions may have to do with the fact that personalities loom large and parties have are in Latvia and Estonia also witnessed declines in the number of contenders between the first and the second election most likely also because the transitional election of 1990 had provided an initial testing ground. In Bulgaria the expansion of the party system did not take place until 1994, after the new institutional arrangements had been put in place. 12 In Romania there are reserved seats for minorities and a far lower formal threshold, which generates a proliferation of minority parties (and arguably a proliferation of self-identified ethnic minorities).
11

10

general been less accepted and respected in many of the former Soviet countries than they are in Central Europe (Wyman et al, 1995; Rose and Haerpfer, 1996).13 - Table 3 about here Turning now to volatility, this measure also exhibits a decline over time (though the paucity of data on fourth and subsequent elections makes generalisation difficult beyond the first two data points). Yet the geographic distribution is the reverse of that for the number of parties. Not only was volatility significantly higher in the FSU states but it has also declined less over time in these countries than it has in Central Europe (see Table 3). We find that, as predicted, variations in the level of support among continuous partisan contenders is in almost all cases lower in the single-member components of mixed systems than in their PR counterparts. Russia 1995-1999 is the only exception (though it is worth noting that the volatility between the Ukrainian single-member elections of 1994 and the single-member component of the mixed system used in 1998 is considerably lower than the FSU average, despite the change of system). This provides some evidence to support the theory that PR should increase volatility more than elections in single-member constituencies, though one must be cautious of making generalisations on the basis of the two parts of a mixed system, as there are well-know interaction effects between voting patterns (Moser, 1997; Montgomery, 1999; Birch, 2000; Shugart and Wattenberg, 2000). Unfortunately there are no instances of volatility in a pure SM system in this dataset that would make it possible to isolate the specific influences of mixed systems on party system change in single-member constituencies. The relationship between volatility and formal thresholds is less clear from the data presented in Table 3; the highest volatility measure - that for the 1990 and 1992 Romanian elections is associated with the lowest average threshold (1.5, as there was no threshold in the first election and a threshold of three per cent two years later), but overall there is no statistically significant bivariate correlation between the two variables. Nor is there a significant association between volatility and the change of threshold between two elections. Data for the replacement index are presented in Table 4. The popularity of new parties has exhibited a decline in tandem with both the number of parties in general and the volatility of support for existing parties. We would expect party replacement to be linked to volatility, as more votes for new political formations will in most cases translate into fewer votes for those partisan contenders that are continuous from one election to the next. Indeed there is a correlation of .476 (significant at .005) between the volatility and replacement scores (though no significant association between the replacement score and the average number of parties that contested the two elections). As with volatility, we find that party replacement is considerably (and significantly) higher in the former Soviet Union than it is in Central Europe. New parties and groupings in former Soviet republics have attracted nearly half the vote on average during the post-1991 period, and the proportion of the vote they have won has declined only moderately over time. This pattern may be self-reinforcing as political entrepreneurs recognise the potential gains to be had in the electoral market by founding a new party and therefore leave existing parties in large enough numbers that voters have fewer familiar options to choose from and are more likely to vote for something new. This high level of replacement undoubtedly reflects the generally high
13

The institutional factors that affect entry are in all likelihood those associated with nomination procedures and campaign finance regulation (see Cox, 1997; Lewis, 1998).

11

availability of post-Soviet electorates. But it can also be taken as an indication of the extent to which politics in the former Soviet states revolves around individual politicians rather than around instituitonalised parties. Table 4 about here

