NATIONAL SCHOOL OF POLITICAL AND ADMINISTRATIVE STUDIES FACULTY OF POLITICAL SCIENCE

DOCTORAL THESIS

PARTY COMPETITION IN CENTRAL AND EASTERN EUROPE: THE CZECH REPUBLIC, HUNGARY, POLAND AND ROMANIA

SUPERVISOR: PROFESSOR DR. ADRIAN MIROIU

PhD STUDENT: RZVAN GRECU

BUCHAREST, 2008

CONTENTS

Acknowledgements ............................................................................................................. 4 CHAPTER 1: INTRODUCTION .............................................................................................................. 6 1.1. Party Competition in Uncertain Environments ....................................................... 6 1.2. Parties, Policies and Democracy ........................................................................... 12 1.3. The Organization of the Thesis ............................................................................. 16 CHAPTER 2: THE ECONOMIC BASES OF POLITICS: A PRELIMINARY VIEW ......................... 17 2.1. The Study of the Political Economy...................................................................... 17 2.2. The Methodological Individualism and the Theory of Choice.............................. 21 2.3. Rationality and Rational Behaviour ...................................................................... 25 2.4. Rational Politicians and Political Action............................................................... 30 2.5. Two Views on Political Competition: Sociological and Economic Models ......... 33 2.6. How the Voters Decide?........................................................................................ 45 2.7. A Crude Model of Party Competition with Downsean Parties and Voters ........... 47 2.8. Summery................................................................................................................ 50 CHAPTER 3: ECONOMIC APPROACHES TO THE THEORY OF PARTY COMPETITION ......... 53 3.1. Introduction ........................................................................................................... 53 3.2. The Voting Paradox as a Puzzle for the Democratic Theory ................................ 54 3.3. The Basic Concepts of Spatial Analysis of Politics .............................................. 58 3.4. The Classical Spatial Models of Party Competition.............................................. 60 3.5. Variations of Unidimensional Spatial Party Competition and Voting: Critiques of Classical Models and MVT .......................................................................................... 69 3.6. Multidimensional Voting: The Power of Mean Positions ..................................... 84 3.7. Competition in Multiparty Format ........................................................................ 88 3.8. Conclusions ........................................................................................................... 91 CHAPTER 4: INFORMATION AND ENTRY AS DETERMINANTS OF PARTY COMPETITION 93 4.1. Introduction ........................................................................................................... 93 4.2. The Models ............................................................................................................ 94 4.3. The Implications of the Models: How Information and Institutions Shape Electoral Competition in Multiparty Settings?........................................................... 102 CHAPTER 5: HYPOTHESES, METHODOLOGY AND DATA........................................................ 107 5.1. Introduction ......................................................................................................... 107 5.2. Rebuilding the Ship at Sea: Party System and Political Parties in New Democracies of CEE .................................................................................................. 109 5.3. The Strategy of Research: The Empirical Hypotheses........................................ 119 5.4. The Data Sources and Data Analysis .................................................................. 121

CHAPTER 6: AN OUTLINE OF THE CZECH, HUNGARIAN, POLISH AND ROMANIAN PARTY SYSTEMS ...................................................................................................................... 136 6.1. Introduction ......................................................................................................... 136 6.2. Parties and Party System in the Czech Republic................................................. 138 6.3. Parties and Party System in Hungary .................................................................. 140 6.4. Parties and Party System in Poland ..................................................................... 143 6.5. Parties and Party System in Romania.................................................................. 145 6.6. Concluding Remarks ........................................................................................... 150 CHAPTER 7: PARTY COMPETITION IN THE CEE: IDEOLOGICAL CONVERGENCE AND CENTRIPETAL COMPETITION ................................................................................. 151 7.1. Introduction ......................................................................................................... 151 7.2. Institutional Barriers for Entry in the CEE.......................................................... 153 7.3. On the Position of Median Voter in the CEE Elections ...................................... 160 7.4. Ideological Convergence and Centripetal Competition in Competitive Elections: Czech Republic, Hungary, Poland, and Romania in Comparative Perspective ......... 165 7.5. Discussion............................................................................................................ 185 CHAPTER 8: CONCLUSIONS: IMPLICATIONS OF ELECTORAL COMPETITION IN THE CEE ........................................................................................................................................ 187 8.1. Introduction ......................................................................................................... 187 8.2. The Implications of Centripetal Competition for the CEE Democracies............ 189 8.3. The Implication of Results for Political Competition ......................................... 192 8.4. The Implication for Political Representation....................................................... 194 8.5. Final Remarks:..................................................................................................... 197 APPENDIX .................................................................................................................... 199 Appendix 1: Formal proof of Wittmans alternative competitive model (Chapter 3, Section 3.5.4) .................................................................................................................................... 199 Appendix 2: Formal proof of Davis and Hinich (1968) (see the Chapter 3, Section 3.6)............... 201 Appendix 3: Party Manifestos Coding Scheme .............................................................................. 202 BIBLIOGRAPHY .......................................................................................................... 213

Acknowledgements

During the academic process that led to this thesis, I have been intellectually influenced by many reputable scholars to whom I owe a great part of my academic achievements. First, I have to thank Professor Adrian Miroiu who has been over these four years an invaluable source of academic mentorship. Not only that Professor Miroiu kindly accepted to supervise a thesis on a topic (party politics) not very relevant for his main academic interests, but he also provided me with theoretical guidance in probably the most crucial part of the doctoral studies: the beginning. I largely owe him the theoretical instrument and the analytical reasoning of this thesis and I personally benefited a lot from the countless discussions we have had during the years about rational choice theory, social choice theory or spatial models of party competition. This thesis represents a small tribute I pay for Professor Miroius efforts to promote rational choice theory among academic disciplines in the Romanian political science departments. I also particularly thank Professor Stephen Whitefield who offered me invaluable supervision during and after my research fellowship at University of Oxford. He patiently accepted my stubborn rational choice reasoning which often clashed with his sociological arguments and guided my empirical research, very often raising questions that helped me to clarify the entire argument of the thesis. His comments and suggestions have significantly shaped the logic of the thesis in many respects and gave me the necessary energy to reformulate and clarify the argument even when I was almost fed up with the entire thesis. Also, this thesis would not have been the same (in the good parts) without the weekly discussions we had in his office at University of Oxford and during the seminar of CEE politics Professor Whitefield organized in the Spring of 2006. Special thanks are due to my good friends, Calvin Mouw and Laurentiu tefan, with whom I have worked very close in various projects during the last five years. They have been always very supportive and ready to provide invaluable feedback; both of them have had a significant influence over my intellectual background for which I have to thank them publicly. Also, my appreciation goes to the numerous friends I have from The Invisible College and The Romanian Society of Political Science, including here Liliana Popescu,

Lucian Catrina, Anca Gheau, Aurelian Muntean, Annette Freyberg-Inan, Irina Ionescu, Arpad Todor and all other persons with whom I spent three wonderful years working for IC and SRSP. I need to acknowledge the tremendous help of Professors H.D. Klingemann and Andrea Volkens who kindly agreed to share with me the manifestos data even before their seminal book was published by Oxford University Press in 2006. Also, Professor Gabor Toka offered me information about 1994 Hungarian elections which I very much needed. I thank them all again. There are also several institutions which have offered me institutional support during the doctoral studies. The Invisible College and The Civic Education Project, Romania, offered me valuable logistic and material support at the beginning of my doctoral studies, for which I thank again Liliana Popescu and Laurentiu tefan. Together with The Academic Fellowship Program, these two institutions have significantly contributed to the development of my professional career in political science. Also, Open Society Institute and The British Council financed my research PhD fellowship at University of Oxford in 2005-2006. Above all, I would like to thank from all my heart to three very special persons. I dedicate this thesis to my parents whom I owe a great education and all these beautiful years from childhood to adulthood. To my father, particularly, I owe my studies in political science and to my mother, her steady support for my professional aspirations. I can not thank enough for how much they have done for me. Last but not least, I need to thank my wife, Georgiana, who always offered me the moral support and who kindheartedly accepted my long absences from home caused by the various research fellowships. Words can not express enough my gratitude and feelings to her.

CHAPTER 1: INTRODUCTION

1.1. Party Competition in Uncertain Environments

The development of political parties and party systems has been a favoured topic for the scholars of political process in the Central and Easter Europe (hereafter CEE). Political parties have been analysed in various regional contexts, such as the development of partyvoter linkages and voting behaviour, government formation process, stabilization of national party systems and electorates and so forth. Yet only little systematic attention has been paid to one of the most interesting political phenomena in the region: the policy strategies and ideological commitments of political parties in the electoral process. The thesis focuses on electoral competition in the CEE. There are two fundamental research questions I address in this thesis. The first is how the political parties compete in the environment of CEE that is notoriously characterized by high political uncertainty? The second fundamental question is what are the political consequences of the pattern, if any, of electoral competition in the region? Concretely, my study explores whether political parties in the CEE do use centrist or divergent electoral strategies in order to maximize their votes in national elections. If a pattern of political competition emerges, what can we say about their causes and political consequences? Also, there are two fundamental propositions in my thesis. The first argues that political parties in the CEE make use extensively of convergent ideological strategies in the national elections and this pattern of competition characterizes both stable (or consolidated) and unstable (unconsolidated) national party systems. Though the influence of exogenous factors like the international organizations (EU, IMF and so forth) can not be fully ignored, my findings point to four main causes of such pattern of electoral party competition: 1) the desires of political parties to win elections; 2) the political uncertainty that has characterized the CEE politics during 1990s; 3) the party dependence of resources (mainly the state funding and media coverage) that come from outside the party organization and

which constrains the political parties in the region to behave as vote-seeking political actors; and 4) the political institutions that emerge out of the desire of self-interested political leaders and parties to control the electoral process, such as electoral thresholds and legislation on public financing. In what concerns the second fundamental proposition of my research, the thesis suggests that the self-interested political parties have produced an electoral competition which may have been responsible at least in part for the success of political democracy in the region, despite the serious political, social and economic transformations that harmed our societies during the transition period. However, for the moment, this proposition has a limited empirical support and the reader should regard it rather as a hypothesis that has to be farther explored with more suitable research design. Currently, it suffers from the same problems that are very common in almost all studies of democratization: there are many key theoretical variables and only limited empirical cases available where we can test their effects. Since in the thesis I have evidence from only four countries in the region, the Czech Republic, Hungary, Poland and Romania, all successfully democratized countries, I can not test properly this argument. However, this second proposition that is suggested by the results the electoral party competition has the potential of being a significant advance in our understanding of democratization and democratic consolidation. The thesis has a third, more theoretical argument. The standard approaches to party competition, particularly the spatial theories, are often blind to the environment and institutional settings in which the competition takes place. As the literature emphasizes, the relationship between parties, voters and policies is indeed fundamental for political competition. Also, the core literature rightly points that institutions influence the political choice and strategies of political parties and voters. A proportional electoral system, for instance, encourages political parties to take on more extreme policy positions because they can obtain parliamentary representation even by speaking to relatively small number of extremist voters. However, less considered in the literature is that the political parties are not just political actors that obey political institutions. They can and habitually they do shape the political institutions so that the institutional system becomes congruent with their political goals. Consequently, I argue in the following chapters that political parties in the CEE structure electoral institutions that give them advantage in electoral competition

against the outsiders (or new parties). These electoral institutions play a role in the convergent strategies of political parties in elections. The political convergence of political parties in the CEE is a puzzling result. After all, the major theories of party systems and political competition have implied that the CEE politics should be a politics of ideological divergence. The sudden end of communist systems in the region let the new political leaders and voters cope with numerous political, social, cultural and economical issues, all of them raising opportunities for political and ideological conflict. Also, the democratic (if any) and the pre-communist history of countries from the CEE were not particularly encouraging for ideological convergence. Nor the institutions of the new democracies were favourable factors in this regard since all countries in the CEE emerged in transition with multiparty systems and with mixed or even pure PR electoral systems. Furthermore, the result is puzzling because the pattern of party competition seems to be significantly different than the one taking place in the Western Europe, where political parties formulate divergent policy platforms to speak to voters. What can then explain the political convergence in the region? As in all relationships in social and political science, there is not a single cause but a set of causes. The Figure 1.1 depicts the theoretical relationships between the four variables I formulate above and the ideological convergence of parties in the national elections. One principal cause of party convergence in electoral competition is the desire of leadership and parties to win the election. This is one of the standard assumptions in the spatial theories of party competition (see the Chapter 3), but it hardly can explain alone the ideological dynamics in the CEE elections. Why the political parties in the region should behave as vote-hunters in elections after all? The argument I advance in the following chapters is that the features of CEE politics constrain the political parties to behave electorally as voteseekers. First of all, the political parties in the CEE compete, as I shall show in the Chapter 5, in a very uncertain political environment. As suggested by systemic indicators such as the electoral volatility, political parties in the CEE do not benefit from the constant support of the electorates. This uncertainty has reduced the capacity of the political parties to foresee effectively their political support during the 1990s and the danger of being wiped

out from the parliamentary party system has partly determined them to behave as voteseekers in elections.

Political uncertainty (high electoral volatility)

Political institutions (electoral thresholds, state funding, other relevant resources)

Political parties as vote-seekers

Electoral convergence

Scarce internal political resources (low membership)

Figure 1.1: Causes of electoral convergence in the Central and Eastern Europe

Other two peculiarities of the CEE politics have been essential in determining political parties to assume vote-seeking strategies. The second variable is the low development of party organization in the region which has had significant political consequences. The political parties have had low party membership that have generated only very limited political resources. Membership fees have been insufficient for sustaining the political activities of political parties, which have become dependent of the political resources coming from outside the party organization. As I argue in the Chapter 5 and the Chapter 7, these resources have come usually from the state, but under one essential condition: the political parties have to win offices (generally parliamentary seats) to be eligible for these allowances. These spoils from the public offices are important resources for political parties in the CEE. However, in obtaining these resources, the established political parties have faced the competition of new political parties that have joined the political competition from time to time. How do they respond to this threat which

endangers their political survival or may give them smaller shares of the political resources? Designing institutions that reduce or eliminate the probability of competing with new political parties (the third variable) has been the typical answer to this question. To certain extent, these institutions reassure the established political parties that they will continue having access to key political resources (state funding, access to media). But these institutions also produce a significant political effect in electoral competition. The standard argument in the spatial theories of party competition, which was formulated as a critique to the findings of the classical models, suggested that the political parties diverge in electoral competition in order to thwart the danger of new entries at the extreme positions (Palfrey 1984, Greenberg and Shepsle 1987). However, as I argue in the Chapter 4, this is not necessarily the case: the political parties can design institutions that keep new parties out of competition and prevent them of winning seats against the established political parties. When the later parties effectively design such institutions and succeed in keeping out new political parties, the type of electoral competition changes to ideological collusion. Theoretically, I approach the question of how political parties compete in the CEE elections by using a rational choice perspective: the spatial theories of party competition. The rational choice theory provides us with a variety of parsimonious, rigorous and elegant theories about competition in different political contexts. Such theoretical approach not only gives us insights about the relevant factors influencing the ideological and policy competition among political parties, but also it provides us with rigorous theoretically driven hypotheses that could be tested in various empirical settings. At the same time, this reach theoretical approach has produced also many conflicting results which determined many political scientists to question utility of the theory for social and political research (Green and Shapiro 1994, Stokes 1963). However, by selecting carefully the key assumptions and variables, the thesis provides support for the claim that rational choice approach has very much to offer intellectually in the social and political sciences. Empirically, I confront the hypotheses generated by the rational choice reasoning in the Chapter 5 with empirical evidence gathered from four countries in the region: the Czech Republic, Hungary, Poland and Romania. These four cases offer sufficient empirical variety to increase the chances that my argument proves incorrect. On the one hand, our

10

cases differ quite substantially in terms of stability of national party systems and patterns of political competition. Up-to-date research on party systems in the CEE regularly point to the fact that the Czech and Hungarian party systems are much more stable and consolidated than the Polish and Romanian party systems, both in terms of the party alliances, government coalitions and links between the electorate and political parties. The competition among political parties is considered to be well-structured (or more programmatic) in the Czech Republic, Hungary and Poland, on the one hand, and nonprogrammatic in Romania on the other hand, some significant differences inside the first group being noted (Kitschelt et al 1999). The partisan affiliation seems to be much more developed in the Czech Republic and Hungary than in Poland and Romania. Also the Hungarian and Czech party systems have been quite stable in terms of party labels, whereas new political parties have emerged within the Polish and Romanian party systems after almost each electoral cycle. On the other hand, there is some variation in what concerns the institutional framework of these four CEE democracies. Poland and Romania are semi-presidential republics that may influence political parties ideological and policy cohesion if Juan Linzs (1994, 1996) argument about the negative role of elected presidents over party systems is correct. The Czech Republic and Hungary are pure parliamentary regimes, less personalized in theory than the semi-presidential regimes, and thus more auspicious for ideologically well-structured party systems. In the first systems, political parties may have more incentives to behave in elections as vote-seekers given that the party candidates for presidential office need to obtain the support of the majority of voters.1 Unfortunately, the cases do not give us much variance in regards to other institutions, such as the electoral systems. Only Hungary differs in the group, having a mixed three-tier electoral system, with Czech Republic, Poland, and Romania using sorts of PR electoral systems. However, pure majoritarian electoral systems are rare and the PR system predominates in the region. At the same, the fact that most of electoral systems encourage a broad representation of
1

A presidential race may have an impact on the party strategies in parliamentary elections. Given that the party candidate for presidential office has to win a majority of votes (or at least a plurality at the later stage) in a majority run-off system, it may determine the parties to take on less extreme policy positions which allow them to negotiate the political support for the second runoff with a larger number of parties. In pure parliamentary systems, parties can focus only on policies which get the support of the largest share of the electorate, so that the strategies may differ between these two groups of countries.

11

social interests is a positive aspect for the present research: the party convergence in election can not be the product of institutional system alone.

1.2. Parties, Policies and Democracy

The study of political parties has been a fruitful venture in political science. Our understanding of how political parties function as agents of political democracy has grown constantly during the last hundred years, as far as the study of modern political parties could be traced back. And yet, in spite of (or perhaps due to) this long period of time, the party politics has proved to be a very rich area of intellectual research, debates and confrontation that have allowed the researchers to come across with new ideas that have further expanded our knowledge about political process. The agreement about the importance of political parties for modern political democracies has not been accompanied by an agreement about how we could study political parties. Far from being a drawback, this has generated a complex and fascinating set of knowledge about the role of parties in our political systems. Generally, the scholars of political parties have used two main approaches that are briefly explored in the next chapter of this thesis. The behaviourist, or sociological approach, focuses on the relationship between political parties and the social structure, arguing that political parties are constrained to represent to some extent particular social groups that offer political support in exchange for policies. The early sociological model (usually called the Chicago model) emphasized the relationship between political parties and cohesive social groups, defined in terms of class, religion, ethnicity or other social characteristics. This theoretical approach has intellectual roots in Lipset and Rokkan (1967) seminal argument about the role of social conflicts and cleavages for the development of party alignments in the West European democracies. Later, as the social and economic developments of modern democracies weakened the group loyalties and cohesion with important consequences for the structure of party systems, political sociologists developed another explanation that

12

emphasized the role of party identification for voting behaviour of voters and their links with established political parties (Dalton: 2002, Campbell et al. 1960). On the other hand, the economic models see political parties and their leaders as rational actors who pursue particular political goals and, contrary to sociological models, enjoy some degree of autonomy in defining these political goals. However, there is a growing tendency in the last decade to combine these two approaches into unified models of party choice and party competition, which include both rational choice and sociological assumptions (Adams, Merrill, and Grofman: 2005). The boundaries between the two intellectual approaches have significantly faded away during the last decades when the empirical research have started to incorporate and test simultaneously sociological and rational choice hypotheses.2 However, for the sake of highlighting the structure of rational choice reasoning, I shall discuss this strategy of research in the following chapter by discriminating theoretically between pure sociological models and pure rational choice arguments. Precisely fifty years ago, Anthony Downs (1957) published the celebrated An Economic Theory of Democracy, a truly manifesto of rational choice theory and economic modelling of party competition. Paradoxically, Downs work was at the same time a major success and noteworthy failure. A success, on the one hand, because An Economic Theory of Democracy has been one of the most cited books of the modern political science, whose primarily merit was to extend the findings of the early social choice theory, particularly the Duncan Blacks Median Voter Theorem (hereafter MVT), to mass elections and electoral competition. Downs work has been the central point of what is probably the largest research program in the history of political science, with hundreds of articles and books addressing the Downsean model of political competition. However, to certain extent, Downs enterprise in the research of electoral competition could be regarded as a failure. For a long period of time his innovative research design for studying political parties has remained confined to only a small group of researchers, most of who have regarded themselves as economists rather than political scientists. In fact, the vast number of scholars of party politics has been divided ever since 1957 into a small group of people who

For arguments in favour of combining the two research strategies, see Hechter (1997), Kiser and Hechter (1998).

13

greatly advanced the Downsean work in the formal theory of political party competition, and a majority of scholars who, although acknowledging and often praising Downs contribution, rejected his method of analysis. But, whatever the shortcomings of Downsean model of political competition and however fashionable is to criticize and even dismiss Downs work as irrelevant, one thing can hardly be ignored: there is impossible to address the issue of electoral party competition and the development of political ideologies without at least referring to the Downsean theoretical framework. At the heart of the Downsean work has been the relationship between parties, voters and policies (or ideology). The logic of democratic politics says that political parties make use extensively of policies and ideology in order to get elected or maintain in power. The parties in power use the governing mechanisms to implement the policies they promise to voters during the electoral campaigns to pursue once they get elected. Conversely, the parties in opposition formulate alternative sets of policies aiming at obtaining the support of voters. Even though the political parties, once in power, rarely implement the full set of the policies they announce before and during the electoral campaigns, the proper function of the political democracy requires a high degree of convergence between the pre- and postelectoral policies. What makes however this relationship between parties, policies and voters function the way it does is the assumption that the voters require policy benefits from politicians in exchange for their electoral support and that these policy benefits are, to certain degree, stable over time. Such assumption represents the bases of accountability principle of the modern democracies which can be formulated in two sentences that have been extensively researched in political science. The first sentence suggests that the political parties which do not act responsively, that is they implement different sets of policies than those advocated before winning the elections, are punished by voters in the subsequent elections because they proved to be unreliable politically. The second sentence is considered to be the core element of accountability: the political parties are constrained in advancing policies by their past record and, in order to remain credible, they need to consider future policy commitments in the light of the policies they advocated in the past.

14

Policies

Voters

Parties

Figure 1.2: Voters, parties, and policies: a systemic relationship

The theoretical relationship between voters, parties and policies is depicted in the Figure 1.2. However, the links between parties, voters, and policies are far less mechanical than the Figure 1.2 suggests. Political parties do enjoy a certain degree of autonomy when choosing what policy platforms are to be presented in elections and voters do not always punish political parties for behaving unreliably. Two reasons are particularly important in this regards. First is that politics is full of ambiguities. Politics and particularly party competition do not usually offer voters choices in environments characterized by perfect information. The political parties often use abstract notions such as ideology, left, centre and right, in order to give voters hints about what policies would implement if getting into power. Only rarely political parties address concrete issues by pointing to what they would do once they got elected. Thus, voters are quite often let to guess what the concrete policies of governing parties will be, a fact which is an unfavourable condition for full accountability of political parties. The second reason, which relates to the ambiguity of politics and favours political parties to be less accountable, regards the imperfect information of individual voters. The voters do not know the positions of political parties on all issues that are relevant to political competition and they often need to make use of informational shortcuts to discriminate between policies advocated by political parties. This gives the political parties a degree of freedom in choosing their political platforms, although they still need to keep in line with the previous ideological or policy stances in order to maintain their political credibility.

15

1.3. The Organization of the Thesis

The thesis proceeds as following. In the Chapter 2, I introduce the basic assumptions of economic research on party competition, starting with a general argument about the specificity of the economic method and moving in the later sections to more specific arguments about the methodological individualism, economic rationality, and how can we apply these concepts to politics and party competition. The Chapter 3 extensively discusses the literature on spatial party competition, with regard to equilibrium outcomes in various political settings. The argument departs from the classical models of party competition introduced by Hotelling, Black and Downs and moves to more complex models of party competition: multidimensional models, models with probabilistic voting and models with imperfect information. It also includes a brief summery of formal models of multiparty competition under uncertainty which represents the core theoretical elements of the thesis. I then proceed to discuss in the Chapter 4 about the importance of information and institutions for political competition, particularly about how the parties react to political uncertainty. The Chapter 5 argues why analyzing party competition in the CEE from the point of view of the theoretical framework that I present in the earlier chapters is correct, pointing to the peculiarities of CEE politics. I also advance there three empirical hypotheses about what pattern of political competition I expect to occur in the CEE elections and I present the data and the methodology I use in the empirical analysis. The Chapter 6 outlines the national party systems of the four countries which constitute the empirical universe: the Czech Republic, Hungary, Poland and Romania. The empirical analysis of party dynamics in the CEE elections is offered to reader in the Chapter 7, where I show the degree of party collision in national elections and I offer support for the theoretical arguments presented in the Chapter 1 to 4. The Chapter 8 presents some possible implication of such pattern of political competition for the CEE societies and offers the concluding remarks.

16

CHAPTER 2: THE ECONOMIC BASES OF POLITICS: A PRELIMINARY VIEW

2.1. The Study of the Political Economy

Collective actors overwhelmingly predominate today in the study of politics. In our inquiries about social and political phenomena, we learn about class, ethnicity, parties, social groups, social movements and so forth, as the elementary units of our analytical approach. Individuals per se are still rare focus in political science despite that other disciplines from the social sciences (mainly microeconomics and psychology) have made important steps forward by using the assumptions of methodological individualism. One of the many difficulties our discipline comes across with is that it aims at explaining macro political processes and it sometimes disregards the micro-foundations of these political phenomena. By focusing on macro political processes, we usually concentrate only on the results of causal process, losing very much of the explanatory power we aim at in political science. Since almost every political phenomenon involves the actions of individuals, in the sense that interactions between individuals lead to certain social and political outcomes, we need to formulate and use theories that ask about why individuals behave the way they do in political contexts. If our aim as scientists is to explain why certain phenomena occur, then we need to pay considerable more attention to these micro-foundations that are of crucial importance for political action.3 One discipline in social sciences underwent an important methodological revolution decades ago. Economics, particularly microeconomics, begun asking about the goals of entrepreneurs and linked the economic behaviour with their presumed economic motivations. The economists argued that endogenous and not exogenous factors were the main determinants of the economic behaviour of firms and they began to pay attention to the goals of economic agents in competitive markets. In this respect, the concept of
3

By explaining a political phenomenon I understand subsuming this particular political phenomenon to a generalizing law. See in this respect W. Riker (1990).