The relationship between electoral institutions and party replacement is less clear. Within mixed systems, there were higher levels of party replacement in Hungary in the single-member constituency voting than in the PR component. In Lithuania and Russia the situation was mixed; in Lithuania replacement was higher on the PR ballot at the first time point and higher on the SM ballot at the second, whereas in Russia the reverse was true. This, combined with the lack of any significant correlation between the replacement index and the formal threshold, makes it difficult to generalise as to the effects of electoral system variables on party replacement. Interestingly, there was little evidence that high levels of instability at election t was linked with high levels of instability at election t + 1. Although there is clear continuity in some countries, overall there are no significant correlations between levels of electoral instability at successive elections. Country-specific dimensions of change were evidently of less importance in the early post-transition phase than factors linked to the transition itself. Over time me may well begin to see country-level factors become more prominent, but the limited data on fourth and fifth elections provide scant material on which to base generalisations. Rather it makes sense to analyse the relevant impact on party system change of the variables already examined. Table 5 presents the results of OLS regressions used to model both volatility and replacement. -- Table 5 about here -The single-member and PR components of mixed systems are entered separately in these equations to test for electoral system effects, and as suggested by the preceding discussion, both the electoral system type (single-member or PR) and the level of the formal threshold had significant negative influences on volatility. Once these factors are controlled for, the number of partisan contenders had a positive effect, but of lesser magnitude. This latter finding is at odds with the experience in Western democracies where, as noted above, a large number of parties is linked with greater volatility. This suggests that the effect of the number of parties on volatility may be curvilinear; when there is a relatively compact party system, the addition of new parties increases volatility, whereas when there is a very large number of parties, voters may be so disoriented with the range of choices on offer and it may be so costly to acquire information on all the parties that they may tend to stick with what they know.14 Two slightly different models of volatility are presented in Table 5. The first model (Volatility I) includes two dummy variables: one for the institution of the directly-elected executive president and the other for having been part of the Soviet Union. Neither variable reaches statistical significance, although the president variable falls just short, being significant at .055. When the regional dummy is excluded from the equation, as is done in the second model
14

Another possible explanation is that the model of ideological space which subtends the hypothesis that party N will be associated with volatility is not as relevant in the post-communist context where the ideological position of many parties may be poorly defined (at least in the eyes of the electorate), and where factors such as party leader popularity and clientelism may play a more important role than ideological proximity.

12

(Volatility II), the presence of a directly-elected executive president has, as expected, a modest positive impact on electoral volatility. Also of interest are the variables that are not significant (and were therefore excluded from the final models presented here in the interests of parsimony and conservation of degrees of freedom). These include major changes in the party system (see Tables 3 and 4 for details) as well as changes in the level of the electoral threshold; a dummy for mixed electoral systems was also not found to be associated with volatility. Two additional variables that were tested but failed to reach significance in the volatility model were party replacement and average turnout.15 The model for party replacement, presented in the final column of Table 5, indicates that entirely different factors are involved in support for new parties from those factors linked with volatility. Contrary to expectations, none of the institutional variables was significantly associated with replacement, nor was the number of partisan contenders. By contrast, average turnout and location in the former Soviet Union had moderate positive effects. This accords with our predictions and what we know about electoral participation generally; the solid core of the electorate that turns out to vote faithfully at every election also tends to be most loyal to their chosen parties, so it stands to reason that more extensive participation should be linked to a greater propensity to defect from existing political groups.16 Higher levels of party replacement in the former Soviet states may well reflect the fact that the Soviet Union was more patrimonial than most of its Central European satellites. A patrimonial legacy can be expected to translate into personalistic relations between voters and politicians during the post-communist period (Kitschelt, 1995). In many post-Soviet states personalism undermines the institutional continuity of parties, because support comes to be based on allegiance to or admiration for an individual rather than with the party as a representative of specific interests or a given ideology. Under these conditions individual leaders have the power to bring their supporters with them if they defect to a different party or opt to form their own political association. Thus any leader who becomes popular or perceives that s/he is popular - has an incentive to trade up so as to be part of a more powerful political group or to have a larger role in the group of which s/he is part. In a comparative study of the three Baltic states, Pettai and Kreuzer (2001) find that candidate replacement the proportion of candidates on party lists at election t + 1 who were not candidates at election t averaged about three quarters of the total corpus. This puts instability of electoral results in context. Average rates of electoral-level party replacement hovering around the fifty per cent mark may seem high, but the voters in this case are actually more constant than the political organisations they are voting for. As the most important measures of party strength, elections are engines of both party system consolidation and division. On the one hand they encourage existing parties to come together so as to make a concerted appeal to as large a sector of the public as possible; on the other hand they also promote the emergence of new contenders on the basis of perceived changes of popularity since the previous election.

Variables were also introduced for the length of time between elections (defined both in terms of numbers of months and the natural logarithm of the number of months) as well as for the proportion of votes won by independents. In no case was any of these variables significant. 16 The remainder of the variables which failed to reach statistical significance in the model for volatility also failed to do so in the model for replacement.