17

revealed preferences of P. Samuelson (1938, 1948), which assumed that the behaviour and strategies of firms tell something important about their economic goals, represented a significant step forward in the economic analysis of firm behaviour. Scientifically, economics has achieved a respectable position within social sciences by using the methodological individualism and the assumption of intentional behaviour as fundamental bases of its methodological approach. Even though many of the economic models seem simplistic or implausible in their assumptions, the capacity of economics to explain has not been matched yet by any other discipline in social sciences. That economics represents a model for other disciplines in social sciences is already a matter largely accepted. Less appealing for other disciplines has been to borrow the instrument that has allowed the impressive scientific progress in economics: the economic method. Several arguments have been advanced to deny the usefulness of applying microeconomics methodology to the study of other social and political phenomena. One argument contests the possibility that social scientists, other than economists, could construct any kind of models for social and political interaction. Such refutation is based on the assumption that social or political interactions are much more complex than pure economic exchanges as it involves values, prejudices, social norms etc. These are generally not of great interest in economic theory but they become unavoidable for anthropologists, sociologists or political scientists. Thus, the complexity of social and political life can not be encapsulated meaningfully in the parsimonious models that are used in economics and any attempt of this sort would be irrelevant for understanding real social phenomena. Another argument against the economic method regards the process that allows economists to arrive at strong conclusions about economic behaviour. The economists use extensively deduction and formalization of the argument in their research. The deductive approach has been usually considered irrelevant for other social sciences for the same complexity that presumably characterizes the social interactions. Also, another argument against the use of economic method in social sciences points to its irrelevance for understanding real social and political phenomena. Economic models are usually highly formalized and theory-oriented, whereas other disciplines in social sciences are concerned with facts and empirical phenomena. The main argument of detractors of economic method suggests that, by using such method, we understand nothing or very little about our

18

societies or about how real people behave in various political contexts because we tend to focus on abstract individuals. For instance, Green and Shapiro (1994, 6-7) argue against rational choice models: [] We contend that much of the fanfare with which the rational choice approach has been heralded in political science must be seen as premature once the question is asked: What has this literature contributed to our understanding of politics? We do not dispute that theoretical models of immense and increasing sophistication have been produced by practitioners of rational choice theory, but in our view the case has yet to be made that these models have advanced our understanding of how politics works in the real world. To date, a large proportion of the theoretical conjectures of rational choice theorists have not been tested empirically. Those tests that have been undertaken have either failed on their own terns or garnered theoretical support for propositions that, on reflection, can only be characterized as banal: they do little more than restate existing knowledge in rational choice terminology.

Yet, in spite of these presumed drawbacks, the economic method has been extensively employed in the last decade in other social science disciplines, particularly in political science. Three reasons were particularly responsible for such academic colonisation. The first reason is that a many political scientists (together with colleagues from other disciplines) were dissatisfied with the explanatory capacity of other approaches and decided to look after available alternatives. Since the reputation of economics among other disciplines from social science has been unrivalled, it is obvious why the choice of some scholars has been the economic method. The second reason why the economic method has been extensively used in political science in the last decades is the need for theories of political behaviour that are internally coherent.4 Formalization and rigorous deduction have allowed economics to achieve such goal. By contrast, many theories in political science are based on scholars intuition and

Since our interest lies in political science I leave outside other social disciplines, although the argument is valid for them too.

19

lack of an internally coherent argument. Consequently, it is common in political science to find incomplete theories that leave aside important theoretical variables or specify incoherent causal mechanisms. A third and perhaps the most powerful reason is the integration of economics and political science in what has been called the political economy. For long time these disciplines were considered separate academic fields since economics was traditionally concerned with production and use of scarce goods, whereas political science inquired about distribution of power and authority within the society. Moreover, the academic segmentation of social sciences also favoured this traditional perspective about economics and political science, viewed as separate, fully distinctive academic disciplines. However, the political economists argue convincingly that at least parts of economics and political science can not be dissociated and they should be considered as an integrated field of study called the political economy. Political leaders take decisions that affect the economy and the economic behaviour of firms. At the same time, the economic behaviour of agents and the evolution of economy have significant political consequences.5 Firms often lobby at governmental level to obtain changes in legislation; governments often pass legislation pursuing particular economic goals and so forth. Moreover, economy and politics were not considered two separate academic disciplines until the 19th century. To break this symbiosis, the political economists argue, it means to miss a great deal of understanding in what regards the economic and political process (Alt and Shepsle 1990). I am largely sympathetic to this argument. There is not any meaningful reason why we should a priori divide disciplines according to the method of research. To large extent, this separation of disciplines is a matter of academic convention. However, rather than using the method to divide arbitrarily the disciplines in social sciences, we should better pay more attention to the field of studies and observe that different disciplines, conventionally regarded as autonomous or even independent, have something meaningfully to say about human behaviour in a particular contexts. Such arbitrary division is less
5

One assumption in voting behaviour is that the shape of economy is an important determinant of the support for political parties. Economic voting literature emphasizes that the support for incumbent political parties varies as a function of the evolution of the national or personal economy. As the economy improves (or develop well), the fortune of the incumbent political party or coalition increases, but when the economic situation deteriorates the incumbents are punished by the voters. For extensive summaries of economic voting literature, see Lewis-Beck 1988, Lewis-Beck and Paldam 2000, Lewis-Beck and Stegmaier 2000).

20

common in natural sciences. Take for instance mathematics, physics or astronomy. All these disciplines use similar methods and instruments to arrive at solid scientific conclusions, although there are significant differences regarding the objects of study among these disciplines. Yet, hardly any physicist, mathematician or astronomer would argue against mathematics as the principal method of research in her field of study. I will restrain myself embarking on a detailed argumentation in favour of economic method and how political science would benefit by using it. My argument in favour of the economic reasoning is rather implicit: I show in this thesis how using a similar argument could explain the political behaviour of politicians and political parties in the CEE. After all, the best argument in favour of a method is to show how it can substantially contribute to the development of our discipline.

2.2. The Methodological Individualism and the Theory of Choice

The typical argument of scholars who support the methodological individualism is that []the elementary unit of social life is the individual human action. To explain social institutions and social change is to show how they arise as the result of the action and interaction of individuals (Elster 1992). The main assumption is that only individuals act in social situations and the group decisions, institutions or other social forms that involve collective behaviour are only indirect results of the decisions and behaviour of individuals. To explain why a particular decision has been taken within a group, why institutions have been created to certain way or why an institution behaves the way, it does means to analyse the decision and behaviour of individuals of that group or institution. Groups and institutions are only abstract entities that do not possess a will of their own. Even when we are interested in the behaviour of such collective groups, we need to recall that the primary units of analysis are individuals and individual behaviour. How an individual acts and what motivates her actions are questions much more controversial than the conjectures of methodological individualism. Every individual has a large set of possible actions at her disposal. How then we can explain why she chooses a

21

particular action out of this large set of alternatives? Elster (1992, 13) offers an explanation which is based on two filtering mechanism. Constrains of any kind (physical, economic, psychological or legal) form the first filtering mechanism each individual faces when she decides about the future course of action. The actions that are not compatible with these constrains are eliminated as they form the set of unfeasible alternatives. The actions that are consistent with constrains form what we usually call the opportunity set. Suppose that an individual A has to decide about how she is going to spend the incoming week-end. Obviously, she almost always has several alternatives from which she may choose. She can decide to rest during the week-end, staying at home. She could also plan a trip to one of the resorts nearby her town or she could visit another foreign city, provided that she is rich enough. Imagine that the set of all alternatives at As disposal is SA={Stay at home; Visit the Carpathians; Visit the Black Sea; Visit Rome; Visit Buenos Aires; Visit Mars}. However, in this set of alternatives there may be some options that are not at all feasible. Take for instance the alternative Visit Buenos Aires. Such alternative is not feasible because presumably the trip needs some prior arrangements such as booking a flight ticket, hotel room or applying for consular visa. Neither the alternative Visit Mars is feasible because we do not have sufficient technological capabilities to transport a person to Mars and back yet. Thus, the feasible set of alternatives for the person A becomes SF[A]={Stay at home, Visit the Carpathians; Visit the Black Sea; Visit Rome}. However, imagine another person B who is less wealthy than A. Then, her set of feasible action does not include Visit Rome, because she could not afford a trip arranged in a short period of time. If she is in a state of extreme poverty, then she probably can afford neither alternatives Visit Carpathians nor Visit Black Sea, and thus her set of feasible alternatives is SF[B]={Stay at home}, and her course of action will reflect this alternative. Understanding the concept of feasible set of alternatives represents an important step forward, yet it is not enough to explain the behaviour (or choices) of individuals. Just because we know that the person A does not have as feasible alternatives Visit Buenos Aires and Visit Mars is not enough to predict what her choice among all alternatives

22

from the feasible set will be. We need additional criteria to help us understanding why she chooses a particular action against other alternatives. These criteria are: 1) the social norms, and 2) the rational choice, both forming the second filtering mechanism (Elster: 1992, 13). Take for instance the rules that regulate the urban traffic. Everybody knows that all cars should circulate on the right side of the road (left on UK or other countries from British Commonwealth), which could be regarded as a social norm, although it may have received a legal status by state regulation. Or, when a group of persons steps up and down on staircase, it usually happens that they choose the right side of staircase to avoid collision. Again, this is a social norm that imposes certain behaviour to individuals. Their behaviour is mostly spontaneous in the sense that it does not involve a cost-benefit analysis of the kind I am going to talk later on. Social norms ask us, for instance, to greet when we enter a room in which there are other persons, because it is socially expected to behave in this manner. In all these situations, our behaviour reflects not a choice made consciously by individuals, but an imposed (loosely defined) social norm. The rational choice deals with other types of actions. Imagine that our hypothetical individual A has to decide where to go this week-end. We established that her feasible set of alternatives is given by the set SF[A]={Stay at home; Visit Carpathians; Visit Black Sea; Visit Rome}. What is going to be her choice over these alternatives? In order to respond to this question, we obviously need to know something more about what the individual A generally prefers, otherwise any alternative from this feasible set could be her choice. Imagine then that we were able somehow to observe that in the last five week-ends our friend A had visited London, Paris, Wien, Prague and Budapest, although she had had in her opportunity set the alternatives Stay at home, Visit the Carpathians or Visit the Black Sea. It is reasonably then to suspect that A likes very much to travel abroad, perhaps enjoying very much the cultural and historical sites of the European cities. Thus, it is very likely that our individual A is going to choose Visit Rome instead of a trip to mountains or seaside resorts. The distinction between behaviours induced by social norms and by rational choice is marked, yet in many situations the distinction could be fuzzy. I argued that cars circulating only on right side of road is originally a social (and later a legal) norm. However, let us imagine that each driver has a choice between driving on the left or

23

respectively on the right side of the road. The driver knows that driving on the left side would incur a substantial cost that is associated with this alternative of action: the probability of a car accident is very high. Then her choice of driving on the right side does not reflect (only) the social norm; in this context it also represents a rational choice. Our interest lies on the cases when individuals rationally choose an action against other possible alternatives, so that we totally disregard the choices that are induced from the existing social norms. The word rationally has here a special meaning to which I will come back in the next section of the chapter. Of interest for the moment is whether there is any criterion that could help us understanding why individuals choose some alternatives against other alternatives. The good news is that such universal criterion exists; the bad news is that many find it unsatisfactory. We, as individuals, have passions and desires. Each individual acts because she aims at fulfilling these passions and desires. Some people prefer to invest time and money to become billionaires, although they earn in one year enough money to live a luxurious life for the rest of their days. Others prefer to become professors of political science, earning less and without any obvious possibility of becoming millionaires at all. Some prefer to pay hundred thousands of dollars for a painting of Picasso, whereas others would be happy to buy only an art album and save the rest for other commodities. Some people enjoy dining to the most expensive restaurants where they can drink a bottle of Champagne wine for several hundred dollars, whereas others would find lovely to go to the local pub where they can enjoy the cheap local beer and watch football games. The sequence could continue endlessly since the behaviours and choices of individuals are very diverse. However, what brings together all these diverse choices is that they reflect the different desires of individuals. We have seen that the two main components of an individual choice are the feasible opportunities and individuals desires or preferences. Or, as Elster (1992, 12) notes [], actions are explained by opportunities and desires by what people can do and by what they want to do. In this context, we need to understand that opportunities are external and objective to individual who can not modify their structure by unilateral action. No matter I do, I can not modify the set of feasible actions from which I have to choose. By contrary, the desires are subjective and internal to individual (Elster: 1992, 20). From this distinction

24

between the natures of opportunities and desires we can derive two main arguments. The first is that what is desirable for one individual is not necessary desirable for another. Thus, since desires vary from individual to individual, the choice of one individual can not be judged in relation with the goals the other individual pursues. The second argument is that individuals usually make mistakes about the real structure of opportunities so that some alternatives which are in the feasible set are let aside whereas other non-feasible alternatives may be considered as available. Thus, what should explain the behaviour of individuals from the perspective of a theory of choice are the desires of individual together with her beliefs about the structure of opportunities (Elster 1992, 14).

2.3. Rationality and Rational Behaviour

As I point out above, a theory of individual choice should think about the desires of individuals and their beliefs about the available opportunities. Certainly, every human being has desires, acts to fulfil those desires and acquires certain knowledge (perfect or imperfect) about the situation in which she takes action. Up to this point of our discussion, the argument about individual choice and behaviour is far too general and we have no useful yardstick to assess how individuals are going to choose in various situations. I have argued above that we are interested in analyzing those social and political situations in which individuals rationally choose a course of action against other possible alternatives. But what rationality and rational behaviour mean in this context? In daily language, the term rationality generally denotes either cognition or a normative imperative. It is common to hear people arguing that it is not rational to act without a proper thinking of what it should be done in that situation. A course of action should come after a careful consideration of all implications of that action, and when we fail to foresee them we usually say that our decisions were irrational. In this context, the term rationality reflects a cognitive exercise of an individual who should be able to predict all the consequences of her action. Thus, a rational behaviour is in this context a complex calculus about the consequences of our decisions.

25

The second meaning, which is to certain extent complementary to the first argument, suggests that the rational persons will choose those actions that are morally superior to others. This is the case of social norms. Take for instance the social norm that says we should keep our promises. If nobody respects the promises, then the institution of assurance would become ineffective and nobody would trust anybody (Miroiu 2006, 36). Such action is not congruent with the principle of morality because it breaks the Kantian principle of universalism (an action is moral if it can be recommended to any actor) and thus it can not be rational because it infringes the principle of morality.6 This meaning of rationality is often brought as an argument for the claim that the voting behaviour of the Romanians in the first elections was totally irrational as it infringed on the prospects of democratic consolidation and economic reforms necessary for the general prosperity. Since such purposes are morally superior to any other particular goals, such voting choice should be regarded as irrational. The first meaning of rationality points to how an individual reaches a decision and the second meaning asks about the goals of individuals. Both of them are very restrictive in the sense that they are impracticable when it comes to the analysis of individual behaviour. We do not know or very often it is impossible to determine how an individual reaches a decision. Also, we certainly do not want to infer that a course of action is morally superior to other course of action because this is merely a philosophical issue, which, though important for normative theory, is less helpful for analyzing human behaviour in social contexts. So we need other criteria for defining rationality in order to analyse the behaviour of individuals. There is a third sense of rationality that can be used to describe and analyse the individual behaviour. According to it, rationality represents the capacity of an individual to undertake those actions that lead her to the fulfilment of personal desires, provided that the object of her desire is in the feasible set of opportunities. According to this definition, the rationality does not refer to the purposes an individual pursues in political, economic or other social contexts or to the capacity of an individual to solve complex equations about the context in which she takes action. Instead, this notion of rationality refers to the process that links the desires (or goals) of individuals to their actions in political or economic
6

For a broader argument of Kantian principle of categorical imperative, see Miroiu (2006, 35-39).

26

circumstances. Rationality in this context is defined instrumentally and it refers to the actions of an individual and not to her purposes. In this regard, we say that the behaviour of an individual is rational if she is able to select those actions that lead to the fulfilment of her desires. Suppose now that we observe the behaviour of an individual A who would like to spend the Sunday evening by going to cinema. We know that individual As house is at the crossroads of three different streets that lead to cinema (street C), to theatre (street T), and to local football stadium (street S). The individual A has to select one of these three streets as a course of action, knowing the destinations of each of them. Since we know (or we infer on the bases of her past actions, for instance) that individual A wishes to watch a movie on the Sunday evening, we say that A acts rationally if she selects alternative C and irrationally if she selects T or S instead. The reader should observe that in our hypothetical example I do not infer which goal the individual A ought to pursue on moral bases and that her feasible set of opportunities is clearly identified as SF(I)={Cinema, Theatre, Football}. As choice in this context depends totally on how she orders these three opportunities on her subjective utility scale. We assume that each of these possible actions produces a certain benefit (or utility, in economics jargon), either positive or negative, to A and she is able to select among the alternatives according to the subjective utilities she attaches to {Cinema}, {Theatre}, and {Football}. There is one big danger that affects the theory of rational choice, for which the discipline has been criticized: the tautological or simplistic explanations. Obviously we wish to avoid inferring that all choices of individuals are rational because each individual has certain preferences and, thus, all her actions necessarily reflect her preferences. Just because an individual chooses a course of action, it is not enough to infer that she made a rational choice and that the goal pursued gave her higher utility than any other feasible alternative. Though explaining any kind of individual behaviour is tempting (after all, the substantial argument of the rational choice is the intentionality of individual behaviour)7, we need to be aware that such reckless argument about individual behaviour produces

We need to distinct between the intentional behaviour and the rational behaviour. Intention is an important component of rational behaviour; however, it is not sufficient to qualify a choice as rational.

27

theories with lower degree of falsifiability. For this purpose, the models of rational behaviour have to be complex enough in order to avoid such sheer tautology. How can we avoid tautological explanations? One answer resides in the internal consistency of rational choice theories. Instead of saying only that any choice with the property of maximizing the utility of an individual is a rational choice, we can formulate some basic requirements that any rational individual choice has to fulfil. These requirements have the property of constraining logically the individual preferences but, at the same time, we do not wish to impose very severe restrictions to individual behaviour, so that some preferences are a priori excluded. Suppose that our individual A chooses again the alternative {Cinema} out of the feasible set SF[I]={Cinema, Theatre, Football}. The feasible set S is non-empty, which means there is at least one course of action available to the individual A. What can we say about As choice? We obviously could say that the individual A chooses the alternative C over the alternatives T and F. Or, in a more rational choice language, that the individual I prefers C to T and F (or C is preferred to any other alternative from the feasible set). We introduce here another crucial element of rational choice theory, which is the concept of preference. This concept is important for the theory because it allows us to establish a certain hierarchy between the alternatives from the feasible set. The choice of an individual reflects her preference relation between the alternatives: the preferred alternatives produce higher subjective utilities than less preferred alternatives. We can say, for instance, that the individual A strictly prefers C to T (CPAT) or C to F (CPAF) or that the individual A weakly prefers C to T (CRAT) or C to F (CRAF), where P and R denote strict or weak preference relationships between two alternatives. When an individual has a preference relation of the type [(CRAT) & (TRAC)], it means that she is indifferent between the two alternatives C and T, which we formally write as (CIAT).8 I have mentioned above that we need to impose certain constrains over the individual preference relation in order to qualify a choice as rational. The very first constrain of any rational preference is the reflexivity. Mathematically, we write this relation as (xRix), which says that a certain good x is at least as good as itself. Sen (1970) calls this

The strict preference relation between C and T (CPiT) may also be written as: C>T. The weak preference relation between C and T (CRiT) can be written as: CT.

28

the condition of mental sanity as it requires an individual i to choose the object j as long as what the individual i looks for is an object that has the same properties as j. Take for instance the example of our individual A who wishes to spend the Sunday watching a movie at the cinema. The condition of reflexivity asks that if the As goal is to go to the cinema, then, ceteris paribus, any cinema is at least as good as another cinema. If the individual A wishes to watch a movie and eat some pop-corn at the cinema, then any cinema that sells pop-corn to customers is at least as good as any other cinema with the same property of selling pop-corn to customers. Second, to formulate a preference of the type I prefer Cinema to Theatre implies that I am somehow able to compare the two alternatives, {Cinema} and {Theatre}. To compare any two or more alternatives means that I am able to order these alternatives in a manner that reflects strong or weak preference or indifference among alternatives. Thus, the second condition any preference has to satisfy is the condition of completeness (sometimes called comparability). The comparability means that, given any two alternatives C and T, we can say that either (CRiT) or (TRiC) or (CIiT). The third condition regards the logic of ordering (or of comparability) of individual preferences. This condition asks that any ordering of three or more alternatives available in the feasible set should be transitive, so that an individual i who prefers C to T and T to F prefers necessarily C to F. Formally, we write this condition of transitivity as:
c, t , f : [c, t , f ] S (cRt ) (tRf ) (cRf ) ,

which reads that whatever three alternatives c, t, and f included in the non-empty set of feasible alternatives, if c is weakly preferred to t that is weakly preferred to f, then c is weakly preferred to f. The relation of weak preference can be changed to reflect a strong preference, but the condition of transitivity can not be abandoned if we talk about a rational decision. Transitivity is a very important condition of rationality. Imagine that one individual has a preference ordering that is not transitive, so that she prefers C to T, T to F, but F to C. Having such order of preferences, this individual will not be able to select a course of action because no matter what alternative she chooses, there will be always another alternative preferred to the first alternative. Such distribution of preferences is called as cyclical ordering of preferences.

29

Now, we are better equipped to define rationality and rational behaviour in the theory of rational choice. We call that an individual i acts (chooses or behaves) rationally if she is able to order the set of feasible alternatives according to her personal tastes or values in a manner that respects the conditions of reflexivity, completeness and transitivity, and after ordering the set of alternatives, she chooses from top of this ordering. Observe that this definition of rationality and rational behaviour encloses two major elements. The first element is the internally consistent ordering of feasible alternatives so that it does not hold logical faults like the cyclical ordering of preferences. Then, once an individual has been able to define and arrange her preferences without violating the conditions of reflexivity, completeness and transitivity, the second element comes into place: the individual will choose according to the maximization principle.

2.4. Rational Politicians and Political Action

A detractor of rational choice theory would fiercely contest that such definition of rationality and the subsequent theory of action are relevant for the study of politics. Her arguments would probably point to the distinction between politics and other daily activities or between the study of politics and other academic disciplines such as economics. Contrary to economics where the goals of firms are very much evident and the struggles of firms for higher profits raise little debate, the goals of political actors are considered to be far more diverse. As I mention above, the typical argument about the peculiarity of politics that allegedly makes irrelevant the rational choices is that values play a very important role in politics. Because the behaviour of individuals in politics is very much influenced by political and social values, a theory of political action that concentrates on the rationality of individuals is going to fail at explaining the political outcomes. Any theory of political behaviour that aims at explaining political phenomena, a detractor of rational choice theory would argue, has to take into account not only the role of political, social and cultural values, but also the diversity of these values in the society. For these reasons, a theory of rational choice in politics is meaningless.

30

I do believe that such argument has serious flaws. Values affect the behaviour of individuals in various other situations, not only in politics. The physicist that studies the origins of the universe has her own values which might be undermined by the results of her scientific research. Nor biologists are not value-free scientists. And yet, there is very much a non-sense to argue that the approach of these scientists should take into account their own values, because it affects the way they describe or explain physical or biological phenomena. Economics is one example of social sciences in which the role of values is quite significant. Welfare economics, for instance, extensively studied in the last fifty years the role of values and other normative issues. Thus, there is no a priori reason to believe that values and norms particularly influence the study of politics while they do not have an effect on other disciplines from social or natural sciences. Another flaw of the argument against rational choice theory is that the notion of rationality is often applied to goals rather than to the process through which an individual attains such goals. To argue that values and norms influence the behaviour of individuals refers, as it seems to me, to the definition of goals and preferences of individuals rather than to the way an individual behaves when she has to choose among different alternatives. Rational choice theory acknowledges that values, norms and tastes have an enormous influence on the formation of individual preferences, and thus different goals result from the diversity of values, norms and tastes of individuals. Furthermore, we acknowledge that some situations are difficult to be framed in models of rational choice or when framed, it leads to tautological explanations. We would not want to explain, for instance, why the terrorists from 9/11 acted the way they did by framing the situation into a rational choice model, because their behaviour complied rather with a social norm than reflected a rational decision. Many rational choice scholars would argue that if the terrorists had a preference relation of the type I discussed above, it would violate the condition of completeness. It is difficult, if not impossible, to argue that the terrorists meaningfully compared the possible alternatives and reached the decision after such comparison. Anyhow, probably models of rational choice bring no further knowledge about the behaviour of individuals when it comes at explaining suicidal or other similar behaviours. But when it comes to situations in which individuals choose without the

31

obvious constrains of social norms, the rational choice theory is a very powerful instrument at our disposal for explaining social or political outcomes. In politics, the rational choice theory treats politicians as being rational political actors. As any other individuals, politicians have their own tastes and desires and we assume in our research that their political behaviour is meant to fulfil such desires. This assumption does carry neither positive, nor negative normative content, usually common in the appraisal of politicians. Contrary to the popular evaluation of how politicians behave, the rational choice theory does not simply regard politicians as pure selfish individuals that are interested only in their own wealth. On the other hand, nor it does argue that the role of politicians is to seek and provide common goods, as normative theories of politics often imply. Instead, it takes a more positive (or neutral) position in this respect, assuming that a politician behaves as any other rational individual would do in political or social context. Thus, any politician is able to observe what is her feasible set of options in a given political situation, to arrange the alternatives from the feasible set according to the utilities they provide with so that the preference ordering is not cyclical, and finally, to select the alternative that lays at the top of the preference ordering. The goals of politicians may also differ from situation to situation. A politician running for a public office in competition with other candidates wishes to win the election, and we say that she acts rationally when she takes those actions that would lead her at winning. She may have personal stakes in running for such position because public offices offer private goods to the politicians who hold those positions. Or, by contrary, she may perceive such office as an instrument for having an impact on politics.9 A politician in the legislature who is interested to pass legislation in a certain domain will lobby at other colleagues, both from the government and opposition, in order to receive political support for the bill. She may have a genuine interest in passing this legislation or she may want to exploit the issue in the future political campaigns, claiming that she was successful in providing her voters with such public good. When member of Cabinet, a politician would probably struggle to receive a greater budget for the ministry she runs because having access to more resources provides more political influence. All the above examples point to

For an extensive argument about the distinction between private and public goods see Olson (1971). For a useful synthesis of the argument, see Miroiu (2007, 117-64).