15

13

IV Conclusion Party system instability in post-communist Europe appears to be the effect of a combination of influences, including institutional design, specific features of sub-regions within the postcommunist area, and factors such as electoral participation and the number of contenders. All in all, the diversity of influences on inter-electoral volatility and party replacement in postcommunist Europe confirm the desirability of separating these two aspects of party system instability. Separating the two variables also yields some interesting findings that challenge previous generalisations about variations in party system development across the region. Once we control for institutional variables, we find that the former Soviet countries have levels of volatility that are not significantly higher than those found in Central Europe. The distinctiveness of the FSU lies rather in the success rates of new parties there. Also of note is that fact that party proliferation per se is not a cause of volatility; on the contrary, multivariate analysis shows there to be a negative association between the two variables. It must be emphasised that this paper represents only the beginning of an attempt to assess the factors that impinge on party system stability and change in post-communist Europe. The analysis presented here has mainly been concerned with how electoral system design affects the stability of party systems. Further research is needed to investigate the relationship between changes in voters political views and changes in the party system. The concept of block volatility (Bartolini and Mair, 1990) may well make it possible to disentangle changes in party institutions from changes in the ideological proclivities of the electorate. Such an undertaking represents the logical next step in the analysis initiated here. Preliminary evidence from a limited number of countries suggests considerable variability among them between the relative importance of total volatility and bock volatility (Kitschelt et al, 1999:400-1; Markowski, 2001); such analyses could fruitfully be expanded to incorporate the wider range of countries examined here. Further work is also needed to determine the role of such factors as entry criteria and campaign finance regulations in affecting the number of contenders, as well as possible direct effects of these factors on party replacement and volatility. However preliminary these findings, they do accord with studies of party system volatility in other parts of the democratising world, notably Latin America, where Roberts and Wibbels (1999) found that institutional factors play by far the largest role in determining levels of volatility. When more comparative analysis is done on party systems elsewhere, we may well discover that Western European party system stability is the exception rather than the rule. Unlike in Western Europe where parties are known for their ability to adapt to change and absorb successive generations of voters (Bartolini and Mair 1990; Donovan and Broughton, 2001), most party systems in post-communist Europe are uncoupled (Lawson, 1999) or floating (Rose et al, 2001). They respond more to lite-level changes in configurations of alliances than to shifts in the electoral base, and in many states individual politicians rather than political parties constitute the basic building blocks of politics. Yet this does not necessarily imply that political competition in post-Communist Europe lacks the systematicity we generally ascribe to Western party configurations; it simply suggests that such systematicity does not reside where we most often look for it in stable patterns of electoral competition between institutionalised political organisations. The challenge for students of post-communist politics is to discover just how these systems do work and where their regularities lie.

14

Table 1: Institutional Characteristics of 12 Central and Eastern European States Country Bulgaria Czech Republic Estonia Hungary Latvia Lithuania Moldova Poland Romania Russia Slovakia Ukraine Sources: See Appendix 2. Electoral System Type Mixed (1990), PR PR PR Mixed (semicompensatory) PR Mixed (parallel) PR PR PR Mixed (parallel) PR Mixed (parallel) National Threshold (for single parties) 4% 5% 5% 4% (1990), 5% 4% 4% (1992), 5% 4% (1994 & 98), 6% 0 (1991), 5% 0% (1990), 3% (1992 & 96), 5% 5% 5% 4%

15

Table 2: The Number of Partisan Contenders Central Europe Country and Election Number of Year Partisan Contenders Bulgaria - PR 1990 ~38 Bulgaria 1991 38 Bulgaria 1994 48 Bulgaria 1997 39 Bulgaria 2001 14+ Bulgarian mean 35.4+ Czech Republic 1990 16 Czech Republic 1992 21 Czech Republic 1996 20 Czech Republic 1998 18 Czech mean 18.75 Hungary - PR 1990 19 Hungary - SM 1990 28 Hungary - PR 1994 19 Hungary - SM 1994 35 Hungary - PR 1998 15 Hungary - SM 1998 26 Hungarian mean 23.67 Poland 1991 111 Poland 1993 35 Poland 1997 21 Polish mean 55.67 Romania 1990 71 Romania 1992 79 Romania 1996 64 Romania 2000 68 Romanian mean 70.50 Slovakia 1990 17 Slovakia 1992 22 Slovakia 1994 17 Slovakia 1998 17 Slovak mean 18.25 Mean of CE states Sources: Appendix 2. 34.00 Former Soviet Union Country and Election Number of Year Partisan Contenders Estonia 1992 17 Estonia 1995 16 Estonia 1999 12 Estonian mean 15.00 Latvia 1993 23 Latvia 1995 19 Latvia 1998 21 Latvian mean 21.00 Lithuania - PR 1992 17 Lithuania - SM 1992 26 Lithuania - PR 1996 24 Lithuania - SM 1996 28 Lithuania - PR 2000 15 Lithuania - SM 2000 28 Lithuanian mean 23.00 Moldova 1994 13 Moldova 1998 15 Moldova 2001 17 Moldovan mean 15.00 Russia PR 1993 13 Russia SM 1993 13 Russia PR 1995 43 Russia SM 1995 43 Russia PR 1999 26 Russia SM 1999 27 Russian mean 27.50 Ukraine SM 1994 32 Ukraine - PR 1998 30 Ukraine - SM 1998 35 Ukrainian mean 32.33

Mean of FSU states

22.31

16

Table 3: Volatility Scores Country Central Europe Bulgaria - PR Czech Republic Hungary PR ballot Hungary SM ballot* Poland Romania Slovakia Mean CE** FSU Estonia Latvia Lithuania PR ballot Lithuania SM ballot Moldova Russia PR ballot Russia SM ballot Ukraine - SM Mean FSU** Mean** First-Second Election 10.94 (9.68 ) 19.56 (6.91) 24.98 (23.87) 26.79 (25.10) 24.26 (17.86) 56.52 (43.83) 16.17 (8.55) 25.56 (18.55) 23.14 (13.20) 41.71 (25.67) 41.49 (20.36) 36.14 (24.35) 78.17 (32.06) 35.86 (24.51) 30.22 (19.45) 19.71 (16.83) 39.10 (22.02) 32.33 (20.29) Second-Third Election 20.92 (9.95) 19.11 (16.11) 30.69 (29.39) 23.43 (22.30) 13.21 (12.26) 16.34 (13.99) 14.21 (12.77) 18.48 (15.16) 23.74 (12.08) 48.58 (22.99) 42.72 (36.74) 35.77 (28.16) 33.98 (22.45) 24.00 (12.77) 30.79 (16.57) 30.88 (16.59) 17.13 (9.42) [20.42 (13.90)] Third-Fourth Election 25.10 (21.98) 8.56 (7.62) Fourth-Fifth Election 33.07 (21.63)

34.59 (20.9`3) 25.80 (17.78) [20.42 (13.00)]

NB: Figures in brackets represent volatility as a proportion of the total vote. A indicates that there was a major change of electoral system between the two elections included in the calculations; see Appendix 1 for an overview of these changes and for further notes on calculations. * Joint FIDESZ-MPP + MDF candidates are counted together with FIDESZ-MPP, whereas joint MDF + FIDESZ-MPP candidates are counted together with the MDF. ** Figures for mixed systems are averaged so that each state is counted once. Sources: See Appendix 2.

17

Table 4: Party Replacement Scores Country Central Europe Bulgaria - PR Czech Republic Hungary PR ballot Hungary SM ballot* Poland Romania Slovakia Mean CE** FSU Estonia Latvia Lithuania PR ballot Lithuania SM ballot Moldova Russia PR ballot Russia SM ballot Ukraine - SM Mean FSU** Mean** First-Second Election 58.09 61.44 5.64 7.90 26.11 41.77 52.00 41.03 44.38 48.77 61.27 44.48 91.55 45.63 45.72 25.70 51.49 46.26 Second-Third Election 55.63 8.07 2.22 4.95 10.88 14.69 22.77 19.27 40.97 63.44 22.80 32.81 36.45 57.66 45.23 Third-Fourth Election 12.34 11.13 Fourth-Fifth Election 56.22

32.89 50.70 [26.77]

44.02 31.65 [26.77]

NB: A indicates that there was a major change of electoral system between the two elections included in the calculations; see Appendix 1 for an overview of these changes and for further notes on calculations. * Joint FIDESZ-MPP + MDF candidates are counted together with FIDESZ-MPP, whereas joint MDF + FIDESZ-MPP candidates are counted together with the MDF. ** Figures for mixed systems are averaged so that each state is counted once. Sources: See Appendix 2.