32

how we expect politicians to behave in different hypothetical situations. If we observe them behaving differently, we may conclude that either their behaviour is irrational (they do not act to accomplish the highest preference) or our model about the individual behaviour in that situation is based on incomplete or inappropriate assumptions. Thus, it is very important to specify carefully the assumptions we have about the goals of individuals in political situations so that our models are not based on implausible postulations about the individual behaviour.

2.5. Two Views on Political Competition: Sociological and Economic Models

The social and political conflict fuelled the debate about the role of political parties in modern democracies for a long period of time. Initially seen as menaces for the republican government because they were conspicuously supposed to function in a manner that spread the political dissent inside the political system, political parties became institutionalized as essential elements of democratic regime once the conflict and the potential for dissent were recognized as essential pillars of the modern political democracies (Dahl 1971, 1989).10 The electoral competition became one of the main institutional guarantees of the modern political democracies and the political parties have been the key political actors that participated in the electoral game since the mid-XIX century. Elections are essential institutions through which a democratic government is formed. They decide what party is going to form the Cabinet in majoritarian democracies or what parties win parliamentary representation and negotiate the government formation in the multiparty systems. They allow leaders to win offices and implement certain policy programs that are important to party activists and voters. Also, they represent an important mechanism of political accountability: the political parties need to maintain their electoral support if they want to have a future in politics. Unreliable political parties, which do not
10

Sartori (1976) offers a comprehensive analysis of the evolution of political parties from loose political organizations (proto-parties in Sartoris terminology) to fully institutionalized political organizations and of the relationship between political parties, party government and the acceptance of the conflict in the modern democracies.

33

govern the way they promised and which are unable to deliver the political goods they pledged, are going to be rejected in future elections by dissatisfied voters. A crucial democratic ideal, that is the responsibility of political elite to the citizens, is set into practice through elections. Though few deny the importance of elections in the modern democracies, the political theory offers a large disagreement concerning the consequences of and the philosophy behind the elections. One school of thought, which I call it the sociological school or model, highlights the relationship between political parties, social groups and social conflicts in the modern democracies. In this interpretation of politics, the elections should give voters a meaningful choice between political parties that represent different social groups and have different ideological orientations. Such argument is based on the normative argument that the government should be responsive toward the society in general and toward different social interests in particular. Responsive parties are linked to responsive government, that is a government [] attentive to, and influenced by, the voice of the people (Sartori 1976, 20). The second school of thought which I call the economic school or model ignores the relationship between political parties and particular social groups and concentrates on the individuals per se. Elections are essentially methods of selecting a government, not instruments of substantial political representation. Election times are periods when political leaders are interested in getting votes and not offering meaningful choices to voters.11 I offer below a short description of how the two logics of explanation theorize about the party development and political competition.

11

My classification intends not to comprehensively include all main approaches to electoral competition but to summarize two different conceptions about the dynamics of electoral competition which I detail below. Recall that in the Chapter 1, I have shown that these models are far from being mutually exclusive and that today it is customary to combine these approaches in unified explanations. Thus, a summery of the literature using these two main approaches is obviously incomplete and to a large extent it suffers of arbitrariness. Rather, this sharp differentiation of the approaches intends to clarify to a reader unaccustomed with the rational choice how an economic model of party competition is set up, in contrast with the structure of sociological explanation. The former highlights the economic (cost-benefit) analysis for vote choice and the rational response of political parties, whereas the later emphasizes the importance of social links and social conflict in electoral arena.

34

2.5.1 Parties, Social Revolutions and Institutional Context: A Short Introspection of the Sociological Model(s) Much of what we know about emergence and behaviour of political parties comes from the studies of political parties that use sociological paradigms. The sociological models assume that the society or the social groups are able to define common political values. Building on the presumed relationship between the social groups and political actors, sociological models suggest that the strong religious commitment makes a Catholic to regard herself as being part of a social group which values the Christian principles. This religious commitment or any other politically relevant social characteristic are very likely to influence her political choices. Equally, a blue-collar worker has the capacity of identifying her social and political affiliation and she acts politically in accord with those social predispositions. As a part of a socially deprived group, and conscious of her social identity, a blue-collar will vote for one of the parties that reinforce the sense of belonging to her social community. Lipset and Rokkans (1967a) seminal work provides the most famous example of sociological explanation of party development. In their explanation of genesis and development of the West European party systems, Lipset and Rokkan bestow a central role to the issue of social conflict. The authors argue that political parties, especially at the beginning of the competitive politics in the Western democracies, underwent two important social functions: the integration of local communities in the national politics and the mobilization of voters in contested elections, especially after the full extension of voting rights (Lipset and Rokkan 1967b, 4-5). Political parties helped strengthening the national identity in the Western Europe by organizing cross-local channels of communication and played an essential role in the institutionalization of national political systems. Also, they encouraged voters to distinguish between their loyalty towards the political system and their own attitudes to politicians who competed for the set of public offices (Lipset and Rokkan 1967b, 5). However, the political parties are not simply agents of integration and mobilization of social strata in the national political arena. They reflect social conflicts, but at the same time they reflect different, even antagonistic, social and political values. Lipset and

35

Rokkans (1967b, 5) definition of political parties as alliances in conflicts over policies and value commitments within the larger body politic shows the intrinsic relationship between parties on the one hand, and the social groups, social conflicts and political values on the other hand. But perhaps the best way to exemplify the role of this relation in the sociological paradigm is to offer the original Lipset and Rokkans description of social and political functions that political parties perform in modern democracies:

For the sociologist, parties exert a double fascination. They help to crystallize and make explicit the conflicting interests, the latent strains and contrasts in the existing social structure, and they force subjects and citizens to ally themselves across structural cleavage lines and to set up priorities among their commitments to established or prospective roles in the system. Parties have an expressive function; they develop a rhetoric for the translation of contrasts in the social and cultural structure into demands and pressures for action and inaction. But they also have instrumental and representative functions: they force the spokesmen for the many contrasting interests and outlooks to strike bargains, to stagger demands, and to aggregate pressures (Lipset and Rokkan 1967b, 5).12

Building on Talcott Parsons classification of the functions of a social system, Lipset and Rokkan argue that the social cleavages in the Western Europe can be arranged in a two-dimensional space generated by a territorial-cultural and a functional dimensions. On the territorial-cultural dimension, there are oppositions between local communities trying to resist to the dominant national elite and bureaucracy, and a competition within the national elite for the control of the political system as a whole. On the functional dimension, there are oppositions that cut across the territory of nation-state over the allocation of wealth and goods within the political systems and over the value commitments (such as religious or ideological movements). The emergence of the Western European political parties and party systems resulted from two major transformations that occurred in the European political systems between
12

Italics in the original.

36

the 17th and 19th centuries. The first of these transformations, named revolutions by Lipset and Rokkan, constituted the downfall of the European empires and the construction of a series of national states in Europe. The national revolution (mainly during the 17th and 18th century) made that many local communities which previously had enjoyed a large autonomy became under the strict control of the centralized power. Other institutions, such as the Catholic Church, that traditionally were autonomous of the secular power also became subjects of the same centralizing effects produced by the national revolutions in Europe. Thus, Lipset and Rokkan argue that the national revolutions produced two major lines of conflicts within the European societies. The first line of conflict opposed the centralizing state to a series of local communities with strong cultural identities, or between the centre and the cultural periphery of the new political systems. On the second line of conflict, newly born national states tried to impose their authority over institutions like the Catholic Church, generating a strong conflict within the society between the supporters of secular state and those with religious commitments. The second transformation that shaped the space of European politics was the industrial revolution, between the 18th-19th centuries. In a first phase of this transformation, the technological advances made the agricultural-based economy become under the pressure of new developing industries. More people occupied the new jobs in the industrial sectors and more land that was previously an important resource for agriculture became the locus where the new industries developed. Thus, a line of conflict between the supporters of the old and new economies emerged within the western societies. However, once the new industries developed and many people were employed in industrial sector, another conflict line emerged (19th - 20th centuries) between the owners and workers of these industries. The Figure 2.1 depicts the basic relationship between the social transformations (the national and industrial revolutions) and the lines of conflict they generate within the European political systems. Lipset and Rokkan provided a framework suitable for analysing the pattern of competition that characterized the Old Politics: a political competition between two antagonistic Left and Right. The empirical research emphasized that the class cleavage, which is a concept linked to Lipset and Rokkan analysis, has been the most important determinant of voting behaviour in electoral competition during the most part of the 20th

37

century. The parties and the social groups of the traditional Left (blue collars at the social level, the Communist, Socialist or Social Democratic parties at the political level) competed against the social and political traditional right (industrial owners and middle class, at the social level, and the Conservatives at the political layer) for policies of economic redistribution. Also, empirical studies show that religious denomination has been another important element that has shaped the party competition in the Western Europe. The disagreements over religion divided the west European societies between the supporters of the Church and supporters of secularism. The first groups (Protestants and Catholics) have constantly voted for political parties with pro-Church political appeals such as the Christian Democrats, whereas the secularists appealed to the support of the anti-Church parties, mainly the Socialists and Social democrats.

L a n d O w n e r s s

I n d u s t r i a l O w ne r s

The Industrial Revolution

The National Revolution

Center Secular power I n d u s tr i a l O w b e r s W o r k e r s

Periphery Religious power

Figure 2.1: Social Revolutions and the lines of conflict in European societies

38

This classical approach links parties and voters who form the respective social groups in a very deterministic way. One evident shortcoming of the classical approach has been the impossibility to explain the short-term political change occurring from one election to another, given that the social structure remains theoretically unchanged. There is also empirical evidence which argues that the Old Politics approach is largely obsolete: we are noticing a decline of the importance of class voting in the western democracies, particularly after 60s, and a decline in the cohesion of traditional social groups (Franklin 1985, Franklin et al 1992, Dalton 2002). In newer sociological approach, the scholars from the University of Michigan formulated a more complex explanation for party-voter links based on a combination of sociological and psychological factors. They formulated a funnel of causality in order to explain the voting process. Social divisions and economic conditions are the first factors that affect the vote choice by influencing the group loyalty and value orientation of voters (the second layer of the funnel). Group loyalty and value orientation (defined mainly through the process of political socialization) determine the voters party attachment, which represents the key element when judging the political candidates and issue orientation. At the end of this funnel of causality, candidate image and value orientation (determined by the partisan affiliation) impact on the voters choice (Campbell et al. 1960). All these sociological explanations of party choice and voting behaviour have in common one important feature: the inextricable relationship between the individuals, the social groups to which they belong, and the political parties that appeal to individuals based on their group identities. This symbiosis has both descriptive and normative connotations as political parties reflect the existing social conflicts and behave as agents of political representation in political competition.

2.5.2 Economic Models of Political Parties and Party Competition Contrary to sociological (behavioural) approach, which highlights the role played by the social interests and values in shaping the behaviour of voters and political parties in competition, the rational choice models focus on the decision-making process of

39

individuals. The methodological individualism that rests at the heart of this intellectual tradition pays no attention to concepts like social groups, social interests or social values. This ignorance is well-intended: since the individual is the only agent who acts in various social and political contexts, concepts like those enumerated above are meaningless for any social or political theory, the rational choice theorists would argue. Thus, we need to concentrate our research on the actions of individuals and look for potential explanations in the preference set they have. At the heart of economic approach lies the notion of model. A model represents a theoretical construct, only that contrary to a theory, it is less developed. We can define a model as [] a collection of simplifying assumptions about the reality (Miroiu: 2006, 31), which is meant to reduce the complexity of political phenomena so that the researcher could concentrate and analyze only the relevant theoretical aspects. Models are stylizations of the real world and they are theoretically useful because phenomena of real world are often impossible to be understood directly. As Shepsle and Bonchek (1997, 9) point: [We] depend of a stylized model to provide us with insights and guidance that will shape analysis of these events. A useful model for understanding social and political phenomena should include three major components. The first component consists of the assumptions we have about the preference of individuals and about the mechanisms of decision making. For instance, we consider that ceteris paribus a buyer prefers to buy a certain good at the lowest price possible. A voter is likely to vote for a political party that advocates a policy program which resembles her policy expectations. A national state advocate for a barrier-free world market if its national entrepreneurs are likely to benefit from the inexistence of exchange taxes; otherwise, it will impose high duties on imports in order to protect the national firms. A second component of the model represents the implications of the assumptions we have about the rational behaviour of individuals. The analysis implies a careful consideration of individual preference and of decision making process. Consider the hypothetical social situation in which two individuals (suspected of committing a major crime) have to decide individually what will be their behaviour when confronting a law-

40

enforcing agent.13 Assuming that each individual is interested in maximization of her payoffs (or minimization of the loss, if winning is not possible), then each person has a dominant strategy in admitting her participation in that crime. The expected social result is that both rational individuals confess, so that the group formed by these two suspects ends in the worse social situation.14 The third component of the model is the conclusions. Given the basic assumptions we have about the rational behaviour in that situation and about the mechanism of decision process, what can we generally say about the circumstances like the ones we just modelled? In the case of Prisoners Dilemma, one general conclusion is that rational decisions of the individuals do not necessarily lead to efficient social situations. If groups wish to improve their social situations, they need to design efficient institutions which constrain the rational individuals to play cooperative strategies (Hardin: 1968, Olson: 1965). However, modelling the behaviour of politicians and political parties requires more concise definitions. One legitimate question is what is the central unit in the analysis of political competition? Is the behaviour of individual politician? Is the behaviour of the political party? If the answer to this question is the later, how can we solve the problems raised by our assumption of methodological individualism? Answering these questions requires a careful definition of what a political party is. Definitions of this type are legion, each of them pointing to features that seem defining: ideology, territorial organization, representation of social groups and so forth. Yet, a useful definition from a rational choice perspective needs to: 1) specify the problem of methodological individualism, and 2) approach the issue of rationality of political agents.

13

This situation is a well-known non-cooperative game called the Prisoners Dilemma that fascinated the community of game theorists for decades. The narrative of this situation is the following: two suspects of committing a major crime are interrogated by police in separate cells, without having the possibility to discuss about whether to confess or remain silent. The police tell each of the suspects the following: We know that you and your accomplice did this crime, but we do not have enough hard evidence to convict you, unless at least one of you confesses. But, nevertheless, we have enough evidence to put both of you in prison for two years for a minor crime (possessing illegally a gun, for instance) if none of you confesses. If one confesses while the other keeps silence, then the first is set free for cooperating with us and the former is convicted alone for the crime to ten years in prison. If you both confess, then each of you will be convicted of six years in prison. 14 Since the Prisoners Dilemma is the best known game, I will not commit myself to a deeper explanation of the game, which is available in almost any basic text on rational choice, social choice or game theory. See for instance Shepsle and Bonchek (1997, 202-204).

41

Downs offers a definition of political party that deals with these two conceptual and methodological problems; therefore there is no need for innovation in this regard. He argues that [] a political party is a team of men seeking to control the governing apparatus by gaining office in a dully constituted election (Downs: 1957, 25).15 The aspect of rationality in this definition is given by the goal of the politicians who seek the control over the public offices through the means of the elections. According to the definition, rational politicians will adapt their political behaviour to maximize the chances of winning the elections. The requirements of methodological individualism may seem problematic though. Downs assumes that every politician who is member of a political party derives some utility from winning of the election (and, subsequently, of the public offices), and thus, every politician from the party prefers winning to losing in elections. As members of a team, the politicians agree on all their political goals and, hence, each member of a political party has the same preferences as every other member (Downs: 1957, 25). Thus, each political party can be assumed of having a single, consistent preference ordering in what regards the political goals of the members of the respective party (Downs: 1957, 26). Downs assumptions do not come without problems, of which two are particularly relevant for formal political competition. The first discussion is about the goals of political leaders and the congruency between the desires of leaders and aspirations of lower political activists. The definition ignores the role of political activists within political parties, at least in what regards the possible conflicting goals between party elite and ordinary party members. One critic would probably argue, not without reason, that even ordinary party members may have enough political means to constrain the party leaders to define political objectives that are valuable not only for the party elite.16 Party members offer important
15

The two definitions of a political party Downs (1957, 24-25) offers, party as a coalition of men and party as a team of men originally belong to Jacob Marschak (1954), yet the posterity gave to Downs the merits of these definitions, although a proper citation can be found in Downs text. 16 We often assume that ordinary members of the parties are policy-oriented and they do not extract utilities from any public offices hold by their own party, unless the party, by using the offices it holds, is able to impose policies that are consonant with their policy objectives. On the other hand, the party leaders who hold public political offices extract utilities not only from the policies they implement when in office, but also from the simple fact that they occupy that public office. The implication of these assumptions is of great importance for the theory of political competition as one of the central questions raised in the theory refers to the goals of political parties. Do political parties value offices instrumentally, as means to attain some policy objectives, or they are goods per se for political parties, and therefore, policy programs are instruments to obtain public offices? is the question that troubled generations of rational choice theorists. Downs (1957)

42

resources to party organization, from membership fees to voluntary activities, and political leaders can not ignore the policy goals of lower rank members. Why then to formulate a very elitist definition of political parties and concentrate exclusively on the goals of political leaders? In this thesis I concentrate on the desires of political leaders, ignoring the complex relationship between the leaders and activists for defining the goals of political parties. Implicitly, I assume that leaders have the power or ability to impose their own political goals to the party organization. Why is this assumption reasonable? On theoretical grounds, one could offer two main reasons for why to construct a theory of party competition on the assumption that leaders political goals prevail. The first reason is the asymmetry of political power between leaders and political activists. Holding the key positions inside the party organization, the party leaders are better equipped to impose their political will over the ordinary members than vice versa. When defining the political objectives of their political party, party elite have numerous ways to persuade and even compel the party organization to accept some political strategies. Not exceptionally, some political goals of the political party are exclusively defined by the party elite, without prior consultation with ordinary members. The second reason is the asymmetry of information between political leaders and ordinary members. Political leaders are generally sophisticated individuals when it comes to political activity and political process. They also hold key positions from which they can filter the information that is later available for the public (ordinary party members included). They often place a sense of ambiguity in their political messages, which does not always permit the ordinary members to judge what the real political position of the party is. This asymmetry of information between party elite and party members works in the very same way like the asymmetry of positions inside an economic firm. The management and the regular employees, not mentioning the shareholders, have different sets of information. The shareholders have a very general view about the firm: they judge the profitability of the firm according to the amount they receive as dividends. An employee has an in-depth

argues that offices are the primarily goals of political parties, a point which I endorse in the Chapter 3 with arguments from the theory of collective action. By contrary, other authors disagree and argue offices are only instruments for policy implementation (Witman 1973).

43

knowledge about her own department. A top manager, though, has the most comprehensive view about the firm activities, profitability, future developments and so forth. Coming back to Downs definition of political party, another questioning assumption is that the party elite function as a team. By team (which is contrasted in the original text with the notion of coalition), Downs suggests that each member of the party leadership has the same order of preferences as any other member, and thus, the party does not have to struggle internally for defining its own political strategy. Moreover, no member of the team extracts a negative utility from the fact that her party wins the election, even though she may not have directly access to private benefits resulting from the public offices. Of course, we acknowledge that empirically, loosing an election does not always bring a negative utility to all members of the party leadership. While the top members of the elite are severely damaged when the party losses the election, as often they need to step down from the positions they hold, for middle rank party leaders this could be an opportunity that would boost their intra-party political career. Although reasonable to some extent, such counter-argument can not be incorporated into a model of inter-party competition. Models, as defined above, are crude simplifications of the reality, and no scientific model could account for all possible empirical situations. Intra-party competition is indeed an important variable for analyzing the political behaviour of political parties, but my main focus remains the inter-party electoral competition. On the other hand there are good reasons why Downs definition seems reasonable. Winning elections provides political goods not only to one politician, even when winning means getting a parliamentary seat. Other political benefits, such as non-elective offices in institutions, are available for other party members too, who then may have stakes in winning the election. Even when winning means that an individual candidate of the party obtains an office in a constituency, this situation is still one of Pareto-superiority for the respective party.

44

2.6. How the Voters Decide?

These assumptions about the behaviour of political parties in electoral competition are necessary but still not sufficient when working with economic models of party competition. The second set of actors about whom we need to make some assumptions is the electorate. In general, rational choice theory regards voters utility as a function of two variables. The first variable is whether the candidate or the party whom voter i votes for wins the elections, case in which the voter i has a positive utility. Conversely, the voter obtains a negative utility when the party for which she votes losses the election. The second variable is the policy utility the voter extracts from the policies of the winning candidate.17 Since there is always uncertainty about whether the winning political party or parties really implements the policies promised during the electoral campaign, it is more accurate to talk about the expected policy utility and to presume that once elected a political party acts responsibly.18 At the same time, Downs (1957, 42) argues convincingly that rational individuals are not interested in policies per se, but in their own utility incomes. Thus, rational voters may decide to vote for a change if they regard their present utility income as being very low comparing with any other possible future situation. The spatial models of electoral party competition usually regard the voters utility function as dependent of the utility income a voter could extract if her favourite political party wins the election.19 We usually employ in our models five major assumptions about the individual voter:
There is also a third variable: the probability of voter i to make a difference in election. Research taking into account this variable argues that a voter extracts utility from the act of voting only if her vote is decisive, that is the vote which makes the difference between winning and losing. Since both the subjective and objective probabilities that an individual impacts on the results of mass elections are slim, the rational choice theorists explain why a good part of voters do not vote in mass elections by arguing that voting in mass elections is irrational. Voting in committees, however, differs from voting in mass elections, since the objective probability of affecting the committees decision is much higher, and thus, we rarely see people abstaining in such context. 18 Newer developments in the spatial theories of electoral party competition argue that voters discount the possibility of political parties to fully implement their electoral promises (Adams, Merrill, and Grofman 2005). Discounting the probability of fully implementing the electoral policy platform drags the political parties away from the political convergence towards the position of their political constituencies. Models with policy discounting are discussed in the Chapter 3. 19 Recent scholarly research has attempted to include into the voters utility function not only the policy distance between the voter and political party, but also other non-policy aspects, such as partisan loyalties, socio-demographic traits, retrospective or prospective evaluation of party performance and so forth (see Adams: 2001; Adams, Merrill III, and Grofman 2005). In the next chapters, I offer an explanation why I
17

45

a.

Each voter has an ideal policy point. Interpreting this assumption in Downsean terms, the ideal policy position is the point where the voters policy utility function is maximized. Voters know the positions of political parties in political space.20 Each voter votes for the party that promises to implement a policy that is the closest to her ideal policy point.

b. c.

d.

Each voter has symmetric policy preferences. This condition suggests that two policy alternatives located at equal distance of voters ideal policy position offer the same utility income to that voter. The later is indifferent between these two positions.21 Every voter votes sincerely.22

e.

The assumptions a) and b) point out that voters and parties have a common view about the political space. The assumption e) asks voters to vote on the bases of their preference ordering so that political choice reflects the ideological proximity and not other strategic issues. The assumption c) and d) need however a much more substantive discussion. That voters have a greater utility when voting for political parties that advocate policies which are more similar to voters ideal policy positions seems to be an intuitive statement in politics, although, as we shall see in the Chapter 3, the directional theory of voting challenges this view. But an important question regarding voters utility function is what form does it take? Often the literature relies on the assumption that voters utility function takes the form of the quadratic utility function: Ui(k) = - (xi sk)2,

(1)

regard such models inadequate for party competition analysis in the Central and Eastern Europe, and implicitly, why I favour models that incorporate only voters policy motivations. 20 Some formal research, however, alter this assumption in order to see what is the effect of political uncertainty on political choice and on parties electoral strategies. 21 I show in the Chapter 3 that some spatial models of party competition work with assumptions that differ substantially from the proximity and symmetry principles. 22 Sincere voting means that every voter votes for the party that proposes a policy which is the closest to voters ideal policy point. Strategic voting, which breaks this rule of voting, is a widespread political phenomenon, documented particularly in committee voting. It consists of voting for the alternatives that are not in the top of the preference set if those alternatives have no chances of winning.

46

where Ui(k) means the utility of voter i when voting for party k, xi is the ideal position of voter i, and sk is the advocated position of party k on that ideological dimension. The alternative to the quadratic utility function is the linear utility function: Ui(k) = -(xi sk), (2)

where Ui(k) means the utility of voter i when voting for party k, xi is the ideal position of voter i, and sk is the advocated position of party k on that ideological dimension, and expression -(xi sk) means the negative absolute distance between voter i and party k policy positions. There is no strong empirical evidence, however, that voter utility function follows quadratic or linear pattern and, thus, it is difficult to argue substantially in favour of one or another. The quadratic utility function is more common in the literature because it is more convenient mathematically for the most of applications (it is symmetric around, it has a unique maximum at the individuals ideal point etc).23 However, the decision to rely on one or another formula is rather arbitrary.