18

Table 5: OLS Regression Models Variable Electoral System (dummy for single-member system) Threshold Number of partisan contenders Directly-elected executive president Location in the FSU Turnout N = 32 Adjusted R2 Volatility I -2.732*** Volatility II -2.903*** Replacement

-2.815*** -.994*** .345 .290

-3.037*** -1.249*** .541*** .681*** .448** .573 .414

.599

NB: Coefficients are standardised betas. * = p < .05; ** = p < .01; *** = p < .001

19

Appendix 1: Notes on Methods and Assumptions used in Calculations System changes: Bulgaria switched from a mixed to a fully PR system after the 1990 Constituent Assembly elections; Ukraine switched from a single-member to a mixed system between 1994 and 1998. Prior to the 2000 elections Lithuania switched from absolute to relative majority for elections to the single-member seats of its mixed system; Poland introduced a threshold of xxx following the 1991 elections. Calculations of indeces of party system change are based on the most comparable figures; thus the PR component of the Bulgarian 1990 mixed system is used to calculate volatility figures for the 1990-1991 elections (given that a fully PR system was used in 1991), whereas the single-member component of the Ukrainian 1998 mixed system is used in calculating volatility between the 1994 and 1998 election (given that a fully single-member system was employed in 1994). In absolute majority systems in two rounds (the single-member races in Hungary and Lithuania 1992 and 1996, as well as the Ukrainian election of 1994), results from the first round are employed in calculating volatility, given that the first round is the only one in which all seats were contested. Independent candidates were discarded from the vote totals and percentages were recalculated on the basis of the remaining votes. Decision rules used to determine the continuity of parties between elections: Coalition partners will be considered to have retained their identity, meaning that the parties that make up the coalition will be considered to have contested two consecutive elections, if they contested the first election as separate parties and the second election as a coalition, or vice versa. In this case the votes of the relevant parties will be added together in both elections. In the case of mergers: Mergers that result in a party which contains the original names (substantially unchanged, in the sense that they are still identifiable) of the parties which merged will be treated like a coalition, as continuity of identity is indicated by this name. Mergers that involve party A absorbing party B, such that the resulting political organisation retains the name of party A and party Bs name is lost entirely, will be treated as the extinction of party B (and party A will be considered to have maintained continuity of identity). Mergers that result in the creation of a new party with a new name substantially different from that of either of its component parts will be considered to be a new party, and the parties that merged to form it will be considered extinct. In the case of splits: If the resultant parties both (all) acquire new names following the split, they will be considered new parties and the party which spawned them will be treated as extinct. If, as is most commonly the case, a splinter group breaks away from an existing party and forms a new party, the new party will be treated as such, and the remaining rump party will normally be treated as being continuous with the party as it was before the split. If a party splits and both factions claim ownership of the original name, then the decision of the judicial and/or other body

20

which adjudicates in the dispute will be observed, and the party which loses the right to the original name will be considered a new party.17 If a party is grouped with different coalition partners from one election to the next, continuity is not assumed, as such shifting coalitions indicate lack of consistent identifiability. In the case of name changes: Parties that change their name without experiencing either a split or a merger will be considered to have maintained continuity. In some cases individual parties will have experienced a number of the above identity changes, either simultaneously or over time. Where possible, the above rules will be applied to consecutive changes. When there is possible ambiguity as to how the schema has been applied in practice, this will be indicated in the form of notes. No attempt is made to make a clear distinction between parties, coalitions, and other types of electoral contender (with the exception of independents which are excluded, as indicated in note 7 above). Notes on individual countries: cases which cannot be determined unambiguously according to the above rules and exceptions Bulgaria: Between 1990 and 1997 there were three main forces in Bulgarian electoral politics: the Bulgarian Socialist Party (BSP), the Union of Democratic Forces (SDS) and the Movement for Rights and Freedoms (DPS). In 1997 each of these parties formed a coalition with other smaller parties that had previously part of different coalitions. Nevertheless, the continuity of the three main parties will be assumed. The Czech Republic: The Left Bloc of 1992 is assumed to be continuous with the Communist Party of Bohemia and Moravia (KSCM) in 1996 (not the Left Bloc party). Estonia: The Left Alternative in 1992 is considered to be continuous with the Justice Alliance in 1995.
17

It may at first seem inconsistent that continuity is attributed to parties that merge but not to those that split. But upon reflection, this is reasonable, If a party splits there is bound to be less continuity of organisational resources as the resultant splinter organisations will mostly have to establish new branches, publications, etc, whereas mergers allow for the consolidation of existing resources. Moreover, party supporters will find it easier to follow their party if it merges with another one than they will if it splits, in which case they will have to decide which group it supports.