2.7. A Crude Model of Party Competition with Downsean Parties and Voters
Summing up, the condition of rationality asks politicians to formulate policies that maximize their electoral support, that is to obtain the support of the largest possible number of voters given the strategies of other political parties. It also requires voters to vote for the political party that offers a policy which resembles their ideal policy positions. However, the spatial models of party competition build upon another crucial assumption that integrates all other assumptions we have formulated above about parties and voters: the politics can be interpreted geometrically. This assumption is crucial for understanding the interpretation of political competition is spatial terms. We assume that the political space
23

Some research reported that when finite policy scales (1-7 or 1-10 points) or thermometer scores are used, linear utility gives better results. However, two problems arise. The first is that is dubious to interpret the results derived from such scales as utilities. The second problem is that although the classical view is that utility scales can not be constrained, those scales are finite and thus constrain the positions of respondents. For more, see Erikson and Romero (1990), Alvarez and Nagel (1998), Adams, Merrill III and Grofman (2005, especially footnote 1, pg. 17).

47

consists of N-policy dimensions, where N1, which are important for political competition. On these N-policy dimensions, each voter has an ideal policy position or a vector of ideal policy positions if N>1. Observing the distribution of voters in policy space, political parties take on those policy positions that they expect to obtain the support of the largest possible number of voters. At the same time, every voter interested in the revenues resulting from the policy implemented by the winning party offers her own vote to the political party that situates the closest to her ideal policy point.24

Party A

Party B

Economic Left

Economic Right

Figure 2.2: A stylized model of spatial competition with two parties and three voters

To exemplify how voters decide, the Figure 2.2 presents an electoral competition in which two political parties, A and B, compete for the political support of three voters who have the ideal policy positions in X, Y, and respectively Z on a policy dimension concerned with the production and allocation of economic goods within the society.25 The party A is a leftist party, arguing that state should own a good part of economic units and should involve in greater redistribution of the wealth within the society. The party B, however, is pro-market and stands for free enterprise and minimum involvement of the state in economy. Both parties are interested in winning the elections (to get the support of at least two voters), and the voters X, Y, and Z, having symmetrical preferences, vote for the party with the most similar policy program to their ideal points in the Euclidean space. Thus, we have several hypothetical situations:

24

That is the Euclidean distance from voter is ideal policy point to the policy point of that party is smaller than any other distance from the voter is ideal policy point to any other political party position. 25 For the sake of simplicity, I shall call these three voters according to their ideal policy points.

48

Situation 1: The distance between the party A and the voter Y equals the distance between the party B and the voter Y. According to this situation, the party A gets the support of the voter X, the party B gets the support of the voter Z, and the voter Y is indifferent between the party A and the party B. If none of the parties modifies its position, then a stalemate occurs: none of the parties wins. Situation 2: Suppose now that party A tries to win by modifying its policy program and announces a policy position which is righter than in the Situation 1. Now, the party A is closer to the voter Y, that is the distance between Y and the newer policy position of A is smaller than the distance between Y and policy position of B. If the party Bs policy position remains unchanged, then party A gets the support of the voters X and Y and wins. Situation 3: Now suppose that B acknowledges that A has changed its position and there is the danger to lose the elections. But the party B also knows that if it is to win, it needs to get the support of voter Y, which is the median voter on this dimension. Thus, it replies to this new situation by formulating a policy position that is closer to the median voter Y than the previous position announced by the party A in the Situation 2. In the end, in their fight for the support of median voter, both parties will propose programs that will be very much similar (or identical) to the policy preference of median voter. In rational choice jargon, the position of median voter in this example is said to be the equilibrium of the game. The notion of equilibrium is of great importance in microeconomics and game theory. It denotes a situation from each no player or actor wishes to deviate by taking unilateral action, because if doing so, there would be no gains for the respective player. The notion of equilibrium asks that no player can deviate from that position and improve her payoffs. In the example above, neither party A nor party B could gain additional votes if they shift unilaterally from the position of median voter, Y. By contrary, the party that unilaterally shifts away from the equilibrium becomes worse-off in a situation like the one I depict in the Figure 2.2. Such crude example offers us a hint about what we should expect to happen in a two-party competition: the political parties will offer similar political programs to voters. At the same time, it is reasonable to suspect that a good part of the voters will abstain from voting either because political parties signal policies that are further away from their ideal

49

points, or because the uselessness of voting from individual point of view: no matter for whom one votes for, she gets the same (or very similar) package of policies. The simplicity of the model from Figure 2.2 could be regarded by readers as a merit, but also as a deficiency. One could argue that is too simple to match the complexity of real politics, which is inevitably true. However, as I argue above, the nature of the model requires us to make our theoretical constructs less complex than real politics. At the same time, I shall point in the next chapter to results of spatial models that are noticeably more complex than my crude example. Also, evaluating the model from a different angle, some readers could be unsatisfied with the result: it appears that nobody wins. If both political parties in my example of twoparty competition take on the same policy position, they should obtain similar shares of votes according to the rules I enounced above. Two answers can be provided to this criticism. The first is that winning an election is very much context dependent. Very often there is not the ideology that wins (or loses) an election, but the personality of the candidate, reputation of the candidate and of the party, the strategic errors during the electoral campaign, the economic situation, the international context and so forth. The second argument is that this kind of models is designed principally to study the dynamics of political ideologies, not to predict the result of the elections. Although appealing for any political, the power of foresee electoral results in all political contexts using such models is inevitably weak.

2.8. Summery
I have presented in this chapter the fundamental characteristics of economic approach to the study of political phenomena. I have started from the general requirement of a rational choice explanation, that is the methodological individualism, which, without being an exclusive attribute of this approach, is nevertheless a marked element of any explanation of this type. The methodological individualism requires that individuals should be the focus of social research and group decisions have be considered in relations with the individual tastes.

50

I have also discussed here the problem of rationality. Contrary to day-to-day language, the concept of rationality has a special meaning: it denotes the ability of individuals to take those actions that lead them to the fulfilment of personal desires. Individuals have the possibility to arrange the feasible choices according to the utilities provided by alternatives and they have to select from the top of such ordering. Moreover, some logical constrains imposed to any preference ordering, such as the reflexivity, comparability and transitivity, were discussed in relation with the notion of rationality. I have examined then how we can analyse the behaviour of parties and politicians in the analytical framework I sketch above. In this regard I have contrasted, somehow arbitrarily but for the sake of conceptual clarity, the sociological and economic approaches to party competition. If sociological explanations analyse the party behaviour in relationship with the social structure, the rational choice argument departs from some basic assumptions about the goals of political actors. Based on these assumptions, it tries to build on an argument about party behaviour in a certain context (two or multiparty competition, perfect or imperfect information, unidimensional or multidimensional party competition and so forth) and draw some implications of the model. Also, whereas in the sociological explanation political behaviour is constrained by social structure (in the sense that social predispositions influence the value commitments of political actors), in rational choice explanations the individual behaviour is always intentional. Political actors take a course of action because this provides them with some utilities or gains. At the end of the chapter, I have designed an example of Downsean party competition in order to highlight the structure of rational choice approach to party competition. It is very important for a reader unaccustomed with the economic theory to recall that according to this interpretation, politics can be envisaged geometrically. Electoral market, for instance, can be conceived as dimensions (or scales) on which parties and voters assume policy positions. Voters vote for a political party that presents policies they like most. On the other hand, political parties formulate in elections policies that are likely to be positively viewed by most of the citizens. Also, the context and the institutional setting are crucial for understanding the behaviour of political parties: they behave differently in two-party competition than in multiparty competition, information and uncertainty produce also different political effects in electoral competition etc.

51

In the next chapter, I approach in detail these formal models of political competition, highlighting how different assumptions produce different theoretical results. I shall start with simple (classical) models of party competition and end up with models of multiparty competition under uncertainty from which I extract the core empirical hypotheses I formulate in the Chapter 5.

52

CHAPTER 3: ECONOMIC APPROACHES TO THE THEORY OF PARTY COMPETITION

3.1. Introduction
In the previous chapter, I have presented some of the foundations of rational choice research. As we have seen, the rational choice models are strategic interactions of individual behaviours in which self-interested individuals intentionally pursue particular political goals. The aims of this chapter is to answer the following question: Given the assumptions we formulate about the motivation of political parties, what theoretical expectations we have about the way party competition develops in two and multiparty competition? Should we expect ideological convergence or divergence to occur in a political environment that resembles the characteristics of the CEE politics, such as multiparty systems, high political uncertainty, unstructured electoral institutions and so forth? In the next pages, I proceed as follow. First, I talk about the paradox of voting and its consequences for political process. Though the paradox has been known for several centuries, its consequences for political process are still largely ignored by most of the social science community. The paradox of voting is important for my research aims because it shows that the outcomes of political process do not deterministically emerge from the preference of political actors. As the paradox argues, institutions and strategic behaviour are important determinants of political outcomes. Second, I introduce the Downsian approach to party competition where I discuss the MVT and its implications. The next section of this chapter offers the standard critics of the Downsean model of party competition: 1) it ignores the role of alienation that makes voters abstain in elections if parties are too far from their voters ideal policy positions (e.g. Smithies), 2) it ignores the potentially deterrent element consisting of the new entries in

53

competition that force the established political parties to avoid ideological collision (e.g. Palfrey), 3) it ignores the crucial importance of non-policy factors in party competition (Adams, Merrill III, and Grofman), 4) it ignores that politicians are interested in implementing policies and not winning offices per se (offices are just instruments of implementing policies, e.g. Wittman), and 5) it is based on false assumptions about how voters decide in electoral competition, favouring proximity instead of direction (e.g. Rabinowitz and Macdonald). I present then models of multidimensional two-party competition, showing why the importance of centrist strategies still holds. At the end of the chapter, I discuss a model of multiparty competition under uncertainty (Lin, Enelow and Dorussens 1999), which represents the core theoretical argument of the thesis. It provides the theoretical foundation for the empirical hypotheses I formulate in the Chapter 5, pointing to the essential role of uncertainty for the outcomes of electoral competition. A brief discussion ends this chapter.

3.2. The Voting Paradox as a Puzzle for the Democratic Theory

A truism about political parties tells that modern democracy is party democracy. The statement emphasizes that modern democracy is inextricably linked to political parties, but it hardly explains why it is so. In this respect, two arguments have been generally advanced. The first argument is that modern democracies, because of their complex organization, large territories, high specialization of functions etc, can not function anymore to the same manner as the ancient democratic poleis did. Direct democracy is not anymore possible and perhaps it is not even desirable because of the permanent danger of the tyranny of majority (Madison: 1987, Dahl: 1989). Thus, the political parties become the main political agents that aggregate the political preferences of voters and advance policies on behalf of political community. Thus, one of the main functions political parties perform is the representation of citizens. A second argument emphasizes the role of political parties in overcoming a puzzle for modern democratic theory: the paradox of voting. The paradox of voting tells us that in 54

large communities that have to take a decision from large sets of potential alternatives, it is not always possible to arrive at social decision by the means of majority rule. The paradox suggests that if three or more persons have to decide by voting what the social decision of their group is, and there are at least three alternatives from which they need to choose, then in some social situations is possible that the group can not agree about a social decision. The Figure 1 illustrates one of the possible preference distributions for a threemember group (A, B, and C) which has to come to a social decision by choosing between policy alternatives a, b, and c. The vertical ordering of the alternatives a, b and c shows the individual preference between such alternatives. For instance, the individual A prefers the alternative a to b, and the alternative b to c. All three preference orderings are transitive, as defined in the Chapter 2 of the thesis.

A a b c

B b c a

C c a b

Figure 3.1. The paradox of voting

In such situation, if the committee has to reach a decision by using the simple majority voting, then the group can not arrive at any final decision. Looking at the distribution of individual preferences from the Figure 2.1, we see that the society prefers a to b (because A and C prefer a to b), b to c (A and B prefer b to c), and c to a (B and C prefer c to a). Then the social preferences between a, b, and c are intransitives, which makes impossible for the group to decide which alternative from the set is the groups choice. This paradox is known in the literature as the Condorcets paradox and it has been extended by another more stunning finding in social choice: Arrows impossibility theorem. Kenneth Arrow (1951, 59) demonstrates that in large communities that have to come to a social decision, there is no method of voting which allows the community to reach a decision preferred by the majority of people to any other alternative from the

55

feasible set. The social decision is then either imposed or dictatorial.26 Thus, some authors argue that the role of political parties in modern democracies is to overcome this paradox of voting by aggregating the individual preferences and formulating a more limited set of policies in competitive elections (Klingemann et al: 1994). Arrows terminology, however, can be misleading in the sense that the notion of dictatorship covers a wide range of possible alternatives, from only one person to about 49.9% of the groups population. In political theory, the imposed preference of only one person is called dictatorship and, at normative level, it is the least desirable political situation. We also know from Aristotelian typology of political regimes that the imposed preference of few members of the group is called either aristocracy, or oligarchy. However, we have practically no explicit political theory to address the problem raised by the social choice research: a larger group, a minority yet, imposing their preference on other (smaller) groups. The question why we do not have such an integrating theory, even though the implications of voting paradox were noticed more than half a century ago, is interesting and theoretically relevant. Although this question should be addressed much more consistently than I am able to do it in these pages, I argue that the incapacity of the modern political theory to address properly this question has roots in our excessive normativism with which we regard political phenomena. We still consider that democratic decisions should somehow reflect the views (or tastes) of majority of citizen body, ideally of all citizens. We still regard politics, including political competition among political parties, as a politics of social representation, and that individual tastes and desires should be mirrored by social decisions. However, as Riker (1980, 443) point out, if such equilibrium of tastes does exist only rarely, then there is necessary to find this equilibrium in the study of the interaction between institutions and tastes. Tastes, including the individual preference for certain outcomes and policies, are unlikely to bring alone the modern political science to the discovery of a fundamental equilibrium. Riker (1980, 443) argues that:

The meanings of imposing or dictatorship are a little bit different than in the usual political vocabulary. In Arrows terms, a social decision is imposed if certain social preference orderings are a priori excluded, irrespective of individual preferences (Arrow: 1951, 28-29). By dictatorship, Arrow understands a situation in which the social decision of the group is the preference ordering of the individual j irrespective of the preference orderings of the other individuals of the group (Arrow: 30-31).

26

56

The outcome, then, of the search for equilibria of tastes is the discovery that, failing such equilibria, there must be some institutional element in the regularities (or actual equilibria) we observe. We are forced back, therefore, to the eclectic stance that the political scientists have typically taken. Outcomes are, of course, partially based on tastes because some persons tastes (not necessarily a majority of peoples) are embodied in outcomes. But the ways the tastes and values are brought forward for consideration, eliminated, and finally selected are controlled by institutions. And institutions may have systematic biases in them so that they regularly produce one kind of outcome than another. In this sense, therefore, both institutions and tastes contribute to outcomes.

Is the paradox of voting an inescapable danger for political process? Black (1958) advances two methods of dealing with the paradox of voting. The first emphasizes the importance of the player who decides the voting procedures. Because no alternative can obtain the endorsement of simple majority over other alternatives, we may come to a solution if we could observe who the agenda setter is and what preference she has over the set of alternatives. The order of voting is important because, as Black (1958, 40) argues, the later one alternative enters in the pairwise contests, the more chances it has to be selected. Thus, with intransitive social preferences, the power over the agenda control is a very influential resource in politics. The second line of reasoning has been very influential in social choice theory. Black proposed an ingenious solution to the paradox of voting, which although constrains the distribution of individual preferences, it eliminates the problem of social intransitivity. Black argued that if each of individuals has transitive, single-peaked preferences, then the paradox of voting does not occur. He defines the single-peak curve as being the particular individual distribution of preferences over the available set that is always upward-sloping, downward-sloping, or is upward-sloping until it reaches a certain point (the optimum) and downward sloping afterwards (Black: 1958, 7). As we shall see further below, this singlepeaked property is a very important result for formal models of party competition.

57

The paradox of voting has serious implications both for democratic theory and for the study of political parties. In respect to the later, it suggests that political parties could perform the social functions in a manner that is opposite to the claims of normative theory of democracy. If normative statements do not hold with respect of individual tastes, why should they hold at the level of political parties and party competition? The paradox of voting raises an interesting question to the modern political theory, which Riker (1980, 445) acknowledges: outcomes of political phenomena are not always the reflection of tastes or institutions, but to large extent they depend of the will of politicians who manipulate agendas and institutions in order to obtain advantages from the disequilibrium of tastes.27 Politicians invent issues, avoid issues, create institutions, dissolve institutions and so forth because they can obtain some political advantages, given that no actual majority of tastes is likely to oppose. The implication of this argument suggests that if the role of politicians is so important for determining the outcomes of political phenomena, we should pay a particular attention to the tastes and desires of politicians in the different institutional frameworks. Political competition is one of the domains in which political manipulation is almost always present, and yet, the normative approaches to this topic are intellectually dominant. One essential question that raised debates in the literature is to what extent political parties offer a real choice to voters in electoral competition, so that the preferences voters are reflected in the competition among political parties? Do parties offer substantially different policy programs and do they appeal to substantially different electorates? Or by contrary, do they tend to offer similar policies, competing for the same electorate?

3.3. The Basic Concepts of Spatial Analysis of Politics

When the revolutionaries of 1789 chose to take seats on the left side of the French Constituent Assembly, they probably never thought at the impact of their political action in politics and political science. Consciously or not, this was the first time when the history
27

Riker (1980, 445) is also sceptical about the possibility of prediction, due to this problem of political manipulation. He argues that: [] And just what combination of institutions, tastes and artistry will appear in any given political system is, it seems to me, as unpredictable as poetry.

58

recorded the use of spatial political symbols by politicians in their political fights and ever since the left and right have been key political labels in politics and political science. But, the logic of spatial analysis had been used for intellectual purposes much earlier than the French Revolution. The classification of political regimes of Aristotle (1981), for instance, represents an example of a similar logic. Aristotles differentiation between political regimes according to the number of citizens who take decisions in the polis represents an interpretation of politics in spatial terms. Spatial theories have achieved significant reputation in economics and political science due to the parsimony and elegance of the argument. The usefulness of such method resides in the fact that it is not a pure descriptive approach but it goes to the heart of what a scientific theory is required to provide: an explanation which is subsumed to a generalized law (Riker: 1990, 168). As I have briefly argued in the Chapter 2, the spatial models of party competition make several assumptions about the political environment. The first is that we deal with two sets of actors, which are the parties and the voters. Being a particular tradition of research inside the rational choice theory, the theory considers these actors as being rational and behaving rationally, as defined in the previous chapter. Given this attribute of rationality, the theory assumes that each of the actors involved in the political competition behaves in a way that maximizes her perceived utility. The spatial theory also assumes that the positions of voters and parties can be arranged on a scale, usually called policy issue or policy dimension. The distinction between political issue and political dimension is a very important conceptual clarification particularly since these two concepts are sometimes used indiscriminately in the analysis of party competition. The political (or policy) issue is a (very) narrow, specific political problem that is high on the voters or parties agenda at a certain time. It may be persistent over time or just a temporary controversial matter. Such specific political issues are of great variance and one can find lots of such issues discussed in the elections. Some examples could be the level of taxation or whether to allow the former communists to hold public offices in transition period or the type of educational or health system (public or private) etc. On the other hand, a policy dimension is a more abstract and more inclusive concept that incorporates multiple policy issues of similar type. When parties, voters or scholars use left and right in their political or scientific discourse, they all refer to a political

59

dimension and not to narrow political issue. Thus, a dimension includes information about the positions of parties/voters on a variety of similar political issues. In the following chapters, I approach party competition on such competitive policy dimensions that combine various policy issues on which political parties take stances. Though not as concrete as the policy issues, the policy dimensions are more suitable for comparative analysis of party behaviour. Moreover, since issues affecting voters choice are many, inclusive policy dimensions are more elegant and theoretically useful for my research purposes. Instead of working with dichotomous variables, the spatial analysis views the political space as a continuum which allows actors to take intermediate positions between two extremes. In the case of ideological placement, for instance, parties and voters can place themselves not only at the left or at the right, but they can take intermediate ideological positions. This is another important feature of spatial theories of political competition as it assumes that parties and voters can introspect their preferences and place themselves on the dimensional/issue scale accordingly. Because each voter has an ideal position on the respective political dimension, she is able to compare the positions announced by the political parties competing in electoral market. By doing systematic comparisons between her most preferred policy position and the positions of political parties, a voter can determine which party is the closest to her position in the Euclidean space and consequently, she votes for this party. Accordingly, the parties know the distribution of voters policy preferences and thus they can choose what seems to be the best ideological position for winning the elections.

3.4. The Classical Spatial Models of Party Competition

Harold Hotelling (1929), Anthony Downs (1957) and Duncan Black (1958) were the first to introduce the rigorous spatial geometry into the study of political parties and electoral competition. It was the merit of Hotelling to be the first who formulated a spatial model of political competition, although his effort was somehow limited from the perspective of political science. His argument concerned mainly the spatial economic competition among 60

firms and the model he developed had only marginal applications for the analysis of party competition. The Hotelling model functions under several assumptions: 1) There is an economic market characterized by inelastic demand. The price is determined by the total quantity the firms issue on the market and by the transportation cost. This is supposed to include other elements that could make buyers buy from one firm or another. 2) The costs of production of two firms engaged into economic competition are set to zero.28 3) Unit quantity of commodity is consumed at each unit of time at each unit of the length of the line;29 4) The buyers are uniformly distributed along a line of a length l, which represents the market (Hotelling: 1929, 45). Because the demand is inelastic, the quantity Q demanded by consumers does not decrease as the price increases. Both firms move towards the equilibrium point of the market that is the centre of the line with the length l.30 According to the definition of the equilibrium given in the Chapter 2, this point has the property that, given the amount of quantity available on the market, it maximizes the profits of both firms. Since the demand is inelastic, the firms have incentives to move the price above the equilibrium in order to keep profits high in the long run (Hotelling: 1929, 48). However, such strategy has a weak point: each firm has to agree to raise the price for the quantity it delivers on the market; otherwise the firm that moves above the equilibrium loses customers and money since the buyers prefer to buy goods from the other firm. If the price of the good is set up above the equilibrium point, then each of the competitors can increase the payoffs by lowering the price and capturing a part of the customers from the other firm. However, the later has also the strategy of decreasing the price even further, and thus we assist to a war of prices until they reach the equilibrium point. From this point no player has the incentive to lower the price any further, since her
28

This condition allows a meaningful comparison of firms behaviour as it supposes that both firms have similar costs of producing one unit of commodity. 29 According to this assumption, the consumption is uniformly distributed on the market. 30 Recall that the transportation cost determines the price, according to the assumption 1. The higher the distance between buyers location and firms warehouse, the higher the price the buyer has to pay for that good.

61

business would become profitless. Hotelling (1929, 49) argues that although prices can be maintained above equilibrium (based, of course, on insecure agreements), they can not be set below such point for longer period. Although the author acknowledges the possibility of a "price-war" (the possibility that one firm, which is stronger on the market, sets up such a low price than the other competitor has to retire because it can not be competitive anymore), he argues that such a point is an equilibrium of the competition since "[] stability is by definition merely the tendency to return after small displacements" (Hotelling: 1929, 50).31 There are several implications regarding the competition of firms in such environment. The first one is that the price depends upon the unit price of transportation costs, and thus, a trader should increase the costs of the other firm that wishes to transport good into his part of the market. The second implication, particularly relevant for the theory of party competition, is the hypothetical situation in which one player has fix location and the other can choose freely the place of her business. In such situation, the second firm should come as close as possible to the location of the first so that it captures a larger share of the market. If the first firm may also choose the place of its business, then it has to move towards the centre of the market to avoid the circumstances in which the second firm increases the payoffs on its. Based on these findings, Hotelling explains the standardization of goods into markets. Moreover, he argues that such tendency is common not only in the economic exchanges but also in politics. Observing the political competition between Republicans and Democrats in the USA, Hotelling brings this argument to explain why the American parties use political platforms that are strikingly similar (Hotelling: 1929, 54). Hotellings argument is no more than a proto-explanation in the case of political competition. Downs (1957) was the first author who linked the argument of economic rationality, the economic models of spatial competition and the party competition in mass elections. The basic assumption of Downsean approach, as shown in the Chapter 2, is that politicians struggle to be (re)elected and political parties goal is to winning the election. As noted in the previous chapter, each political party is regarded as "a coalition of men seeking to control the governing apparatus by legal means" (Downs: 1957, 24) or "[] a team of
31

Italics in the original text.

62

men seeking to control the governing apparatus by gaining offices in a dully constituted election" (Downs: 1957, 25). As noted in the Chapter 2, one of the most contested assumptions of his definition is that the goal of political parties is to win elections and to control the government. 32 There is a serious amount of literature which argues very powerfully that political parties have other political goal than it is suggested in the Downsean definition, for instance advancing public policies (see below Wittman 1973, but also Budge et al 1987, Klingeman et al 1994). Their arguments come mainly from extensive empirical research showing that in many empirical situations parties do not comply with the assumptions of vote-seeking models but they behave as a sort of office or policy maximizators. In the case of officemaximization, the standard argument is that political parties know the governmental offices are distributed among parties that agree to form a political coalition. Usually these political coalitions are formed taking into consideration the ideological proximity among political parties. Thus, political parties could refrain themselves from pursuing the votemaximization strategy, hoping that being closer ideologically to other political parties would increase their chances in negotiations for coalition government formation (De Swaan 1973, Dodd 1976). On the other hand, the policy-maximization argument suggests that either the top leaders compete for offices in order to implement policies or that the ordinary activists of parties are sufficient influential to determine the leadership to maintain the ideological and policy identity of the party in exchange for their support in elections. However, three arguments can be brought to defend the assumptions of classical models of party competition about the goals of political parties. The first argument, which hints to the opposability between office and vote maximization, is that electoral strength is an important advantage in negotiation for government formation. Thus, political parties can not ignore the electoral game and electoral outcomes even when they are interested exclusively in obtaining public offices. Also, this argument assumes that parties have significant knowledge about the identity of future winners and losers and about their coalition preferences, which is hardly the case. Furthermore, the argument of a trade off between votes and offices ignores that electoral competition and coalition formation are
32

See footnote 14 of Chapter 2.