21

Secure Home in 1992 is considered to be continuous with the Coalition Party-Rural Union (KMU) in 1995. The Popular Front in 1992 is considered to be continuous with the Estonian Centre Party + the Estonian Entrepreneurs Party (EEE) in 1995 Our Home is Estonia (MKE) in 1995 is considered to be continuous with the Russian Party in Estonia (VEE) in 1999. Hungary: The Hungarian Socialist Workers Party (MSZMP) in 1990 is considered to be continuous with the Hungarian Workers Party (MP) in 1994. Latvia: The United List of the Latvian Farmers Union (LZS), Latvian Christian Democratic Union (LKDS), and Latgale Democratic Party in 1995 is considered to be continuous with the Latvian Farmers Union (LZS) + the Latvian Christian Democratic Union (LKDS) in 1993. The coalition of the Latvian National Conservative Party and the Latvian Green Party in 1995 is considered to be continuous with the Latvian National Independence Movement (LNNK) + the Green List (ZS) of 1993 Moldova: The Alliance of Democratic Forces Electoral Bloc (AFD) in 1998 is considered to be continuous with the National Liberal Party (PNL) in 2001. The Electoral Bloc for a Democratic and Prosperous Moldova (PMDP) in 1998 is considered to be continuous with the Democratic Party of Moldova (PDM) in 2001. Poland: Solidarity Electoral Action (AWS) in 1997 is considered to be continuous with Solidarity + Fatherland + the Coalition for the Republic + the Non-Party Reform Bloc + the Confederation for an Independent Poland + the Centre Alliance + the Peasant Alliance The Union of Political Realism (UPR) in 1993 is considered to be continuous with the Union of the Right of the Republic in 1997. Slovakia:

22

The Movement for a Democratic Slovakia (HZDS) in 1992 is considered to be continuous with the Movement for a Democratic Slovakia and Peasants Party of Slovakia (HZDS + RSS) in 1994 The Common Choice block in 1994 is considered to be a coalition of the SDL, the SDSS and the SZS The DS in 1994 is considered to be continuous with the DS + the ODU Romania: The Democratic National Salvation Front (FDSN) is considered to be continuous with the Romanian Party of Social Democracy (PDRS) in 1996. Russia: Russias Choice (VR) in 1993 is considered to be continuous with Russias Democratic Choice United Democrats (DVR-OD) in 1995. The Congress of Russian Communities (KRO) in 1995 is considered to be continuous with the Congress of Russian Communities (KOR) + Movement of Yuri Boldyrev in 1999 The Liberal Democratic Party of Russia (LDPR) in 1995 is considered to be continuous with the Zhirinovski Block in 1999. Ukraine: The Ukrainian Christian Democratic Party (UkhDP) in 1994 is considered to be continuous with the Forward Ukraine! block in 1998

23

Appendix 2: Data Sources Data on electoral results and electoral system design are taken from the Database on Central and Eastern European Elections: Results and Legislation at www.essex.ac.uk/elections. Data on parties are taken from the following sources: Szajkowski, Bogdan, Political Parties of Eastern Europe, Russia and the Successor States, Harlow: Longman, 1994. Alan Day, Richard German and John Campbell (eds)., Political Parties of the World, London: Cartermill, 1996. Richard Rose, Neil Munro and Tom Mackie, Elections in Central and Eastern Europe Since 1990, Strathclyde: Centre for the Study of Public Policy, University of Strathclyde, 1998. Notes on Recent Elections section of Electoral Studies, various issues. These general sources were in some cases supplemented by sources listed in the bibliography and for individual countries by the following: Estonia: Vello Pettai and Marcus Kreuzer, Party Politics in the Baltic States: Social Bases and Institutional Context, East European Politics and Societies 13.1 (1999) 148-89. Latvia: Vello Pettai and Marcus Kreuzer, Party Politics in the Baltic States: Social Bases and Institutional Context, East European Politics and Societies 13.1 (1999) 148-89. Lithuania: Vello Pettai and Marcus Kreuzer, Party Politics in the Baltic States: Social Bases and Institutional Context, East European Politics and Societies 13.1 (1999) 148-89. Poland: Frances Millard, Polish Politics and Society, New York and London: Routledge, 1999. Russia: Yu. K. Abramov and T. Yu. Golovin (eds.), Politicheskie partii I dvizhennya Rossii 1998: yezhogodnik, Moscow: Press Ltd., 1998.

24

O. K. Zastrozhnoi (ed.), Obshcherossiiskie izbiratelnye obedineniya: nakanune vyborov deputatov Gosudarstvennoi Dumy Federalnovo Sobraniya Rossiiskoi Federatsii tretevo sozyva: spravochnik, Moscow: Ves Mir, 1999. Grigorii V. Golosov, Political Parties in the 1993-1996 Elections in Vladimir Gelman and Grigorii V. Golosov (eds.), Elections in Russia, 1993-1996: Analyses, Documents and Data, Berlin: Sigma, 1999. Ukraine: Politychni parti Ukrany, Kiev: KIS, 1998.

25

References Bartolini, Stefano and Peter Mair, 1990, Identity, Competition, and Electoral Availability: The Stabilisation of European Electorates 1885-1985, Cambridge: Cambridge University Press. Birch, Sarah, forthcoming, Two-Round Electoral Systems and Democracy, Comparative Political Studies. --- 2001, Electoral Systems and Party Systems in Europe East and West , Paper prepared for presentation at the first conference of the European Consortium for Political Research, Canterbury, 6-8 September, 2001. --- Elections and Democratization in Ukraine, Basingstoke: Macmillan. Broughton, David and Mark Donovan, Changing Party Systems in Western Europe, London and New York: Pinter, 1999. Budge, Ian, 1982, Electoral Volatility: Issue Effects and Basic Change in 23 Post-War Democracies, Electoral Studies 1, 147-68. Colton, Timothy J., 2000, Transitional Citzens: Voters and What Influences Them in the New Russia, Cambridge, MA: Harvard University Press. Coppedge, Michael, 1998, The Dynamic Diversity of Latin American Party Systems, Party Politics 4.4, 547-68. Cotta, Maurizio, Structuring the New Party Systems after the Dictatorship: Coalitions, Alliances, Fusionsand Splits During the Transition and Post-Transition Phase, in Geoffrey Pridham G. Lewis (eds.), 1996, Stabilising Fragile Democracies: Comparing New Party Systems in Southern and Eastern Europe, London and new York: Routledge. Cox, Gary, 1997, Making Votes Count: Strategic Coordination in the Worlds Electoral Systems, Cambridge: Cambridge University Press. Crewe, Ivor, 1985, Introduction: Electoral Change in Western Democracies: A Framework for Analysis, in Ivor Crewe and David Denver (eds.), Electoral Change in Western Democracies: Patterns and Sources of Volatility, London and Sydney: Croom Helm, 122. Crewe, Ivor and David Denver (eds.), Electoral Change in Western Democracies: Patterns and Sources of Volatility, London and Sydney: Croom Helm. Duverger, M.,1959, Political Parties: Their Organisation and Activity in the Modern State. 2nd ed. London: John Wiley. Jones, Mark P. 1995. Electoral Laws and the Survival of Presidential Democracies. Notre Dame and London: Notre Dame University Press. King, Charles, 2000, The Moldovans: Romania, Russia and the Politics of Culture, Stanford, CA: Hoover Institution Press. Kitschelt, Herbert, 1995, Formation of Party Cleavages in Post-Communist Democracies: Theoretical Proposition, Party Politics 1.4, 447-72. Kitschelt, Herbert, Zdenka Mansfeldova, Radoslaw Markowski, and Gabor Toka, 1999, PostCommunist Party Systems: Competition, Representation and Inter-Party Cooperation, Cambridge: Cambridge University Press. Lawson, Kay, 1999, Cleavages, Parties, and Voters, in Kay Lawson, Andrea Rommele and Georgi Karasimeonov (eds.), Cleavages, Parties and Voters; Studies from Bulgaria, the Czech Republic, Hungary, Poland, and Romania, Westport, CT and London: Praeger, 1934.