63

theoretically two separate arenas of competition. Both are nevertheless very important loci for political competition, but the argument of vote maximization offers us a better tool to analyse the party behaviour in electoral markets. A second argument in defence of the classical models regards the debate between policy and vote maximization. As pointed above, the advocates of policy maximization offer mainly an empirical argument to support their claims: parties seem to be linked to particular policies, which indicate that they are either policy maximizers or they balance between these two extreme goals (Budge et al 1987, Klingeman et al 1994). However, a theoretical counterargument to this claim is that parties that stick to relatively stable policies do not have necessary policy maximization as a cause of their behaviour. Other factors, such as the interaction between leadership and activists, risk-averse vs. risk taker, short vs. long political aspirations, could influence the behaviour of self-interested political parties, making it similar to policy-maximization behaviour. But probably the most important argument against policy maximization as a motive for human action in politics comes from the nature of policy as public good. The problem of collective action is a fairly established and accepted phenomenon in a modern political theory (Olson 1971). As almost all goods provided by the government, public policies are collective or public goods. By definition, a public good has the property that once provided to the members of the society, no member of the group can be prevented from consuming it. No matter whether she contributed or not to achieve this public good, a person can freely enjoy the public benefits of the good (Olson 1971, 15). When considering joining the collective action that is necessary for producing a public good, every individual of the group compares her individual cost of participation with her individual benefit obtained from consuming the good. Apparently, the rational decision of individual is to join the collective action that provides the public good, provided that the individual revenue exceeds the individual costs. It seems that the individual effort within the collective action is worth only when it is decisive for achievement of that public good. But individual effort is almost never decisive in collective action. Every individual knows that, since excluding the people who did not contribute at the production of the public good is impossible or unfeasible, she can enjoy consuming the public good without bearing any individual cost. Thus, the benefits of a public good are not sufficient to

64

motivate people to join the collective action! In order to determine an individual to join the collective action, the societies, particularly the large ones, have to provide their members with other private goods, called selective incentives (Olson 1971, 60-61). The theory of collective action provides us with a reason to regard as problematic the assumption that political parties (and especially political leaders) are policy oriented. If we consider that political leaders are rational actors, as I claim in the Chapter 2, the policy motivation can not be the only reason of their political actions. By contrast to policies which are pure public goods, winning elections and political offices provide positive selective incentives to political leaders that determine them to join the collective action of governing. Coming back to motivations of political leadership, Downs argues that:

Thus politicians in our model never seek office as a means of carrying out particular policies; their only goal is to reap the rewards on holding office per se [italics in original]. They treat policies as purely as means to the attainment of their private ends, which they can reach only by being elected (Downs 1957, 28).

But policies are valuable for politicians since they represent instruments through which political parties appeal to voters. Interested in getting elected, politicians have to formulate policies that are appealing to large number of voters. On the other hand, a rational voter chooses the political party that provides her with the maximum policy utility. Downs argues that each voter makes a comparison between political parties based on the following subtraction: E(UAt+1) - E(UBt+1), where: E(UAt+1) is the expected utility of voting for party A at the time t+1, and E(UBt+1) is expected utility for voting for party B at the time t+1. The above equation is called the expected party differential. If this party differential is positive, then voter votes for party A; if negative, she votes for party B; if zero, she abstains (Downs 1957, 40). However, the expected party differential is not sufficient to explain voting behaviour, since a voter has to take into consideration not only the future but

65

the present situation as well. Downs suggests that the vote based on the actual situation of the voter is much more rational than a vote based on future expected utilities, and therefore, the current party differential should be a much better predictor of vote that the future party differential. The current party differential is given by the equation: (UAt) - E(UBt), where (UAt) is utility the voter receive at the moment t, and E(UBt) is the utility the voter would have been received if the opposition had been in power (Downs 1957, 41). In order to accommodate the present utility and future utility, Downs introduce two modifiers. The first one is the existence of a trend, suggesting that the current party differential is modified by the existence of a particular trend in performance which influences the future expected utility of voting for that party. The second modifier comes into action when there is no policy base to separate the two competing parties. In such situation, in order to come to decision, the voter considers the difference between the performance of incumbents and the past performance of the predecessors in power. Coming back to the Downsean model of electoral competition, there are other two important assumptions the author makes use of. The first one is that political parties and voters can be arranged on a single dimension (most common the left-right continuum), and there is an agreement about the positions of political parties on this continuum. The second important assumption is that voter's preferences are single-peaked and the slope of preference curve is monotonically downward on both sides of the peak (Downs 1957, 116). Borrowing from Hotelling model, Downs argues that the rational behaviour requires parties to move towards each other, until their positions converge and the voters are not able to perceive any significant ideological differences. The convergence occurs at the position of median voter in the case of two-party competition, since the MVP has the property of maximizing the number of votes for each of political parties.33 The formulation of MVT has been one of the most significant scientific results of Hotelling and Downs approaches. The MVT states that, under several assumptions (two political parties, one competitive dimension, vote-driven political parties, single-peaked preferences of voters), the political parties end up proposing very similar political
33

See also the example at the end of Chapter 2.

66

platforms. However, the maximization property of MVT is conditioned mainly by the existence of two political parties competing in elections and a single salient dimension for political competition. If we alter these assumptions, the MVT might not hold. Indeed, moving from two to multiparty competition, Downs argues that political parties have incentives to diverge ideologically because of the institutional peculiarities of these party systems.34 As he writes down, the equilibrium outcome for multiparty competition is not the position of median voter:

Thus it is likely that in multiparty systems, parties will strive to distinguish themselves ideologically from each other and maintain the purity of their positions; whereas in two-party systems, each party will try to resemble its opponent as closely as possible (Downs 1957, 125-27).

Consider an example with three competing political parties: A, B and C. For the sake of generality, consider also that B is the centre party, whereas parties A and C are the leftist respectively the rightist parties. By moving closer to the position of party A, the party B loses votes in favour of the party located at its right (party C), as some of the voters who intended to vote for B would be disappointed with Bs new ideological location. Similarly, the parties from extremes are afraid that moving at the centre would make them losing the extreme voters which might turn towards new political parties or abstain from voting. Then, in the case of multiparty competition, the equilibrium outcome is the policy divergence. Contrary to the case of two-party competition, in multiparty settings with PR electoral systems, political parties can successfully appeal to extremist voters and gain political representation in parliament. Since extremist strategies are rewarding in multiparty format, political parties formulate distinct policy appeals to voters, thus increasing the ideological distance between political parties. However, the argument of Downs seems to be a special case of multiparty competition. Multiparty competition produces ideological divergence only under the condition of perfect information about the location of voters and

Usually competition in multiparty systems takes place under electoral systems based on proportional representation. Contrary to electoral systems based on simple majority formula, a PR electoral system offers representation to large and small political parties, so that the later can successfully compete in elections and get representatives in the Parliament.

34

67

parties. The argument is quite straightforward: consider three political parties A, B, and C that have the same location, k, on a policy dimension. If we assume that voters vote sincerely and evaluate only the policy distance between their ideal positions and the policy position of each political party, then every party obtains 1/3rd of the votes.35 However, this position k is never an equilibrium in case of competition under full information. Since parties have perfect information about party locations and voter distribution on the competing dimension, everyone can attempt to increase its vote share by deviating unilaterally from the initial position k. Thus, there is no equilibrium in multiparty competition under full information that involves party policy convergence. It seems that information rather than the number of political parties, as it is usually believed, drives away the competitors from convergent political strategies. As I show later on this chapter borrowing mainly from the scholarly work of Hinich and Enelow and in the next chapter using a simple model of multiparty competition under uncertainty, the level of information is crucial in determining the electoral strategies of competing political parties, regardless the number of parties within the system. Policy convergence may occur in multiparty settings provided that the level of information of parties is low. The findings of Hotelling and Downs are backed by the scholarly works in another area of research: committee voting. Contrary to mass elections, committee voting involves only one set of players: the politicians who vote for or against of a policy proposal. Consider five politicians A ($50), B ($100), C ($200), D ($500), and E ($1000) debating over a budget for public school subsidizes, with the numbers in brackets denoting the ideal individual budget, in millions USD.36 Knowing only the ideal individual budget is of little help for predicting what budget this committee will award for education next year. But the ideas of single-peakness and symmetry give us a great leverage over predicting the result of voting in this committee. If the individual sets of preferences are symmetric and singlepeaked, then every politician from the committee prefers budgets that are closer to her ideal point over budgets that are further away. If the conditions of single-peaked and symmetry hold, then the choice of this committee voting is $200m, the position advocated by the politician C, the median voter in this committee.

35 36

Or no votes at all, since voters have incentives to abstain when parties converge. So that ideally politician A would give $50m, politician B $100m and so on.

68

The Hotelling, Black, and Downs findings highlight the importance of the median position when we deal with unidimensional policy space. Their prediction for unidimensional political competition is the ideological convergence, with political parties approaching the position of median voter. However, we should recall that the simple majority rule conditions this finding: only one option is selected through the means of the majority rule voting. If more than one option is to be selected (like in the case of Downsean multiparty competition), then theoretical predictions of classical formal theories is that convergence does not occur.

3.5. Variations of Unidimensional Spatial Party Competition and Voting: Critiques of Classical Models and MVT

The findings of Hotelling and Downs, particularly the MVT, came rapidly under the scrutiny of several other scholars in formal spatial modelling. Although it is usually acknowledged that the position of median voter plays a significant role in shaping the electoral strategies, some critical evaluations of classical models attempt to prove that equilibrium at median voters position is an artefact of the assumptions we have about the party competition and the goals of political parties. In this regard, five major counterarguments have been formulated against the MVT in the literature of spatial theory of party competition.

3.5.1. The Role of Abstention in Electoral Competition The first major counterargument is that the classical models are grossly underspecified. The typical argument against Downsean model is that it ignores important assumptions about the party competition which, if they had been included in the model, they would have changed the dynamics of competition in electoral arena. This argument is very

69

straightforward and it deserves we pay a special attention to its implications for political competition. Smithies (1941) argues against Hotelling suggesting that the later explains why parties ideologically converge, but it does not offer an explanation for why parties diverge in certain empirical situations. He claims that Hotelling approach is suitable for modelling only a special case of electoral competition, which is the perfect competition. According to this view on economic and political competition, the firms can fix both the price of commodity and the location on the market, assuming that the price and the place of the other competitors are unaffected by their choice (Smithies 1941, 424). Smithies uses similar assumptions like those employed by Hotelling in his seminal argument. He considers for instance a linear market bounded at both ends, with only one price being set in every point of the market. This condition is similar to the assumption from Hotellings model that requires the price of the commodity to be determined only by the transportation cost. Additionally, there are two competitors having only two locations on the market that are subjects to constant marginal cost. The fixed costs the firms face when producing a unit of commodity is ignored in the model (or set to zero)37 and the firms have the freedom of choosing their locations on the market at no cost. Also, there is a uniform freight rate which is independent of the distance and of the price and quantity of the commodity and the relationship of the freight rate with the demand is such that there are sales at every point of the market (Smithies 1941, 425-426). Smithies analysis of spatial competition concerns three hypothetical situations. The first situation regards the case when, making adjustments, every competitor assumes that the opponent will adopt a location that is symmetrical in spatial terms with its own location and that the rival will set up a similar price for commodity unit. Smithies calls this situation full-quasi cooperation because he argues that in such case, the equilibrium outcome of the competition is similar to the situation in which the two competitors would act jointly as a monopolist on the market (Smithies 1941, 427). In the second situation, called quasi-cooperative as to prices and competitive as to location, each firm assumes that the rival will set up the same prices but it will keep the location of the business unchanged. The last

37

Note that this condition of zero cost of production is also a part of Hotellings model. See above in this chapter for further details.

70

situation is the one analyzed by Hotelling, in which the firms assume that the price and location of the rival are set independently of their own choices (Smithies 1941, 428). A monopolist, argues Smithies, has the privilege of exploiting both sides of the market without any fear of competition. Monopolist firm can set up the location of its business at any point of the market, but the most probable location is the centre of the market, so that the prices will be identical for any two points located at the same distance from the monopolist business. If the monopolist has two plants instead of only one, then each of the plant should be located at each quartile of the market (Smithies 1941, 430). In the case of two competitors, each firm tries to accumulate as much as possible from the market, whereas the monopolist tries to maximize the profits in both sides of the market (to the left and the right of its current location) (Smithies 1941, 430). The problem is that the further they go away from the hinterland38, the less successful they are in exploiting these regions, since the freight rates increase. In the case of the full quasicompetitive, Smithies (1941, 430) claims that no competitor is able to exploit more than a half of the market because the opponent will set up a similar price and will adopt a symmetrical location. Because there is no possibility to win more than half of the market, each of the firms will be concerned with maximization of profits in its half share of the market. But because there is also a freight rate involved and the price is dependent of this freight rate, then profits are not maximized if the firms locate themselves at the median point of the market.39 Thus, the maximization of the individual profits is reached when the firms act as in the case of a monopolist with two plants. However, in the case of quasi-cooperation as to prices and competition as to locations, each firm expects that the rival will set up a price similar to its own. At the same time, every firm anticipates that the location of rival remains stable while it will move its location towards the centre of the market. But if each follows the same logic, they will end

Smithies calls hinterland the spatial region that goes from the location of the firm on the market to one end of the market, in the opposite direction of the rivals place. Thus, each of the firms has one and only one hinterland, where it faces no competition from the other firm, and which is able to exploit it alone. By contrast to the hinterlands, there is only one competitive space, between the locations of the firms, where the market war takes place. 39 The price P is given by the equation P=aq + b, where a is the freight rate, q is the quantity delivered to that point and b is the intercept. Because there is a uniform freight rate per unit of distance (see Smithies 1941, 425), then the price increases if the distance from a point of the market to the firms location increases.

38

71

in symmetric positions, closer to the centre of the market than to the quartiles (Smithies 1941, 431).40 In the case of full competition (the Hotelling model), when firms change independently both the price and the locations, the equilibrium is again closer to centre than to quartiles of the market, because the profits are maximized by cutting prices and by expanding the market shares (Smithies 1941, 432). However, the equilibrium point is not the centre of the market again because of the freight rates. Smithies argues that such equilibrium at the centre of the market has been achieved because of the characteristics of the demand, which is totally inelastic in the case of the Hotelling model. In such case, every firm is able to pass to consumers the entire cost of delivering the commodity from the centre of the market to extreme points, because the demand is totally irresponsive to price. But when such constrain on demand is not included in the model, the equilibrium outcome is not the centre or the median point of the market.41 Smithies argument is straightforward. If buyers (or voters in the case of political competition) can refuse buying an expensive product, then a shift of the firms towards the centre would incur a loss at the extremes of the market. The argument is similar if the buyers have the opportunity to substitute the good with another good. In both cases, the firms are worse-off if they move towards the centre of the market. Translating the economic words in political language, if we allow the voters to abstain in two-party competitive elections, then political parties maximize their vote shares by diverging from the centre/median position. Adams, Merrill and Grofman (2006) have expanded this theoretical argument showing theoretically and empirically how political alienation plays an important role in determining political parties to diverge in two-party competition. At the same time though, the policy divergence Smithies obtains in his model of party competition is largely dependent on the assumption that voters are uniformly distributed among the market. Such assumption gives weight to the support of extreme voters, which become alienated as the parties converge to the electoral mean. However, if the distribution of voters is bell shape (unimodal), then Smithies result might not be robust.

Again, the impact of the fright rate is visible. Because of it, the location of the firms will not be at the centre of the market. 41 In the case of Hotelling model, the mean and median of the market are at the same point.

40

72

3.5.2. The Role of Entry Threat on Electoral Strategies: Divergence Occur Palfrey (1984) provides a similar counterargument to Hotelling/Downs findings. He argues that equilibrium outcome in two-party competition can not be party convergence given the impact of potential new entry on the strategies of the established political parties. Palfrey proposes a model of three parties competing in elections, in which two parties are established [or dominant, as Palfrey (1984, 141) calls them] and the third is the entrant. The voters have single peaked preferences over the policy space and the distribution of voters is unimodal. Each voter votes for only one party and the party choice is determined by the distance between voters ideal policy position and the positions of parties. The parties choose their positions in the following order: the dominant parties choose simultaneously their locations and the entry decides on its position later so that it tries to maximize its vote share (Palfrey 1984, 141). The established parties face a dilemma. Should they move closer each other on the policy space, so that they will maximize their shares of votes, or should they diverge from the median position, making the entry ineffective? Under simple majority voting system, Palfrey (1984, 152) argues that convergence of the dominant parties at the point of median player is never an equilibrium outcome. Furthermore, under such conditions there is no equilibrium outcome that involves the convergence of established parties at any point of the electoral spectrum. The proof is almost trivial and needs no advanced mathematical equations. By logical reasoning, if the two dominant parties adopt very similar platforms at the median position, the entry may enter near the location of one establish party and wins. In order to avoid such outcome, the established parties have to diverge from the median position, though not very much because this would make the space between the dominant parties large enough to be appealing for the new entry. If the policy space takes the form of the interval [0;1], then the outcome of the game is [1/4;3/4] for the dominant parties. Under

73

unimodal distribution of voters and with the dominant parties located at these points, the entry never wins (Palfrey 1984, 152-54).42 However, it is not clear enough how general Palfreys argument is. The significant problem raised by his model is whether the findings still hold if the dynamics of the game change. It seems that the findings of Palfrey are valid only if parties enter in electoral market sequentially, with the third party joining the competition after the established parties have chosen their locations. As I shall show Chapter 4, if we model the multiparty competition non-sequentially, taking place on a single dimension and under the assumption that parties can not leapfrog the position of other parties, then political parties have a maximizing strategy in moving towards the centre of electoral market. They end up by proposing very similar policy programs to voters.

3.5.3. The Role of Non-Policy Factors on Voting Choice: The Parties Diverge The core element of the classical models is arguably the policy distance between voter and competing political parties that determines the vote choice. In the classical models, the party choice is exclusively a function of distance between voters and parties, disregarding other non-policy factors, such as the evaluation of individual candidates, the pre-established party loyalties and so forth. Since the party choice is a function of ideological proximity, according to Downsean model every voter i votes for party k given that: Vi(k)= max (Ui) = min [-(Xi - Sk)2], where

(3.4.3.1)

Vi(k) is the party choice k of voter i, Ui represents utility of individual i which is maximized when voting for party k, Xi and Sk represents the policy position of voter i and party k.

42

The proof is trivial. If established parties take positions at .25 and .75, the entry may enter in competition at three possible locations: between 0 and .25 (at the left of first party), between .25 and .75 (in between the two parties) and between .75 and 1 (at the right of the second party). If the entry decides to enter at the extremes, it obtains less than .25, and loses (the established parties obtain more than .25 votes). If the entry decides to enter between the established parties at .5, for instance, then the entry gets exactly .25 if voters are uniformly distributed, and established parties obtain more than .25. With established parties located at .25 and .75, the entry never wins. QED.

74

Recall from the Chapter 2 that the Downsean utility of a voter i when voting for a party k is given by the quadratic utility function: Vi(k) = - (xi sk)2

(3.4.3.2)

However, extensive empirical research on voting behaviour found that voters have multiple reasons for why they vote for a particular party, among which the policy proximity is only one factor. Extensive research point out that one major factor that has influence on voting behaviour, particularly but not only in American context, is the partisan affiliation (Campbell et al 1960, Miller 1991). The scholars of voting behaviour see the partisan affiliation as a long-term, affective, psychological identification with a particular political party, which results mainly during early socialization process and intensifies with age (Jennings and Niemi 1973; Dalton 2002, 175-78). Voters with partisan affiliation have their utilities affected not only by the policy distance between them and political parties, but also by the psychological attachment to a political party. In contrast with the policy-only model of party competition, Adams (2001) and Adams, Merrill and Grofman (2005) formulate two models of party competition, which include both policy and non-policy factors, to account for the electoral strategies of political parties in four established democracies: Great Britain., USA, Norway and France. The first model is a probabilistic model of voting that incorporates policy and non-policy factors together with a random utility term: Ui(k) = Vi(k) Xik = - aj(Xij-Skj) + bktik +Xik ,

(3.4.3.3)

where Xik is a voter-specific random utility which represents unmeasured components of voter is utility for candidate k, aj is policy-salience parameter for jth issue (which is estimated from data), tik is a vector of non-policy variables, and bk is a vector of salience parameters of non-policy variables (Adams, Merrill and Grofman 2005, 22-3). According to (3.4.3.3), the utility provided by a party to a voter is given not only by the policy distance between them, but also by the voters psychological attachment to that party and other non-policy evaluations of party leadership, political reliability, politics performance

75

etc. For instance, a voter who identifies with party k might have the tik parameter 1 and zero otherwise. The implication of this unified model of party competition is that voters might vote for a party which is not close to their ideal positions in policy terms, if the non-policy factors such as partisanship are sufficiently salient. Also, political parties know that voters have non-policy predispositions when they choose the party to vote for and that policy strategies might be ineffective for changing voters choice if the expression bktik is large enough. Thus, instead of trying to persuade the partisans of other parties, a vote-seeking political party will propose policies that resemble the policy expected by its constituencies in order not to alienate them. The second model accounts for voters uncertainty about what policies political parties implement after elections. The model originates in Grofmans (1985) pioneering research about the role of discounting in politics.43 The idea behind the model is that voters can not fully compare all the party platforms in elections and so they compare what political parties would do once they were elected. But voters know that political parties can not fully implement the platforms they present in elections and so they discount the partys policy promises. They take into account the position of status-quo (the policy which is already in place) and the positions the parties advocate, discounted at a particular rate. The idea behind discounting is very simple: imagine that the status-quo is located at 5 on a 7point policy scale. The leftist party L advocates a policy located at 1, whereas the rightist party R proposes a policy located at 6. If all voters discount the ability of parties to implement their promises with a rate of 50%, then they will perceive the position of L at 3 (instead of 1 - the official position of L) and the position of R at 5.5 (instead of 6 the official position of R). Note that discounting has an impact on party choice: a voter with an ideal policy position in 4 prefers R to L without discounting, but when she discounts parties promises with 50%, her preference shifts to party L. If political parties know that voters discount their policy appeals, they will end up presenting policies that are further away from the status-quo. Thus, the second model of party choice, the unified discounting mode is:

43

The idea of discounting political promises has been supported by extensive empirical research. See Merrill and Grofman (1999) and Kedar (2002).

76

Ui(k) = Vi(k) Xik = aj[Xij (SQ + (1 - dk) (skj SQ )]2 + bktik +Xik (3.4.3.4) where, dk is the discounting factor for voter i (Adams, Merrill, and Grofman 2005, 23-6). The unified and the unified discounting models produce different predictions than Downsean model, in the sense that non-policy factors and discounting drive political parties away from the centre. The policy-only (Downsean) model produces a convergence of parties at median position. By contrast, the unified model of party competition forces political parties to appeal to their political constituencies. If voters have non-policy reasons to favour some parties, the other parties can not overcome them by presenting policies very close to voters ideal locations. If fact, they may lose votes since some traditional voters may become alienated with the new party location, whereas the voters targeted by the party have partisan affiliation that determine them to vote for other parties. By consequence, political parties choose to advance policies that are similar to those preferred by their constituencies. Finally, the unified discounting model produces equilibria in which political parties diverge even more than in case of the unified model, appealing typically to even more extreme positions. However, there is no consensus about the role of partisan affiliation on party strategies either among political sociologists or among rational choice theorists. McKelvey (1975) proves that the existence of political partisans in two-party competition drives political parties towards the median voter, provided that in the electorate there is a certain amount of policy related voting. He argues that:

[] It is found that if the distribution of the citizens preferred points is symmetric and unimodal, a sufficient condition to guarantee the existence of an equilibrium is that the candidates strongest supporters are always those citizens whose preferred position is closest to that offered by the candidate. If the distribution of the citizens preferred points is any symmetric distribution, then a sufficient condition to guarantee the existence of an equilibrium is that the candidates strongest supporters are extremists. In both cases, the equilibrium is at the median of the distribution of preferred points (McKelvey 1975, 835).

77

McKelvey (1975, 835) argue there is a tendency in elections for candidates to move away from the strongest supporters and, even when the extremists vote more consistently for the preferred candidate, the outcome equilibrium is still at the median position. The argument, emphasized also by the classical studies on partisan affiliation, suggests that when voters have non-policy reasons to vote for one party (they are ab initio biased towards one party), political leaders may have sufficient place of political manoeuvring.44 They might attempt to move towards the centre of political dimension in order to obtain the support of undecided voters who do not have any partisan affiliation but who can be attracted by a set of policies which resembles their ideal policy positions. Also, as I argue in the Chapter 4, the non-policy motivations such as the partisan affiliation are very powerful obstacles that obstruct the new entry to join the electoral competition, favouring thus a centripetal pattern of party competition.

3.5.4. Parties as Policy-Seekers Smithis, Palfrey, and Adams et al offer critiques to the Downsean model largely relying on the Downsean assumptions. They do comply with most of the core assumptions of traditional models of party competition (parties as vote-maximizers, unidimensionality of political space, proximity as the main criterion of party choice etc), but they point out that the classical models are grossly unspecified. They argue that the classical models ignore many important assumptions which, if included, would make the findings of formal models resemble more the conclusions of empirical research. Thus, the absence of these important theoretical variables in the classical models produces a marked dissimilarity between the theoretical predictions of formal theory and the empirical results we get from voting behaviour studies. However, there is another type of argumentation against the MVT which is substantially different than the above critiques and regards a different understanding of political process. In many respects, these arguments do not function as refinements of
44

I thank Professor Whitefield for suggesting me this idea about the role of partisanship in sustaining the convergent pattern of party competition.

78

Hotelling/Downsian models but as competing genre. The major claim of this reasoning is that parties have not (pure) vote or office interests, but they are oriented towards the fulfilment of policy objectives (Wittman 1973) or have mixed political goals (Wittman 1983).45 Wittman (1973) advances two alternative models in which parties maximize policies and not votes. In the first model, called the alternative competitive model, Wittman (1973, 495) assumes the parties are interested in promoting certain policies and winning is just a mean to this end. According to this argument, political parties are interested to maximize the expected policy utility. The difference between the assumptions of the traditional models and the assumptions of the alternative competitive model can be best explained using the analogy with a business firm. As Wittman points out, a firm maximizes the profits and not necessarily the volume of sales, because profits can be increased by reducing the number of goods sold on the market and raising the price of commodities. Using the example of an election that is concerned only with the allocation of income, based on intransitive social preferences and with voters interested only on personal income, Wittman (1973, 496) shows that even in the case when parties are interested in policies, there is a tendency of political parties to collude against the interests of voters. However, if the level of information is very low among voters, then rational parties aimed by policy goals diverge (Wittman 1973, 497). Wittman argues that when parties are interested only in winning the election, collision is impossible, because the political game is a zero-sum game. However, if parties are interested in policy outcomes, then collision is a rewarding solution (Wittman 1973, 497). If voters are completely uninformed about political parties platforms, then parties can keep all the income to them and provide nothing to voters. Nevertheless, such strategy is risky because it leads to a situation in which each party has 50% chance of winning all the income and 50% chance of winning nothing. By presenting similar policy platforms, parties reduce the risk and may obtain 50% of the income with certainty. Such model is called the restricted competition model. In a later approach, Wittman (1983) proposes another model of party competition in which parties have both policies and office goals. Not surprisingly, the synthesis model
45

By mixed I mean a certain combination of policy and office goals.

79

leads to different equilibrium outcome than the traditional model of party competition. If parties have mixed political goals (offices and policies) and they maximize the expected utility of winning the elections and implementing the preferred policy, then parties can trade off the chances of winning for moving closer to their ideal policy position.46 Thus, when parties have policy objectives and they have different ideal policy positions, they will diverge ideologically. Wittman (1983, 145) concludes that when parties are interested in pursuing policy goals, they are less responsive to the interests of voters than when are interested in winning. Wittman claims that the synthesis model of party goals produces better results in the light of the empirical analysis of party competition. At the same, the author does not actually test the formal findings in empirical settings. In sustaining his claims, Wittman relies on empirical analyses of other researchers who found remarkably ideological differences between political parties in electoral competition or between policies of governments lead by left-wing parties and policies of centre-right governments (see Page 978, Ginsburg 1976, Hibbs 1979). On the other hand, the claimed empirical superiority of synthesis model is partly based on American presidential competition which is an environment that produces certain affects to the incumbent. Given that incumbent presidents can not participate in elections beyond the second mandate, there is no direct electoral accountability for the incumbent. Thus, she is relatively free to pursue policies that are far more extreme than it would have been if the incumbent had participated in subsequent electoral competition. Wittman proposes an argument that shows a different perspective over the political process. According to this view, parties are not (exclusively) selfish actors interested in private gains and they are ready to offer policies to the public, which are by nature public goods. As I argued above, the assumption that political parties are policy seeking actors also has important theoretical flaws. Foremost, it can not adequately explain why rational individuals (politicians) join the collective action for pure public goods (policies).

46

Wittman (1983, 142-143) argues that when parties have mixed political goals, they maximize not the probability of winning as in the case of Downsian model, but the following expression: [the probability of winning] times [the utility received from the policies implemented if winning] plus [the probability that the opposition wins] times [the utility received from the policies implemented if the opposition wins].

80

3.5.5. The Influence of Directional Voting on Policy Divergence Recent developments in the theory of voting question the assumption of traditional models about how voters choose in electoral competition. According to the traditional voting models, voters make their decisions taking into account the Euclidean distance between their ideal policy positions and parties advocated positions. As I mentioned earlier, this assumption about how voters decide in elections is based on another assumption which suggests that voters and parties can place themselves on a continuum that links two opposable policy positions. Because the traditional model assumes that voter votes for the closest party, it is often called the proximity voting model. However, Rabinowitz and Macdonald (1989) argue in their seminal article that the proximity voting model has important flaws from the cognitive perspective. The authors claim that the left and the right are concepts of symbolic politics and, in order to have an impact, they should induce sentiments to voters (Rabinowitz and Macdonald 1989, 94). When parties take a stance on a particular issue, they need to make sure they are able to induce emotional reactions to voters, through association with previous experience, which makes the issue politically relevant (Rabinowitz and Macdonald 1989, 94). The authors argue that symbolic politics is associated with emotional reaction of voters to issues brought out by parties in electoral competition and that the issues which have symbolic power are characterized by direction and intention. Thus, argues Rabinowitz and Macdonald (1989, 94) the impact of issues should be modelled in terms of direction and intensity. According to the directional model of voting theory, voters take into consideration the direction of their positions on an issue and the intensity of their response. For instance, if a voter is strongly in favour of a particular policy, when deciding about her choice in elections she observes what party has a similar direction of policy positions and how much intense is the political message of the respective party. For instance if voter X is slightly in favour of governmentally subsidized health system and she has to choose between a party that is strongly in favour of a fully subsidized health system (left party) and another party that advocates softly a private health system (right party), she will vote for the leftist party 81

because this takes a similar direction on this issue with voter X. Even if the right party is closer to X in Euclidean space, she still prefers the left party because it has the same direction of preference on the health issue.47 What about the case when both parties propose the same type of policy but they differ about the level of intensity (for instance, the party A puts much more emphasis on this issue than B)? If voter X attaches high salience on this policy, she will vote for party A, because both the direction and the intensity coincide. However, if the voter X attaches low importance to this policy it does not mean she nevertheless favour the party B. By contrary, Rabinowitz and Macdonald (1989, 98) suggest that probably voter X will not be aware of the position of party B because of the low importance the party attaches to this issue. On the contrary, the voter will know that party A and she have similar policy positions because of the high importance the party attaches to this issue, which is probably signalled by more statements in favour in this policy during the electoral campaign. It is not difficult to observe that directional voting model foresee electoral results that are conflicting with the predictions of the MVT. By emphasizing the direction and the intensity as the main criteria of party evaluation, the directional model gives credit to policy dissent. When voters have preferences distributed over the entire area of an issue, it is expected by this model to see parties taking different and even extreme policy positions. The only possibility for policy congruence is the special case in which there is a broad consensus over a public policy. However, one legitimate question that should be raised when we evaluate the implications of the directional model is why we do not see parties taking very extreme policy positions in the elections? Rabinowitz and Macdonald (1989, 106) argue that candidates or parties need to convince voters of their reasonableness and because they fear of being labelled as extremists, they have to confine themselves to what the authors call the region of acceptability. Building upon this region of acceptability, Rabinowitz and Macdonald (1989, 109) argue that if voters are symmetrically distributed around the neutral point, all strategies in the region of acceptability are equally desirable. If two competing parties are one to the left and the other to the right of the neutral point, then all voters from

47

Notice that while in the proximity model policies are ordered sets on a policy continuum, in the case of directional model policy alternatives are dichotomous.

82

left part will vote for the left party while voters from the right will choose right party. If both parties are on the same side of the neutral point, then the party with a more intense political appeal (but who is in the region of acceptability) receives the support of voters from this side (Rabinowitz and Macdonald 1989, 109). The authors bring two arguments in favour of directional model. The first argument is theoretical and suggests that the assumptions of the traditional (proximity) voting model are unrealistic. Voters do not have much knowledge about the way politics function, Moreover, the level of information of voters about relevant political facts is generally very low. Thus, the assumptions of the traditional model that voters can order the set of policies over the entire policy space and that they are able to place themselves and the parties on such continuum are implausible. The advantage of the directional voting model is that voters do not need to have deep knowledge about the whole spectrum of the policy alternatives on a particular issue. They just need to decide what kind of policy they favour for every particular issue. The second argument Rabinowitz and Macdonald bring about in favour of the directional model is the empirical superiority. The proximity model has been criticized from the early stages of empirical analysis of voting behaviour because it consists of important theoretical and empirical flaws (see Stokes 1963). The refinements of the Downsean model could not eliminate all these flaws, particularly the assumption of how voters decide for what party to vote in elections, and thus a good part of the empirical research in the theory of voting does not fit the predictions of the proximity model. But, the claims of superiority of the directional voting model have not been received without criticism. Although the authors assert the empirical superiority of the directional model in several instances (Macdonald et al 1995, 1998, 2001), there have been important research that refute their claims. One example is Westholm (1997) which argues that directional model is less falsifiable than the proximity model and the claimed empirical superiority rests on tests that presuppose interpersonal comparison of the utility.48 Westholm (1997, 867) criticizes Rabinowitz and Macdonald for leaving unspecified the size of the region of acceptability, which makes the directional voting model less falsifiable
48

This is an important theoretical flaw here because, as theories of individual choice, they should rely on tests of intrapersonal comparison of utility. Also, inter-personal comparisons of utilities have been traditionally considered meaningless in economics because individual preferences are subjective.

83

than the traditional voting model. Moreover, not even the empirical analysis does fully support their claims of superiority of the directional model. When data is set to show intrapersonal and not interpersonal comparison of utility, the superiority of directional model over the proximity model vanishes (Whestholm 1997, 873). By contrary, in such cases, the proximity model predicts better the results of the elections. The directional model has been constantly criticized for operationalization faults. Merrill (1995) suggests that the decision to place political parties on the issue space according to the mean score of party evaluations as perceived by voters is a methodological artefact that produces empirical evidence in favour of the directional model.49 Krmer and Rattinger (1997) also contest the superiority of directional model using the same argument. In the line with Merrill, they argue that when using individual evaluation of party positions, the proximity model is constantly superior in predicting the vote choice (Krmer and Rattinger: 1997, 23). Furthermore, Lewis and King (1999) contest the superiority of the directional model even though the authors also reject the fact that proximity model works better, suggesting that the dispute between the two models is still undecided.

3.6. Multidimensional Voting: The Power of Mean Positions

Another major critique to the Downsean model regards the assumption that competition takes place on only one policy dimension, which is almost never accurate to describe the party competition in modern democracies. Again, extensive empirical research have shown that competition usually takes place on more policy dimensions, which may produce changes in the pattern of party competition. Because voter preferences are distributed on multiple dimensions (vectors), there may be no individuals who occupy the point of median voter on all the relevant policy dimensions. Thus, the main questions raised by these

This criticism is related to the discussion of intra and interpersonal comparison of utility. If directional model is, and it is, a theory of individual of choice, then it has to be based on intrapersonal comparison of utility. However, in this case, Rabinowitz and Macdonald should have taken into consideration the individual evaluation of parties (intrapersonal) and not mean evaluation (interpersonal).

49

84

empirical findings are to what extent and under what conditions (if any) the policy convergence is still an equilibrium in multidimensional party competition? Davis and Hinich (1968) advance a model of two-party competition that takes place on multiple dimensions, with all voters casting their votes (deterministic voting). Every voter has a set of her most preferred policy positions on each dimension, which is a column vector x1 = {x1,,xn}. The distribution of individual preferences is not restricted and there is infinity of such position available. Two political parties compete in the election, having two policy positions in multidimensional policy space, denoted by and . Since most of voters do not have identical policy preferences, each individual has a loss matrix which depends of the distance between her policy positions and the positions of party and on policy space. Let denote x the vector of the most preferred policy points of an individual, and A being her loss matrix, and x and A have a joint multivariate distribution such that the conditional mean preference is independent of A. Thus, given A, there is a vector which is the mean preference position for all population. For the sake of simplicity, I relegate now the formal argument to the Appendix 2, retaining here only the verbal argument. Those who are interested by the formal proof should consult the Appendix. Davis and Hinich argue that a benevolent dictator who would like to introduce a policy into society should minimize the loss of the society: she needs to introduce a policy which is the mean preferred position, , because this position minimizes the average loss (||x-||)2A. In the case of two-candidate competition, provided that all voters cast their vote, Davis and Hinich (1968, 64) show that a randomly chosen individual is more likely to vote for this mean position than for any other position in the policy space. For a single dimension, this is the Downsean median voter. For N>1, the mean preference has a less precise meaning, but Davis and Hinich (1968, 64-65) suggest that, if the policy space is characterized by a multinormal distribution of individual preferences, then the mean position behaves as a median preferred position. Also, even when the distribution of voters preferences is not multinormal, the mean position is very influential. The candidates can not deviate from this point and still win the elections. Davis and Hinich find that mean positions exercise a powerful attraction for politicians interested in winning the elections in two-party systems, with multiple

85

dimensions for competition and deterministic vote. However, their finding is very much dependent of the multinormal distribution of population. Moreover, the centripetal competition at the mean positions does not indicate that the mean position is also a stable equilibrium. Indeed, Plott (1967) shows that equilibrium is almost impossible in multidimensional voting with majority rule, unless we employ very restrictive conditions about voters ideal distribution. Thus, equilibrium outcome is only rare in models with deterministic vote: only if we constrain the individual preferences, we may obtain equilibrium at the mean position in multidimensional voting. Because many voters abstain from the polls in real elections, a new generation of models has incorporated this missing variable. In the probabilistic models of voting, voters are supposed to vote in elections with a probability that takes a value from the interval [0;1]. When the voter is ideal position and party js most preferred point coincide, the probability of voter i voting for party j is 1. Increasing the policy distance between party is and voter js ideal points makes the probability of individual j voting for party i decrease towards 0. With this new assumption, formal models approach more realistically the particularities of real elections.50 The deterministic models of spatial competition use very restrictive assumptions about the individual preference and choice in order to obtain equilibrium. Usually the conditions necessary to obtain an equilibrium in spatial models with deterministic voting are the unimodal distribution of preferences and the symmetry of preferences around the mean. Extending Plotts findings, Hinich, Ledyard and Ordeshook (1973) show that when we relax the assumption of symmetrical distribution of voters, the mean position remains the equilibrium outcome in multidimensional competition with probability voting. They show that even if candidates maximize votes or pluralities, unique pure-strategy equilibrium exists (Hinich, Ledyard, Ordeshook 1973, 166-67). If the distribution of voters individual preferences is concave (but not necessary symmetrical) and the probability of voting is of similar function for each voter, then this unique pure-strategy Nash equilibrium is the mean position (Hinich, Ledyard, Ordeshook 1973, 170). If we impose the restriction

50

The probability of voting does not depend only by the policy distance between parties and voters. Others non-policy factors may be relevant too.

86

that candidates do not leapfrog but we allow convergence, then parties converge at the mean position. This result yields a very interesting argument. The model explicitly assumed that voters have identical probability voting function p, which means that voters are not biased by non-policy features of political competition. If partisan bias affect the probability of voting p and the probability of voting does not correlate with the policy distance between voter and parties, then political parties might not converge (Hinich, Ledyard, Ordeshook 1973, 173). Enelow and Hinich (1984) offer further support that probabilistic voting leads to policy convergence at the mean of voters distribution in multidimensional two-party competition. Modelling an election where two parties compete along an ideological dimension with multiple issues and with two groups of voters voting, they show that a pure Nash equilibrium exists at the mean position if both groups attach the same importance to the ideological dimension and weight issues in a similar format (Enelow and Hinich 1984, 476). In a later paper, Enelow and Hinich (1989) argue that high uncertainty about party and voter positions are factors which further increase parties predisposition for policy collision in two-party multidimensional competition with probabilistic voting. Candidates policy position may be imperfectly perceived by voters, given that highly informed voters are rare and accumulation of information is not rational from the point of view of utility of voting. Also, political parties may face different forms of uncertainty about the preferences of voters, which comes from various sources: low information, strategic voting, high electoral volatility etc. Thus, if the electoral process depends not only on the policy position of parties but also of the level of uncertainty about voters or parties, or of any other form of non-policy bias (random element), then an equilibrium occurs in policy convergence. More interesting, the authors argue that with such random element, the dimensionality of space is irrelevant: policy convergence occurs in uni or multidimensional competition (Enelow and Hinich 1989, 110). The difference between deterministic and probabilistic models of spatial competition is striking. Equilibria are much more robust in two-party competition on multiple dimensions with probabilistic voting than in the case of two-party, multidimensional competition with deterministic voting. Equilibrium generally occurs at

87

centrist positions and shows the power of centre in two-party competition even when this takes place on multiple dimensions or issues. Moreover, the probabilistic models work with more relaxed assumptions, so that models approximate better the significant variables that occur in empirical party competition.

3.7. Competition in Multiparty Format

Although pure two-party systems are rare among parliamentary democracies, there is a big cavity in the literature between the formal models of two-party and multiparty competition. The literature on spatial models of multiparty competition is thin and only few articles produced in the last two decades could be mentioned as valuable scholarly work. One of the early works on spatial multiparty competition is Greenberg and Shepsle (1987) who extends Palfreys model of party competition to multiparty format. Recall Palfrey (1984) who suggests that Downsean convergence could not occur in two-party competition because parties need to thwart a potential entry at extremes. Greenberg and Shepsleys model of multiparty competition works with three restrictive assumptions. The first assumption assumes that the set of alternatives can be ordered on a policy line, and

the set of alternatives can be either finite or infinite. The second assumption is that every voter has a preferred policy position on and she prefers policy options that are closer to

her ideal point to policy alternatives that are farther away. Finally, the third supposition assumes that an individual voter casts her vote for an alternative if that alternative is the closest of his ideal policy position (Greenberg and Shepsle 1987, 527-28). If in an election characterized by these assumptions a society needs to select a number of alternatives n 2 from a set of alternatives Q 3, then no equilibrium (or K-equilibrium as the authors call it)51 exists in such electoral competition. The argument of Greenberg and Shepsle is straightforward and does not require a sophisticated mathematical proof. They argue

51

A K-equilibrium is a set of alternatives A, [which] it consists of K alternatives and if no other alternative given A and this additional alternative, is supported by more voters than any of the original alternatives in A (thus, no potential candidates can guaranteed themselves seats) (Greenberg and Shepsle 1987, 529).

88

essentially that no two or more alternatives can be chosen by the means of majority rule so that it could beat any other possible alternative (Greenberg and Shepsle 1987, 529). The implication is that multiparty systems are generally unstable. Once there is no equilibrium from which the main parties do not wish to deviate, there is always possible for an entry to join the competition and displace one of established parties. Contrary to Palfrey who argued that the entrant (should) maximize(s) votes, Greenberg and Shepsle (1987, 535) suggest that the entrant should maximize ranks so that it could relocate one of the major parties in the subsequent elections. In multiparty competition, established parties compete on two separate fronts: a competition for the central positions (voters), another for deterring any possible entries at extremes. This model points at the role played by the institutional or non-institutional barriers that thwart the danger of a new entry in competition. Those barriers could be institutional, such as high electoral thresholds, public finance for main (parliamentary) parties, or high numbers of registered members required for registration of a new political party. On the other hand, these barriers could be psychological, in the sense that established parties have also established reputations so that new parties could find difficult to compete with old parties (Greenberg and Shepsle 1987, 535). Furthermore, old parties are psychologically advantaged by the superior exposure to media communication. Other non-policy factors, such as partisan affiliation are also particularly important forms of political barriers. I shall come back to this very important issue in the next chapter. However, further formal research have shown that equilibrium in multiparty competition is not difficult to be obtained. Extending the previous research of Enelow and Hinich (1989), who prove that in a multidimensional two-party competition a sufficient condition for an equilibrium to exist is to model voting probabilistically, Lin, Enelow and Dorussen (1999) show that an equilibrium exists in multiparty competition if we allow for the same conditions and we impose the fairly unrestrictive condition that objective function of political parties is concave. In their model, a number of candidates M compete in a compact and convex subset X of a k-dimensional policy space Rk (M 2). All candidates are assumed to be votemaximizers and they choose a position in X accordingly. Voters have fixed policy

89

preferences and they vote probabilistically and sincerely in election. Each i of the voters has a utility for voting for party j, denoted Uj[i], which is a linear function of Dj[i], where: Dj[i] = Cj Vi, where Cj is the position of party j and Vi is the position of voter i. Thus, Dj[i] is the distance between voter i and party j. Given that we model a case of party competition under uncertainty, the utility for candidate j is given by: Uj[i] = Ej Dj[i], where Ej is a random element representing uncertainty about party j and which is independently distributed for all j.52 The authors show that if the expected vote for each party j is concave, then this is a condition for existence of global equilibrium, which always exists if Dj[i] is everywhere twice differentiable and it is unique if the expected vote for party j is strictly concave. They argue that in multicandidate probabilistic voting, with candidates maximizing the expected number of votes, and with expected vote function for every candidate being concave, the point that minimizes the sum of the distances from all voter ideal points is a convergent equilibrium if this point is unique (Lin, Enelow, and Doressen (1999, 70). Depending on how we measure these distances, the authors show that the equilibrium is at the mean position in all directions (with squared weighted Euclidian distance), at Ni=1 e[i] = 0, where e[i] is an unit vector pointing from the equilibrium to voter i ideal point (with simple weighted Euclidian distance), or at the median position (weighted city-block distance) (Lin, Enelow, and Doressen (1999, 73). The major finding of Lin, Enelow, and Doressen, but also of De Palma, Hong and Thisse (1990), is that convergence is always the equilibrium in multiparty competition if the level of uncertainty (about voter or party positions or both) is sufficiently high. The result of such convergent equilibrium is extremely robust, and the conditions necessary for such equilibrium require either a higher degree of uncertainty in party system, or sufficient low saliency of party policies compared with other voters non-policy motivations (Adams 1999, 260).

52

For the formal proof, refer directly to Lin, Enelow and Doressen (1999).

90

3.8. Conclusions

Whether political parties converge or diverge ideologically in spatial electoral competition has been one of the most fruitful debates in rational choice theory. Much of it gravitates around the location of equilibrium or equilibria in various institutional formats of party competition, which presumably shows that electoral competition follows a certain rationale and the result of competition is not a random process. In the previous pages, I have shown how the political equilibria change when we move from two-party to multiparty competition, from unidimensionality to multidimensionality of political space, from deterministic to probabilistic voting, and from perfect to imperfect information. One general finding of the literature has been that median (on single dimension) or the mean (on multiple dimensions and multiparty competition) positions exert a powerful influence on party strategies. In general, vote-maximizing parties have a dominant strategy in moving towards these points, under several other basic assumptions about citizens preferences and voters distribution. The models of probabilistic multiparty competition under uncertainty provide the theoretical foundations of this thesis. In the light of their predictions, I shall analyse the electoral party competition in the CEE. There are good reasons why I rely on these particular types of models to guide my analytical approach. As I argue in the next chapters, the level of political uncertainty is considerably higher in the CEE than in the established democracies, a phenomenon that is expected to influence the political process in the region. As I mentioned in the Chapter 1, this uncertainty of electoral process is the result of peculiarities of the politics in the region: the high number of competing parties, not very stable political links between parties and electorate which results in high electoral volatility, lack of internal resources and so on. In the next chapter, I come back discussing one inconsistency that comes out from the survey of literature on formal political competition: the conflicting predictions about the patterns of party competition which result when analysing the effects of uncertainty and entry. The models of political competition under uncertainty suggest that political parties

91

approach convergent pattern of competition, while models with entry argue that parties deviate from centre in order to thwart any potential entry at the extreme positions. I shall theoretically argue that political parties have a very powerful instrument against other potentially new competitors: they can build institutions to prevent new entries. In combination with the effects of political uncertainty and other peculiarities of politics in the region, the effects of institutional setting have produced a pattern of electoral competition that is noticeably convergent around median or mean voter location.

92

CHAPTER 4: INFORMATION AND ENTRY AS DETERMINANTS OF PARTY COMPETITION

4.1. Introduction
I have showed in the Chapter 3 that the pattern of electoral party competition depends largely on the interaction between the probability of entry in competition and the level of information political parties have about the game structure of party competition. The role of entry as a barrier for policy collusion has been emphasized by Palfrey (1984) for two-party competition and by Greenberg and Shepsle (1987) for multiparty competition. Their standard argument is that political parties avoid central or median positions because parties have to maintain their political support among non-central voters. On the other hand, the low level of political information about the structure of competition seems to drive the competition towards the centre of political dimension. Enelow and Hinich (1984, 1989), Hinich and Ordeshook (1970) or Lin, Enelow and Dorussen (1999) show that multiparty competition under a high level of uncertainty about party or voter positions becomes centripetal if parties behave as vote maximizers. The results seem contradictory and the goal of this chapter is to examine closer this apparent contradiction within the literature of spatial analysis of party competition. The aim is to discern how the danger of entry and the imperfect information influence the electoral strategies of political parties in multiparty competition. The argument of this chapter is that, given the influence of uncertainty, political parties approach convergence as the most probable winning strategy under low information. At the same time, they are likely to respond to the danger of entry by devising political institutions that minimize this source of political uncertainty. Methodologically, I analyze in this chapter the impact of these two variables on the dynamics of party competition by designing two simple models of three political parties competing in the elections under a set of Downsean assumptions. The results show that: 93

1. Under perfect information, when voters vote deterministically and under some limitations imposed to their electoral strategies, political parties approach convergence in multiparty competition, but formally they do not assume the same policy position; 2. Under imperfect information and with no restriction imposed on party strategies, political parties converge at MVP if the uncertainty is large enough; 3. Political institutions (the electoral system PR formulas and other provisions such as the electoral thresholds, public financing etc, and the coalition government) are likely to be important elements that allow the equilibrium outcome of electoral competition be at median voter position. 4. Political parties design institutions (electoral threshold, electoral systems which favour big parties, party financing laws etc) whose main goals are to diminish the level of uncertainty, by decreasing the potential number of players in competition. At the same time, these institutions decrease the probability of new entries in competition and allow the established political parties to compete centripetally.

4.2. The Models

Consider an electoral competition taking place on one policy dimension, with voters uniformly distributed along it.53 Every voter i has an ideal policy position on this dimension, and her utility function is given by the formula: U(i) = - (Xi - Sk)2,

(4.1.),

where Xi represents the position of voter i on that policy dimension, and Sk represents the position of party k on that policy dimension. Each voter has a set of preferences that are reflexive, complete and transitive. According to the utility function (4.1.), every voter has a set of policy preferences that is single-peaked and her utility function is symmetrically increasing and decreasing to the left and to the right of her ideal policy point. From the

I choose this assumption about voter distribution because it is the least favourable for policy convergence, because the number of extreme voters equals the number of median voters.