26

Lewis, Paul G., 1998, Party Funding in post-communist east-central Europe in Peter Burnell and Alan Ware (eds.), Funding Democratization, Manchester: Manchester University Press, 137-57. --- 2000, Political Parties in Post-Communist Eastern Europe, London and New York: Routledge --- 2001 (ed.), Party Development and Democratic Change in Post-Communist Europe: The First Decade, London and Portland, OR: Frank Cass. Lijphart, 1994, Electoral Systems and Party Systems: A Study of Twenty-Seven Democracies, 1945-1990, Oxford: Oxford University Press. Mainwaring, Scott and Timothy R. Scully, Building Democratic Institutions: Party Systems in Latin America, Stanford, CA: Stanford University Press, 1995. Mair, Peter, Party System Change: Approaches and Interpretations, Oxford: Oxford University Press, 1997. Markowski, Radoslaw, 1997, Political Parties and Ideological Spaces in East Central Europe, Communist and Post-Communist Studies 30.3, 221-54. --- 2001, Party System Institutionalization in New Democracies: Poland A Trend-Setter with No Followers, in Paul G. Lewis (ed.), Party Development and Democratic Change in Post-Communist Europe: The First Decade, London and Portland, OR: Frank Cass, 5577. Miller, William L., Stephen White and Paul Heywood, 1998, Values and Political Change in Post-Communist Europe, Basingstoke: Macmillan. Montgomery, Kathleen, 1999, Electoral Effects on Party Behaviour and Development, Party Politics 5.4, 507-23. Moser, Robert G., 1997, The Impact of Parliamentary Electoral Systems in Russia, Post-Soviet Affairs 13.3, 284-302. --- 2001, Unexpected Outcomes: Electoral Systems, Political Parties, and Representation in Russia, Pittsburgh: University of Pittsburgh Press. Olson, David M., Party Formation and Party System Consolidation in the New Democracies of Central Europe, Political Studies 46.3, 432-64. Olson, David M. and Philip Norton (eds.), The New Parliaments of Central and Eastern Europe, London and Oregon: Frank Cass, 1996. Pedersen, Mogens, 1979, The Dynamics of European Party Systems: Changing Patterns of Electoral Volatility, European Journal of Political Research 7, 1-26. Pettai, Vello and Marcus Kreuzer, Institutions and Party Development in the Baltic States, in -- 2001 (ed.), Party Development and Democratic Change in Post-Communist Europe: The First Decade, London and Portland, OR: Frank Cass, 107-25. Pridham, Geoffrey and Paul G. Lewis (eds.), 1996, Stabilising Fragile Democracies: Comparing New Party Systems in Southern and Eastern Europe, London and new York: Routledge. Rae, Douglas., 1971, The Political Consequences of Electoral Laws. 2nd ed. New Haven and London: Yale University Press. Roberts, Kenneth M. and Erik Wibbels, 1999, Party Systems and Electoral Volatility in Latin America: A Test of Economic, Institutional, and Structural Explanations, American Political Science Review 93.3, 575-90. Rose, Richard, Neil Munro and Tom Mackie, 1998, Elections in Central and Eastern Europe Since 1990, Strathclyde: Centre for the Study of Public Policy, University of Strathclyde.

27

Rose, Richard, Neil Munro and Stephen White, 2001, Voting in a Floating Party System: the 1999 Duma Election Europe-Asia Studies 53.3, 419-43. Rose, Richard and Christian Haerpfer, 1996, New Democracies Barometer IV: A Ten Nation Survey, Studies in Public Policy 262, Strathclyde: Centre for the Study of Public Policy, University of Strathclyde. Shugart, Matthew Soberg and John M. Carey. 1992. Presidents and Assemblies: Constitutional Design and Electoral Dynamics. Cambridge: Cambridge University Press. Shugart, Matthew Soberg and Martin P. Wattenberg (eds.), Mixed-Member Electoral Systems: The Best of Both Worlds?, Oxford: Oxford University Press, 2001. Taagepera, Rein, 1998, Nationwide Inclusion and Exclusion Thresholds of Representation, Electoral Studies 17.4, 405-18. Taagepera, Rein and Matthew Soberg Shugart, 1989, Seats and Votes: The Effects and Determinants of Electoral Systems, New Haven and London: Yale University Press. Toka, Gabor, Political Parties and Democratic Consolidation in East Central Europe, Studies in Public Policy 279, Strathclyde: Centre for the Study of Public Policy, University of Strathclyde, 1997. Wyman, Matthew, Stephen White, Bill Miller and Paul Heywood, 1995, The Place of Party in Post-Communist Europe, Party Politics 1.4, 535-48.