53

94

voters utility function, we may infer that every citizen evaluates identically two policy positions located at a distance from her ideal policy position. See, for instance, the Figure 4.1 which depicts two policy programs, X1 and X2 located symmetrically to the left and to the right of the ideal position of voter Vi. Given that individual utility function is symmetrically increasing and decreasing over the policy space, voter Vi is indifferent between these two alternatives. At the same time, another condition I impose is that every voter votes in the election according to her ordered set of preferences. Thus, every voter votes sincere (strategic voting is not allowed).

X1

Vi

X2

Figure 4.1: Symmetry and voters choice

In order to analyse the implications of uncertainty and entry, I start by considering the simplest case of multiparty competition, in which three political parties XA, XB, and XC compete. Every party is interested in obtaining the largest possible share of votes. All three parties know that a fourth party can not enter in electoral competition and every party knows that the other two parties also know the same information. Also, every party evaluates neutrally the policy positions from the dimension, which means that, ceteris paribus, every position of the dimension may be chosen by parties to be advocated to voters. This condition suggests that parties are not ideologically oriented and they choose the policy position that maximizes the number of votes they receive in election. Additionally, I consider that parties face no other cost than losing votes when they change policy positions. Another assumption is that political parties can not leap-frog ideologically (they can not pass over the policy position of other party) because they would lose credibility in the eyes of voters and by consequences they would end up losing votes.

95

In what concerns the level of information, we assume that there is a competition in which the actors have perfect information. Every party knows its position on the policy dimension, it knows the position of other parties and also it knows the distribution of voters on this political dimension. Additionally, every voter knows her ideal policy position on this dimension and the positions advertised by political parties that compete in elections. Given these preliminary assumptions, we can now analyse the dynamics of party competition in a multiparty setting, when parties are vote-maximizers, voters have singlepeaked policy preferences, and parties and voters are fully informed. Consider three parties (X1,X2,X3) taking initially three policy positions (j, k, and l) on this policy dimension which, without the loss of generalization have the following distribution:

j<k<l

(4.2),

where j, k, l are three policy points from the policy dimension.54 At this point, I shall not discuss the positions of parties regarding the MVP, but rather I shall concentrate on the political dynamics of those three parties. Later on, when I discuss the implications of the model, I shall take into consideration the influence of the MVP for the inter-party competition. Consider a political space with the characteristics mentioned above (uniform distribution of voters, with all voters voting in the elections) and three parties which take policy positions according to formula (4.2). I show now that under these conditions, there is an equilibrium outcome and this equilibrium is given by the position of party X2, although parties do not fully converge at the location of party X2. Proof: Consider a policy space that takes values over the interval [0;1]. If the condition enounced by the formula (4.2) holds, then we have the following political sub-spaces for each individual party: a. For party X1: [0; (k-j)/2)
54

Inequality (2) says that party X1 takes initially position j (and thus is the left party), party X2 takes initially position k and is the centre party, and party X3 occupies the policy position l and is the right party in competition.

96

b. For party X2: ((k-j) / 2; (l-k)/2) c. For party X3: ((l-k)/2; 1].55 Given the condition I set that parties can not leap-frog the positions of other parties in competition, it is evident that party X2 has no unilateral strategy that allows an improvement in the number of votes. A shift to the left (in the direction of party X1) gives to X2 a loss of votes in favour of party X3.56 However, because the all voters vote deterministically, parties X1 and X3 do not lose votes at the extremes of the political space.57 Even though the policy distance increases, the extreme voters still prefer the positions advocated by X1 and, respectively X3. Thus, parties X1 and X3 can improve the electoral score by moving near the position of party X2. This situation in which political parties agglomerate around the position of central party is an equilibrium. Note that formally political parties take divergent positions since they do not fully converge at the position k. If all parties assumed the position of X2, they would obtain 1/3 of all votes each. But the extreme parties would be better off if they come close to, but not at, the position of party X2. Yet, the strategies are notably convergent, as political parties move closer towards each other. We can see that under several conditions, the findings of Hoteling/Downs model of two-party competition partially hold. Even in multiparty competition political parties may offer quite similar policy programs in order to attract votes from the electoral body. However, it seems that the findings of the model contradict one assertion of Downs, which suggests that in multiparty systems parties have incentives to differentiate themselves, either to better individualize in the context of party competition or to avoid a loss at the extremes in favour of new entry. Under the conditions formulated above, the political parties have incentives to move closer to each other, although, as Downs predicts, they do not fully converge. Also, while the above model finds a political equilibrium around the point of middle party, it does not imply that the middle party takes median position on that
55

According to the assumptions I have made about the individual voter functions, voters with ideal policy positions inside these intervals will vote for the respective parties. At the intersection of these spaces, voters are indifferent to the alternatives, so that no party occupies those positions. 56 Recall the condition that voters are uniformly distributed on the political dimension. 57 For distinction between deterministic and probabilistic vote, see Hinich and Ordeshook (1970). I choose the case of deterministic vote because it is assumed to be less favourable to party convergence, unless very restrictive conditions are applied to voters utility functions. Enelow and Hinich (1984) point to the importance of probabilistic vote for ideological convergence of parties, both in two- and multiparty competition.

97

dimension. So, the equilibrium point and the position of median voter might not be the same. It is relatively simple to prove this statement. Look at the inequality (4.2) and imagine a particular case when the position of median voter differs from the position of the middle party, X2. According to the argument I gave for a more general situation, X2 has no incentive to modify its initial position as it can not improve its electoral fortune. Because voters are uniformly distributed along the policy dimension, the party X2 is constrained in competition by the positions of the extreme parties. Since we have formulated the condition that parties do not leapfrog, X2 has not available an individual strategy to make it better off if it shifts. Thus, X2 will be indifferent between any two policy positions included in the interval delimited by the extreme parties and it will not move towards the position of median voter if the party plays a strategy oriented only towards maximizing its electoral result. Instead, the other two parties, which take initially more extreme positions, move towards the position of party X2. So, none of the parties attaches any electoral value to the point of median voter. However, it may happen to see such competition in equilibrium at the MVP if we extend the discussion beyond electoral competition. Often it is forgotten that political institutions matter in party competition. Not only that they regulate the political process, but they also shape the incentives of political parties in electoral competition. Take for instance the case of centre party, X2. It knows that it can not improve its electoral fortune during the electoral period because it is in a situation of restricted competition. But at the same time, it knows that electoral competition is only one of arenas of interparty competition. Parties, for instance, may win few votes in election but at the same time they may have a favourable position that allows them to negotiate offices in the government. If elections are regulated by an electoral system based on proportional representation, the centre party knows that even though its electoral hopes are doomed, there are some political opportunities located beyond electoral arena. The most evident examples of such opportunities created by political institutions within the political system are offices in government. Although the party can not obtain many votes, it has hopes that under a PR electoral system it may obtain other political revenues.

98

However, in order to remain confined in the limits of spatial theory, it is necessary to show how such political institutions (which offer political opportunities outside of electoral arena) shape the electoral behaviour of parties. Coming back to my initial example, if the party is office-seeking party, then X2 has incentives to make sure that no party is going to win the elections with qualified majority that allows it to form the government alone,. Thus, under a PR electoral system, it has to occupy for itself sufficient political space (and to locate in a strategic position) so that no other party will obtain more than 50% of the votes. But the party can prevent other parties to win a majority of votes only if it takes on the position of median voter.58 By doing so, no other party will win a majority of votes and presumably there will be negotiations among political parties to form a coalition government. Here, the pivotal position of X2 may allow the party to play a more important role than in electoral competition. This first model of competition offers us some interesting conclusions. One is that although an equilibrium exists with parties offering divergent policy positions, party strategies are to certain way convergent: parties agglomerate the policy space around the position of middle party. Yet, this equilibrium is largely the result of the condition that political parties can not leapfrog. If we eliminate this conclusion, then no convergent equilibrium occurs since the party X2 will cross over the position of one extreme party in order to avoid the centre. But even under this assumption, equilibrium may occur at any point of the dimension, because the party X2 may assume any policy position on it. Thus, as emphasized in the literature, the models of party competition under perfect information and deterministic voting are generally unstable and offer equilibria only under very restrictive conditions. Under uncertainty and probabilistic voting, things are substantially different. Consider now the same three political parties competing on the same policy dimension
58

I have previously argued that position of centre party becomes the Nash equilibrium of the party competition, under the assumptions of the model I used. But if the centre party moves to the position of median voter, then the extreme parties have to move as well near the position of median voter. Ceteris paribus, none of the players would like to deviate from their positions (recall that parties can not leap-frog), which is then the new Nash equilibrium. Otherwise, the assumption we impose in the model that no party can cross over the position of other party would be violated. Also, following Downs argument, political parties can not leapfrog because they need to develop ideologies and reputation and thus they should be consistent in developing policies over time (Downs 1957: 97). However, there is an additional condition that is necessary to have equilibrium at the position of median voter: the median position should be located between the positions of extreme parties. Otherwise, the equilibrium position remains the location of middle party.

99

delimited by the closed interval [0;1]. Again we assume that voters are uniformly distributed along the axis of competition, so that one voter takes position at every 0.1 division of the dimension. Thus, there are 11 voters, with the voter number 5 being the taking the median/mean position. Contrary to the previous example, we model now the utility of voters not only as a function of party-voter proximity, but to include two additional components. The first component is the probability of voting in the respective elections given the partys policy position, P, which takes values between 0 and 1. Also, every voter i has an unknown parameter ik that includes party ks non-policy evaluations, such as candidate evaluations, political performance of the party and so forth. Thus, the voters utility for voting for party k changes to: Ui(k) = - Pi(Xi - Sk)2 + ik,

(4.3)

Political parties are unaware about the real values of Pi and ik. Thus, if voters have utility functions similar of (4.3), if voters are uniformly distributed along the policy dimension and if political parties are interested to obtain as many votes as possible, we can show that there is an equilibrium outcome at the point of median voter.

Proof

If any voter i evaluates the parties A, B, and C according to the equation (4.3), then the utilities provided by parties to the individual i are: Ui(A) = - Pi(Xi SA)2 + iA, Ui(B) = - Pi(Xi SB)2 + iB, Ui(C) = - Pi(Xi SC)2 + iC, where, Xi is the ideal policy position of voter i, Pi is the probability that voter i votes in the elections, SA,SB and SC are the positions of parties A, B and C, and iA, iB, iC are the unknown parameters of voter i for parties A, B and C. From equation (4.3), we infer that voter i prefers party A to party B if and only if Ui(A) > Ui(B), which means that:

100

- Pi(Xi SA)2 + iA > - Pi(Xi SB)2 + iB (Xi SA)2 + iA > (Xi SB)2+ iB

(4.4)

From (4.3) and (4.4), we observe that voters choice is now a function of the policy distance between voters ideal policy position and party position and of the unknown parameters Pi and Ei. Since voters are uniformly distributed along the axis of competition, the utility of party A when it takes the policy position S is:

10

UA(S) =

P (X
i i =0

S A ) 2 + EiA

For the sake of simplicity, assume that Pi and iA have the same values for all i, but political parties do not know these parameters. Then, UA(S) is maximized when its derivative equals 0. Thus:
10

max UA(S): UA(S) = 0 => ( Pi ( X i S A ) 2 + EiA ) = 0

i =0

Performing a simple calculus, we have:

10 i =0

10

2 Pi ( X i S A ) = 0 ( X i S A ) = 0
i =0

(4.5)

But (4.5) is true only if SA = 5, which is the median/mean voter. Also, from:
10

UB(S) = Pi ( X i S B ) 2 + EiB
i =0 10

UC(S) = Pi ( X i S C ) 2 + EiC
i =0

we may show, using a similar calculus, that UB(S) and UC(S) are maximized when SB and SC are located at the MVP, under the condition enounced above. If party utilities are maximized when SA, SB, Sc equal 5, then the position of median voter represents the equilibrium point in competition.

101

It is relatively simple to see why. Contrary to the case of deterministic voting with perfect information, now the vote is conditioned not only by the party position but also by the unknown parameters Pi and Ei. Thus, a shift from the MVP to any direction does not bring necessarily additional votes, as in the case of competition under full information and with deterministic voting. Since parties are uncertain about Pi and Ei numbers, they are also uncertain about the expected utilities given by any other position. At the same time, what every party knows is that median or mean position minimizes the policy distance between party and all voters. Thus, any political party interested to win votes, but unaware of other relevant parameters, should take median/mean position.

4.3. The Implications of the Models: How Information and Institutions Shape Electoral Competition in Multiparty Settings?
The models from the previous section bring out two findings that have theoretical value. The first finding is that political parties have incentive to collide ideologically even in multiparty elections. The extent of convergence depends largely of the level of information. Under perfect information, even though parties do not take positions exactly at the MVP, they nevertheless tend to adopt very similar policy strategies.59 But this outcome is very unstable and it may be obtained only under very restrictive assumptions about party behaviours. By contrary, when we model party strategies under uncertainty, not only that we get a stable equilibrium at the MVP but we do not need to impose any restriction on party strategies. In fact, the models confirm what the literature emphasizes: equilibria are much more stable in probabilistic voting models. The second finding is that Downs is partially right: the position of median voter fades in multiparty voting models under perfect information (unless, of course, the very special case when the party from the middle takes exactly the median voter position).
Technically, the extreme parties come very close to central parties but they do not take the same policy positions. Thus, mathematically, the parties will take distinct positions but very close each other and all around the MVP. I think that interpreting the results purely with a mathematical eye is not a useful strategy for understanding party strategies. Thus, I shall rather pay attention here to the dynamics of electoral competition which is far more illuminating than a pure mathematical interpretation.
59

102

However, as I have shown above, the political institutions modify the behaviour of parties in electoral competition, producing a different equilibrium outcome, even though a pure formal approach does not reveal this equilibrium point. More concretely, institutions modify the incentives of median party that has good reasons to occupy for itself the position of median voter, which in the end represents the equilibrium outcome of the electoral competition.60 But in probabilistic multiparty voting models under uncertainty, centripetal competition is obtained exclusively as result of party uncertainty and we do not need ad-hoc explanations: to minimize the average of absolute policy distance between political party and voters, any party should place it at the mean/median position. There is also a third finding which has both theoretical and political implications: centre party is always losing under perfect information. The political space of the centre party is limited at the space between the extreme parties. The centre party has a sorrow fate as it can not improve its electoral score by making individual decisions. Rather it is in the logic of the multiparty political competition to have the centre party under the constant fear of losing votes in favour of more extreme parties. It is probably only a matter of luck that central party can improve its electoral fortune when information is perfect. It is interesting to note that, in the light of the first model of party competition I advance above, the information does not appear as necessary condition for ideological collusion. Even if we eliminate the assumption of uncertainty, political parties still appear to collide ideologically in multiparty competition. However, considering more carefully the results of the first model, we observe that uncertainty is essential for obtaining convergence in multiparty electoral competition. In a three-party competition under perfect information I obtained a stable equilibrium given the assumption that political parties do not leapfrog. Under this assumption, the competition is very similar to competition between two political parties since only two players have meaningful electoral strategies. However, eliminating the interdiction of leapfrogging the chaos is likely to characterize the party competition, since political parties would crossing continuously the position of other parties in order to avoid the centre of political dimension. Political parties will try to differentiate each other and no stable equilibrium is likely to occur. The same pattern holds for four-party

60

Recall the first model in the previous section of this chapter.

103

competition under certainty too: the political parties from the centre will leapfrog the extreme parties hoping to avoid the political centre. However, the strategic interactions under uncertainty lead political parties at the median position, as I have shown in the second model of party competition. I modelled the uncertainty in two ways here: as a probability of voting and as uncertainty about voters evaluation of other non-policy relevant aspects. But uncertainty may take multiple aspects. One type of uncertainty the political parties face in competition regards voters policy positions, when parties are ignorant about the real policy preferences of voters. Total ignorance is improbable since polling agencies regularly survey the policy preferences of voters, but at the same time some degree of uncertainty is inevitable in politics, particularly when voters have preferences over many issues. Another type of uncertainty regards the policy position of other political parties, particularly because political messages are often blurred intentionally so that they cover a wide range of policy positions. A third type of uncertainty relates to the number of political parties that compete in elections. In this third way, the information of parties and voters could be distorted by increasing the number of players that compete in the system. With the number of parties increasing, the future party positions of opponents are much more difficult to be foreseen by the parties themselves. Also, voters will be less capable to locate every party on the salient issues, which could determine them to choose politically relying on non-ideological issues.61 But there is another type of uncertainty that has detrimental effects on party convergence. As I mention in the Chapter 3, some rational choice theorists argue that the danger of new entry in competition makes the political parties propose distinct political ideologies in elections. This potential new entry is nothing else than a source of uncertainty of type three I have mentioned above. If the probability of a new entry is sufficient high, then extreme parties need to take more extreme policy positions in order to avoid the electoral fate of the centre party. But then, they face another dilemma. Under favourable electoral systems, such as PR, which is the case with the most part of political systems with multiparty system, an entry at the extreme could win relatively easy some political representation. At the same time, if the established parties take on more extreme positions

61

Partisan affiliation, charisma of leaders, clientelistic relations could be the main factors affecting voters decision in such context.

104

to avoid the danger of entry, they lose votes in favour of centre party(is) and make the centre becoming more important from electoral point of view. How are political parties likely to respond to these types of uncertainty in politics? Facing the uncertainty, they would probably try to make some order in the chaos resulting from increasing number of political interactions. When the uncertainty about the electoral results is very high, one good strategy for parties is to set policy programs which are very similar and to turn to non-policy party competition. If the number of parties is sufficient high, voters have no incentives to invest time in analyzing the policy differences between parties.62 Also, the political shortcuts become meaningless from the point of view of policy appeals. With many parties adopting labels as left, right, liberal or social democrat etc, voters face almost the same level of uncertainty about the real, substantive policy proposals of political parties. Political parties have few, if any, available strategies to tackle the low information they have about party preferences or about the positions of other competitors. This uncertainty may have systemic causes such as low institutionalization of parties and party systems, weak party-voters linkages, limited knowledge of voters about political process and many other plausible factors. Also, such uncertainty is more likely to appear in new democracies where voters and parties do not have behind a long of party behaviour or ideological commitments of parties and voters. Thus, political parties are likely to respond to this type of uncertainty by taking policy positions that are very similar to the mean or median voter instead of risking more extreme policy appeals. However, the uncertainty generated by the potential new entry in competition is easier to be challenged. Contrary to what the literature usually emphasizes, parties are not likely to challenge the entry by making use only of electoral strategies. By contrary, they have better instruments to eliminate the danger of the entry and perhaps even some of the smaller established parties. These are the political institutions that offer instruments and channels (electoral thresholds, state finances and media access) to parliamentary political parties to attempt the elimination of undesirable surplus from political competition. The established parties use these political institutions to decrease this category of uncertainty by setting, for instance, high electoral thresholds which affect mainly smaller and newer
62

This is an argument which holds when the number of parties is also small.

105

parties. Or, they may pass legislation that allows them to appropriate key political resources.63 These may be state subsidies for party finance that have become one of the most important political phenomena in the last decades in Europe or privileged access to media coverage for established political parties. Moreover, the uncertainty about voter and party policy positions works together with political institutions against the probability of entry. Consider the second model of party competition, when the voters utility function includes the unknown parameters Eik and Pi. Let assume that that a fourth party D would like to enter in competition. Not only that it need to overcome a certain electoral threshold, but it also need to take into consideration the unknown parameters of voters. Taking on a policy position different than the MVP (position 5 in my example above) does not offer necessarily the support of extreme parties because the electoral support depends also of Eik and Pi. The established parties may have an advantage over the entry in this respect, because they are more know to the voters or because voters may have non-policy reasons to vote for the parties, including here psychological attachments to parties. Thus, combined effects of uncertainty and institutions form serious barriers to new entries, which positively influence the party convergence of established parties. Uncertainty plays a crucial role in politics; this is probably the greatest legacy of Anthony Downs in political science. The uncertainty is undesirable for political actors because it is a source of strategic errors they can make in such environment. Thus, political actors are likely to attempt to minimize it, if possible, by designing political institutions that bring a certain degree of predictability into political system. For my purposes, the substantial argument of this chapter has been that political parties are likely to design political institutions that decrease the level of uncertainty by reducing the number of political parties within the party system and by reducing the probability of new parties to enter in political competition. The combined effects of these institutions with the political uncertainty emerging from other social or political phenomena produce a competition that is likely to be convergent, a hypothesis that I shall test in the following chapters.

Quite recent debates in the literature of political parties talk about cartelization of party systems, as a tentative of established parties to avoid the uncertainty of competition. Cartelization suggests that competition among parties has become much smoother, because parties can survive as major players even though they lose elections (see Katz and Mair: 1995, Koole: 1996).

63

106

CHAPTER 5: HYPOTHESES, METHODOLOGY AND DATA

5.1. Introduction
In the Chapters 3 and 4, I have extensively surveyed the literature on formal party competition in order to obtain some theoretical understandings about the dynamics of party competition in electoral arena. As we could have seen previously, there are numerous formal models highlighting that the dynamics of party competition follow somehow different logic when we deal with party competition between two or multiple parties, or when we talk about one or multiple salient dimensions, or when competition among parties takes place into an environment characterized by perfect or imperfect competition and so forth. Since formal models are very flexible instruments of research, it is no surprise that models of political competition have mushroomed since Anthony Downs published An Economic Theory of Democracy. Almost all refinements of the theoretical framework I make use in this thesis make reference to a very powerful concept in political science: the Median Voter Theorem (MVT). The prediction of this core approach is that political parties adopt similar, centrist ideological positions in elections in order to maximize their political gains and that citizens (voters) are denied the possibility of a real policy choice. This prediction, which I refine it in the following pages making use of contextual data from the CEE politics, is going to be tested as the main hypothesis of this thesis. However, extensive analysis of party manifestos in the west European elections show that this prediction is empirically flawed because parties do not converge at the position of median voter (Laver and Budge 1992, Klingemann et al 1994). Several explanations have been advanced for why this prediction fails in the Western Europe, even in contexts that seemed very favourable for party convergence, such as the British political system. Some of them I reviewed in the Chapter 3, where I talked about the critiques of Downsean model of political competition. In this regard, the most influential factors that

107

drive political parties away from the centre seem to be the partisan affiliation of voters and the close political linkage between political parties and the social structure of the electorate (Adams, Merrill and Grofman 2005). Thus, the obvious question is why we should expect political convergence to occur in the Central and East European elections? As I already outlined in the Introduction and I shall further develop in the following pages, there are the peculiarities of the CEE politics that make the difference. These peculiarities (uncertainty of political process, scarce political resources and blocking institutions) make political parties in the region behave more intensely as vote-seekers in elections than political parties in the Western Europe. Since we know from the theory of spatial party competition that ideological convergence is the outcome of competition among self-interested political parties with imperfect information, I expect to observe such phenomenon occurring in the CEE elections. There are very important methodological and theoretical arguments for why testing the theoretical insights from the Chapters 3 and 4 is essential. Very often, the use of rational choice models to study politics has been criticised for no real scientific value. The detractors of rational choice argue that models remain only abstract, mind-stretching constructs with very limited influence on an empirical science of politics. This is for instance the core argument of Green and Shapiro (1994) against the rational choice theory. Partly, Green and Shapiro are right: quite often formal theories are not confronted with the empirical developments of a particular political phenomenon. However, no rational choice theorist would agree that formal theories offer no information at all about real politics, unless Green and Shapiro mean by information sheer facts about particular political processes. Cox (1999), for instance, argues that there is an empirically successful rational choice theory, ignored by Green and Shapiro because they are biased when selecting parts of the literature with less successful empirical contents. By confronting the formal theories with empirical phenomena, we may provide arguments that the scientific scepticism which surrounds rational choice theory is not justified. There is in the aim of this and the next chapters to perform such undertaking. Although it has been usually said that rational choice theory functions well only in environments that are well structured, I argue implicitly in this thesis that the CEE represents a case where we can meaningfully test some hypotheses driven from the formal

108

theories about of party behaviour in electoral competition. Like in any other established democracy, political parties in the region perform the same political functions and the difference in this respect is rather a matter of degree than of substance. Since rational choice theory is a universalistic theory which assumes that behaviour of individuals is intentional, there is no a priori reason why we could not apply this analytical framework to party competition in the CEE. In this chapter I proceed as follow. First, I highlight the peculiarities of CEE politics, already mentioned in the Chapter 1, that are likely to have an influence on party competition. In the second part I advance three main propositions, based on the theoretical arguments from the Chapters 3 and 4, which I test in four countries from the CEE: the Czech Republic, Hungary, Poland, and Romania. In the third part, I present the data I use in the empirical analysis, how it has been obtained and what are the strength and limits of the methodology I rely on.

5.2. Rebuilding the Ship at Sea: Party System and Political Parties in New Democracies of CEE
At the beginning of 1990s, the enthusiasm that had accompanied the fall of communist regimes in 1989 faded as the difficulties of political transition and economic change became noticeable. Many of the students of political transition and democratization, usually called transitologists after the domain they studied, argued that democracy and democratization may not be a successful story in the CEE, as the countries in the region had to cope with three major processes simultaneously: 1. The political liberalization, meaning the institutionalization of political competition as the main method of selecting the leadership; 2. The economic transformation, that is the withdrawal of the state from the economy and the creation of market economy.

109

3. In some cases, these two processes were accompanied by nation-building process, with new states emerging at the beginning of 1990s from the dissolution of the USSR, Czechoslovakia or former Yugoslavia. The difficulties of these three transformations brought pessimism in the community of scholars of democratic transition (Ekiert 1991, Plattner and Dimond 1995, ODonnell 1995, Rose 1997). In a challenging book that inspired the title of this subsection, Elster et al (1997) argue that the immense effort of the CEE countries to deal simultaneously with these three processes makes the CEE countries similar to a ship that has to be repaired while it is still floating. Obviously, the scepticism about the democratic future in the region has brought about numerous political metaphors to describe the peculiarities of the new political democracies. Disregarding those metaphors, there is much true in the claim that transition to democracy in the CEE has been very different than other similar political phenomena in the Western Europe or North America. First of all, the extent of societal transformation, which included political, social and economic spheres, has been unprecedented in the history of political change. Second, the democratic system had to be implemented in a region where the democratic ideas and practices had not been present at all in the past. Moreover, as Elster et al (1998, 10-11) point out, the former communist ruling elite was not totally discredited as it happened with the Nazis and Fascists in Germany and Italy after the 2nd World War whereas the social solidarity common in these countries after the war was attained only for short period in the CEE countries. The CEE has been regarded as peculiar because of the historical and societal differences from the west European and North American countries that became democratic in the 19th-20th centuries. Very often, these peculiarities of the CEE compared to the established democracies have been formulated in terms of differences in modernization of societies, that is differences in social, economic and political structures. Steamed by such arguments, one of the most debated questions in the party politics literature has been what kind of political competition is likely to occur in the CEE? Is competition likely to be structured by any enduring social conflict resulting from different interest articulation of citizens or the competition among political parties will be fluid, driven by political personalities?

110

Three different paths, summarised extensively in Evans and Whitefield (1993), have been hypothesized as probable evolution of party politics in the region. The first prediction has been a politics substantially different than in the western democracies because of the missing link between the society and political parties. The argument emphasises the lack of social differentiation and the absence of social institutions from which voters could coherently formulate their political interests in post-communism. Several factors are held to be responsible for the tabula rasa in the CEE politics. One factor was the total lack of competitiveness of communist regimes which demanded a total submission of individuals to an official ideology. The total control of ideas in the communist politics and society produced social and political passiveness that were likely to have consequences after the break of communism. Also, the communist dictatorships were political systems without social classes; they produced socially homogeneous societies from which little political and social differentiation could have occurred immediately after the end of the communism. Moreover, the social institutions that in the West have important attributes in social and political articulation of interests, such as labour unions or Church, were either strictly controlled ideologically or suppressed during the communism. Not least, the anti-politics feeling on which anticommunist intelligentsia based their political appeals against the communist regimes in many of the CEE countries could produced very low political mobilization and interest articulation during the early transition. For such reasons, the supporters of this missing middle perspective64 conclude that parties and party systems in the CEE are a very different species that is characterised by fluid or open party attachments (Mair 1997). The second perspective clashes with the predictions of tabula rasa. This perspective is the intellectual outcome of modernization theory and argues that party competition in the CEE will approach, under the pressure of market forces, the pattern we notice in the western democracies. Kitschelt (1995, 451) argues that tabula rasa hypothesis is too conservative and pessimistic and political parties in some countries of CEE may develop programmatic appeals similar to those used by political parties in the western democracies. He argues that market pressure and economic change are likely to produce political divisions around distributive policies that will determine political parties to take positions
64

A term originating in Evans and Whitefield (1993, 530).

111

on a dimension similar to the left-right in the west European context (Kitschelt 1995, 458). These socio-economic bases of party competition will occur mainly as a result market differentiation. The young, better educated persons, living in the urban areas will become more adapted to the new market economy and they are likely to be associated with the winners of economic transition. In contrast, large groups of peasants, heavy industry workers, pensioners etc will become disadvantaged by the market differentiation. Thus, they are likely to become supporters of populist parties or of those parties that appeal for greater redistribution of wealth within society. With the time, this differentiation determined by the personal position within the market economy will reinforce the political behaviour of groups and the social linkages of parties with voters. Though Kitschelt acknowledges that other non-economic factors might influence the structure of party competition (ethnicity, religion or nationality), he seems to imply that the effects of these factors will coincide with the divisions produced by economic reform (Kitschelt 1995, but also Evans and Whitefield 1993, 537-39). For these reasons, a third possible evolution of party systems has been advanced to take account of other factors that might provide an ideological base for party competition. Evans and Whitefield (1993) advance another explanation which differs from tabula rasa argument and modernization theory. They argue that in some countries the conflict around ethnic (like Estonia, for instance) or other non-economic issues will inhibit the party competition on economic policies. Conversely, in other countries that are confronted with economic hardship resulting from failing economic reforms, no social or economic factors are likely to form the bases of political competition among parties. Political parties will compete through personalities and valence issues, voters will vote according to the effectiveness of politicians to deliver the public goods and no distinct social or ideological dimension will structure the relationship between voters and political parties (Evans and Whitefield 1993, 538). The tabula rasa argument has several pitfalls. It makes too far-fetched assumptions about the nature of party-voter linkages by arguing there is no structure at all for political competition. Communist regimes aimed indeed at building socially homogeneous societies, using tight political and economic control to reinforce this political utopia. However, there is highly improbable that they succeeded to eliminate all sources of social and political

112

differentiation in the CEE. Studies in voting behaviour showed that such pessimistic assumptions about politics in the region were not at all accurate. Voters have been able, at least in some countries, to find structural sources of social and political differentiation quite early in the transition period (Evans and Whitefield 1995). Whether these structural sources have been of economic type, as Kitschelt suggests, or have been more varied, as Evans and Whitefield argue, remains a question that it hardly could be solved here. If some sort of social differentiation of interests occurred in transition, one important question is whether and how much it influenced the centripetal or divergent pattern of party competition? Social differentiation in itself is neither favourable nor inauspicious to political convergence in electoral competition. Since in rational choice we treat preferences as exogenous (individual preferences are given), there is of little importance for the spatial theory how voters formulate their political preferences. Moreover, the existence of social differentiation of voters is implicitly assumed in their different policy positions. Voter positions on economic dimension, for instance, should reflect to large extent their social or economic status, even though inside models of party competition these specifications are not included. It is very plausible that a Catholic voter extracts her policy positions out of the social experience inside her social group. Thus, a socially structured voting behaviour is a priori neither favourable nor discouraging for ideological convergence of political parties. Social structure may matter in this regard if it involves strong political cleavages, like ones Lipset and Rokkan define, or strong partisan affiliation. It is very appealing to interpret party politics in the CEE in the elegant framework of political cleavages that Lipset and Rokkan provide for western European politics. Some authors have repeatedly argued against such interpretation (White et al 1997, Elster et al 1998), while others support such analytical framework, at least in what regards some countries from the CEE (Whitefield 2002). The former argument points at the limited social differentiation at the beginning of transition period or at the lack of mezzo-structures which could encapsulate voters along the cleavage lines. The later suggested that some social differentiation occurred in the transition, as a result of economic or social transformations, and this differentiation shaped the voting behaviour of the citizens.

113

However, the answer to the question whether cleavages have structured or not the political competition in the CEE is largely a matter of definition. Many of the arguments claiming the existence of cleavage politics in the CEE either define political cleavage very loosely or use the term erroneously. In most of the cases, authors use the concept of cleavage to denote the existence of simple political divisions (Agh 1998). These divisions do shape political competition but they represent only transitory political divisions and are far from the concept of cleavages in the classical Rokkean understanding. Or, at best, the authors that emphasize the role of cleavage politics use a looser definition, pointing to [] a pattern of political competition embedded in cognitive, emotive or social structures of the citizenry as opposed to one determined by day-to-day issues, evaluation of government performance and personalities (Enyedi 2005, 698). However, if we talk about cleavage structure in the classical sense, then disagreement about the role of cleavages in explaining CEE party politics is futile: it is too early for political cleavages to crystallize, the political parties are still unstable political organizations, electoral volatility is high, the voter turnout is declining and the electoral strategies are oriented toward persuasion rather than mobilization. The effects of partisanship on ideological convergence or divergence, on the other hand, are more difficult to assess because of theoretical inconsistencies. As I show in the Chapter 3, Adams (2001) and Adams, Merrill, and Grofman (2005) prove formally and empirically the detrimental effects of partisanship for political convergence. On the other hand, McKelvey (1975) demonstrates formally that existence of party supporters and policy voting determines political parties to converge ideologically.65 At the same time, some authors suggest that the concept of partisanship, originating from American political context, has less explanatory power in the European context of multiparty competition, where ideology and cleavages have been the salient factors in voting behaviour (Budge, Crewe, and Farlie 1976). Also, while the first generation of articles on partisanship argued that partisan affiliation had roots in early socialization process, it is passed through generations (so called symbolic politics) and gives stability to political choice, newer approaches argue that partisanship is not an exogenous force organizing political behaviour,

65

McKelvey does not use explicitly the concept of partisanship but the context suggests it.

114

is less stable and depends less of early socialization than of political experience (Franklin and Jackson 1983). Because of these contradictory arguments, it is difficult to assess the impact of partisan attitudes on party electoral strategies and I need to leave this argument aside. However, as I shall argue bellow, the partisanship has favourable effects for centripetal party competition because it discourages potential new entries in electoral competition. New political parties may find difficult to compete exclusively on policy bases with political parties that have already developed non-policy links with voters.66 Irrespective to the existence of social links, there is little doubt that political parties in the CEE were at the beginning of political transition political entities created around political leaders. In many instances, those leaders were either members of former communist party elites or members of anti-communist dissidence. Such pattern of emergence of political parties in the region has had important political consequences. Since the parties have been creation of leaders, their power over organization has been noteworthy and in any case higher than in western established democracies. At the beginning of transition, many authors pointed to this peculiarity of CEE political parties (Kopecky 1995, 528-29, Olson 1998, 445-47, Szczerbiak 2001, 105-106). An indirect indicator of this pre-eminence is also the small membership of political parties (see Table 5.1 below). Large party membership has been considered in the literature a measure of influence of ordinary members and activists over the party leadership, and hence it has been considered that mass parties give sounder voice to ordinary members than cadre or catch-all parties. The number of party members is generally lower in the CEE than in the Western Europe as the Table 5.1 reveals. Even in the Western Europe the general trend in party membership is declining after 1970 and this phenomenon is associated with the end of ideological politics and the decline of mass parties (Mair and van Biezen 2001, Dalton and Wattenberg 2002, Webb et al 2003, Mair et al 2004). Furthermore, this steep downward slope of party membership in advanced democracies is accompanied by other complementary political phenomena, such as increasing public financing of political

66

See also the discussion at the end of Chapter 4.

115

parties, professional politics, capital-intensive political campaigning etc (Katz and Mair: 1995). The low party membership in CEE countries is a disturbing political phenomenon for the normative analyses of political democracies. However, together with the leadership predominance, it has important consequences for electoral strategies of political parties. Since parties lack the possibility of extracting political resources from party organization (such as party activists, money for political campaigns or for routine party activities between electoral terms etc), they need to look elsewhere after them. Some of these resources are coming from the public financing of political parties, but I argue that the most important resources, financial but not only,67 are coming from controlling public offices. This need of controlling public offices reinforces the political authority of leadership over the party organization and party members, who have few means to oppose the political leaders who do not conform to their policy expectations. Parties as cohesive political units are determined to behave electorally as vote-seekers in order to gain access to these indispensable political resources.
Table 5.1: Party membership in the CEE in comparison
Country Czech Republic Hungary Poland Greece Year 1993 1999 1990 1999 2000 1980 1990 1998 1980 1989 1998 1978 1988 1999 1980 1989 1998 Electorate 7,738,981 8,116,836 7,824,118 8,062,708 28,409,054 7,059,778 8,050,658 8,862,018 3,776,300 3,941,499 3,993,099 34,394,378 36,977,321 39,215,743 41,095,490 43,180,573 43,818,324 Total party membership 545,000 319,800 165.300 173,600 326,500 225.000 510.000 600.000 275,767 231,846 205,382 1,737,347 1,100,398 615,219 1,693,156 1,136,723 840,000 Membership as % of electorate 7.04 3.94 2.11 2.15 1.15 3.19 6.33 6.77 7.30 5.88 5.14 5.05 2.98 1.57 4.12 2.63 1.92

Denmark

France

United Kingdom

Source: Excerpts from Mair & van Biezen (2001, 15-16). Media coverage is perhaps the most visible non-financial political resource, and its importance for party competition is even greater in-between the elections. During the electoral process and electoral campaign, there is much background noise in the media coverage of politics, because of the numerous political parties competing, especially in multiparty systems with PR, where even smaller, insignificant parties are advertised. But during the period between the electoral periods, the background noise decreases, and only significant parties (that hold public offices) are normally covered by media.
67

116

Is there any evidence that political parties in the CEE are more dependent of resources coming from outside the party organizations? One hint is the low membership rates in the CEE, suggesting that membership fees are insignificant sources of revenues for parties. If party members are few, parties need to seek additional revenues outside the party. The state financing is such source of revenues from outside party organization which is very important for political parties in the region. In the Chapter 7, I highlight the significance of state financing and media access for political parties in the region, arguing that they are important political barriers meant to protect the established parties against new competitors. But there is also another source of revenues, much more sensitive since it is often associated with political corruption: private donations, coming mainly from top politicians who obtain revenues from the state offices and from private corporations and enterprises. There is some evidence that most of the money for political activities comes from private donors, even more than from the state subsidies. However, the reported figures rarely reflect the reality because the process is not always transparent. Moreover, the institutions that control the private donations are often ineffective in monitoring and banning illegal donations. Political parties, on the other hand, have incentives to maintain the silence about the sums they collect from private donors because these are usually limited by law to certain upper thresholds. Thus, instead of showing brute numbers of whose reliability is questionable, I rest my case about arguing that parties in the CEE are dependent of resources outside of party organization to the following argument:

One of the most prominent issues in post-communist states has been the large donors political influence and access to decision makers. In a matter of years fundraising efforts among ordinary supporters have declined and institutional donations have become more significant. Individual contributions from personal income or wealth, as well as corporate donations, have become a major source of income for many CEE political parties and candidates, especially during election years. Even taking account

117

of almost certain under-reporting of political payments, the growing role of large donors emerges clearly (Austin and Tjernstrom 2003).

Even though the data about party financing in the region is scarce and often unreliable, there is important evidence that political parties are constrained to rely on state financing and private donations. As Austin and Tjernstrom (2003, 90) suggest, the later seems to be the most important, although it consists of irregular flows of money for established political parties. However, what is more important than the strict hierarchy among them is that these external sources of money are only available as long as the parties maintain their political viability by maintaining in the Parliament and influencing the political process.
Table 5.2.: Electoral volatility CEE, 1990-199668
Country Czech Republic Hungary Poland Slovakia Slovenia Estonia Latvia Lithuania Bulgaria Year 1990-92 1992-96 1990-94 1991-93 1993-97 1990-92 1992-94 1990-92 1992-96 1992-95 1993-95 1992-96 1990-91 1991-94 1994-97 1990-92 1992-96 Electoral volatility 9.5 19 24 28 29.5 13 7 18 25 26.5 32.5 28 13.5 15.5 24.5 4.5 11

Romania

Source: Rose et al (1998, 119), taken from Lewis (2000, 85).

I argue that these peculiarities of party politics in CEE have important consequences for electoral party competition. Under the influence of uncertainty about party-voter linkages, low membership, lack of internal political resources, the political parties in the CEE have reasons to be even more vote-oriented in elections than political parties in the
68

There are certain differences in various reports of electoral volatility due to different strategic decisions taken by authors when calculating Pedersen index. One example is the case of party splits, in which case some authors consider the splitting parties as new parties, while others continue to judge them as old parties. Such differences between authors are accountable for different figures of the volatility in CEE, but overall, the volatility is higher than in the Western Europe.

118

western Europe because their survival is at stake. The political environment in which the political parties compete is significantly less predictable than the western party systems. The electoral volatility, which is largely accepted as a measure of stability of party systems, is considerably higher in the CEE and in net terms and it raises serious questions about the stability of party-electorate relationships (see Table 5.2 above).69 Such high volatility shows that parties do not enjoy the benefit of having stable electorates and they have to cope with the possibility of losing rapidly the electoral support, which ultimately endangers the organizational survival of political parties.

5.3. The Strategy of Research: The Empirical Hypotheses

Linking rational choice models with empirical phenomena is by no means easy. The parsimonious and elegant structure of the theoretic models often clashes with the complexity of the real politics. As models are based on very strong assumptions about political behaviour of political actors, operationalization of the important variables is rather difficult task. However, the value of looking to real political phenomena through the lentils of a rigorous developed theory rather than adopting ad-hoc explanations overcomes these difficulties (Riker 1990). In order to analyse the dynamics of the CEE electoral competition using the theoretical predictions of the rational choice models, I advance three main empirical hypotheses. In the Chapters 3 and 4, I present some rational choice models whose conclusions argue that political parties do not compete for the median voter because they risk competition at extreme positions. Because parties face such risk, they take on ideological positions which are noticeably non-centrist and non-convergent. However, I
Electoral volatility is an indicator that measures the stability of party system (or of party system change), which takes into account the electoral strength of the parties in two subsequent elections. Pedersen (1979, 6) defines the electoral volatility as [] the net change within the electoral party systems resulting from individual vote transfers. Since the electoral volatility is calculated for two subsequent elections and represents the net change within the electoral party system, the formula of electoral volatility (or Pedersen index) is:
n
69

V=(

p
i =1

i ,t

) / 2 , where pi ,t represents the net gain/loss of one party i in two consecutive elections.

119

implicitly argue in the Chapter 4 that such strategy is very risky, particularly when political parties have imperfect information. Moreover, these models neglect the fact that political parties may employ other means to block new entries: political institutions, such as electoral thresholds, electoral laws, and preferential public resources. The first hypothesis regards this theoretical insight: Proposition 1: The political parties in the CEE build up political institutions which aim at blocking the entry of new political parties in competition and at reducing the level of political uncertainty. The main hypothesis regards the interaction between political environment, policy preferences, political institutions and the maximization strategies of political parties in electoral competition. I shall argue that: Proposition 2: Due to peculiarities of CEE party systems (predominance of leadership, reduced membership, location of relevant resources outside party organizations etc), the political parties in the region are very much constrained to become viable electoral vehicles. Also, given that the established political parties define political institutions that block new entries in competition, I expect to observe a convergent pattern of party competition. Thus, if all above is correct, we should observe over time a high degree of policy convergence of the parties around the point in policy space that maximizes the chances of winning the election (the median voter).70 The structure of party system has important consequences for party strategies. Many parties face difficulty in advocating particular policy positions because their appeals are challenged by other political parties that offer similar political platforms. Such type of competition on the same side of ideological space determines some parties to receive a smaller number of votes in electoral competition, and they need to accept a secondary role in the party system. But, if some policy spaces are available, vote-maximizing parties will shift to these unoccupied ideological space. Thus, the third hypothesis is: Proposition 3: The structure of opportunities constrains the electoral strategies of political parties in the CEE (Mair 1997). If a certain viable policy space is available while

70

Since analytically I treat competition as taking place on a single dimension for party competition at a time, the convergence pattern should occur around the position of median voter. But since the median and the mean voter are very close each other in CEE elections, convergence to one point means convergence to the other as well.

120

several parties compete for other viable part of the policy space, I expect to see one or more of the parties shifting to these available positions. I have to emphasize that the reader should considered temporally the issue of convergence or divergence of political parties around the point of median voter. I do not expect political convergence occurring at every election time. After all, parties need to consider other issues as well, such as political reputation or persuasion of activists and compliance with the policy preferences of ordinary members. In some cases, political parties confronted with inevitable electoral loss might decide to strengthen their ideological orthodoxy rather than following the maximizing electoral strategy. Nor the parties compete in a world of full information about the preferences of voters and their preference distribution. Such imperfect information makes impossible for political parties to adapt perfectly their electoral strategies to their political goals; so that it is plausible that the point of median voter is never reached in empirical situations. However, since parties know that such position is rewarding electorally, we should observe empirically over time a high degree of party convergence around the point of median voter. Also, I would like to contend that maximization strategies do not bring better electoral result, irrespective to political context. Rather, I argue through these hypotheses that centripetal competition is one necessary element but by no means a sufficient condition of political success. Other elements such as credibility of party platforms, efficiency of political parties, the credibility and effectiveness of political leaders etc., play an important role in determining the electoral result.

5.4. The Data Sources and Data Analysis

5.4.1. Comparing Different Sources of Data for Political Party Competition In the behavioural analyses of political parties and party competition, researchers have extensively used four main types of data to locate the policy positions of political parties. Two of the data sources are indirect, while the other two are direct measures of party

121

positions. In this section I shall start discussing their merits and weaknesses, in order to justify the empirical results I present in the Chapter 7. The indirect sources of data about political parties are the mass and expert surveys. In the mass surveys, respondents are usually asked to place political parties on different policy scales, the most standard being left-right and liberal-conservative, the last particularly used in the American context.71 The underlining assumption is that the perceptions of voters reflect to some extent the real party positions on the issue dimensions. One advantage of such strategy is the high number of entries that allows the use of advanced statistical techniques. Also, this kind of data about party positions is often useful because it tells us something about the perception of voters regarding the political parties appeals in election. Moreover, due to that the number of respondents is high, the degree of confidence in the final results is considered to be also high. However, mass surveys have several big pitfalls for the study of political parties. One is that the substantial knowledge of the public about political parties and their policy positions is usually low. People are generally less interested in politics, and even those very interested tend to be quite ignorant about what parties advocate in elections. There are two other inter-related shortcomings. The first, which is surprisingly rarely acknowledged, is that data from mass surveys offer information about the perception of voters rather than information about party positions. The second drawback, which has been for quite a long time in the attention of scholars of mass politics, is the problem of endogeneity. Voters may position political parties the way they do because they like or dislike the respective political parties. This problem has been fervently debated in the context of the proximity vs. directional model of voting, with the supporters of the later claiming that relying on voters individual evaluation of party position is faulty because of the rationalization problem: voters bring parties they like closer to their own policy positions and ignore the real positions of parties.72 Using results obtained from the aggregation of individual evaluations is a technical solution that deals with the issue of rationalization, but it offers results that may be questionable in what regards their reliability.

71

Data of this kind is the Comparative Studies of Electoral Systems (CSES), available at http://www.cses.org. See Chapter 3, Section 3.5.5 for more discussion of this issue. In other contexts but with the same meaning, the problem is also known as projection.
72

122

The second indirect type of data is the expert survey. Expert surveys follow an opposite logic than mass surveys. These surveys do not rely on high number of respondents, as mass surveys do, but on their in-depth knowledge about party system and individual political parties. The experts are usually individuals who work intensively on the respective field of study, and thus, they are supposed to have substantial knowledge about the positions and policy dynamics of political parties. Moreover, experts have a broader vision about the party behaviour: in their estimations, they rely not only on the official statements of political parties, but also on discourses of political leaders, on how party votes in the legislative, on coalition behaviours and so forth. Other advantages of expert surveys are the cost-efficiency ratio and their accessibility in comparative research. This is not a negligible factor given that not always other sources of data are accessible at the time they are needed. Such advantages are noteworthy and they provide a good reason why the number of expert surveys expanded very much in the last decades (see Castles and Mair 1984, Laver and Hunt 1992, Huber and Inglehart 1995, Marks and Steenbengen 2004). However, both mass and expert surveys suffer of a serious methodological shortcoming. What they both measure is not the political phenomenon per se (in our case the policy positions of political parties), but the perception of individuals. Usually, the expectation of researchers is that such surveys tell something about the placements of political parties, even though the potential for measurement errors is relatively high. Such methodological illness is much more serious in the case of mass surveys, because it is often impossible to control for potential important variables that might affect the individual assessments. In order to increase the reliability of data, the surveys usually integrate several controlling variables (such as the question how much the expert or the voter likes/dislikes particular political parties), under the assumptions that regular inter-correlations between such controlling variables and evaluation of party positions would signal potentially error judgments or biases. These indirect sources of data about party behaviour are supplemented in the literature by other two direct approaches: party elite surveys and manifestos research. Contrary to mass and expert surveys, elite surveys and manifestos are direct (or more authoritative) sources of data on party positions. In the case of elite surveys, the party leaders tell directly what the policy positions of their parties are. The data from this kind is

123

authoritative particularly because party elites represent the group which defines the party positions and the evaluation of party positions is not anymore filtered by perceptions of external individuals. However, errors impinge heavily on this type of strategy too because party leaders are not always (many would say they are not most of the time) reliable sources of data. They may try to dissimulate their position by taking more orthodox ideological stances or by exaggerating their support for particular policies that are high on the public agenda. Furthermore, the surveys of party elites are rare because they are time consuming and financially expensive, which makes such data not easily available. In fact, compared to other sources of data on party positions, elite surveys are extremely rare and they are seldom used in comparative politics. The electoral manifestos have the advantage that they are authoritative documents issued by parties themselves just before the time of elections, which signal the positions of political parties on presumably important issues for party competition. Since they are public documents, issued at least once at every four or five years, manifestos are generally easily accessible source of data. Obviously, they are not binding documents and perhaps not even important for electoral process, as only a limited part of voters are well informed when they go to polls. However, they remain valuable sources of information because manifestos are the only documents in which the political parties address in almost exhaustive manner the political issues that are important at the time of elections. However, since the procedure of decoding the electoral manifestos is extremely complex, the potential for errors is also high. The logic of this approach asks that the length and the style of the text should not affect final results. Thus, in the case of manifestos research the basic unit of analysis is not the phrase or the sentence, but the coding unit, which is a part of a phrase or of sentence that has a policy meaning of its own.73 Three critiques to manifestos as reflection of party preferences have been addressed in the literature. The first critique argues that electoral manifestos do not influence significantly the results of elections and hence their impact is negligible. Since parties have to prove to voters that they have a policy program in case of winning, they adopt manifestos that are not necessarily a reflection of their policy preferences. The second critique relates to the first and argues that party manifestos do not tell anything about the
73

I detail the decoding procedure below.

124

policy preference of political parties, because once in government, parties may implement substantially different policy programs. The third critique, as I wrote above, suggests that since manifestos are not binding documents for political parties, they are nothing more than cheap talks at the electoral times. However, recent research have shown that electoral manifestos are generally good predictors of policies implemented by political parties in government (Klingemann et al 1994). Moreover, by stressing particular policies in their public stances, opposition parties may have an influence on the policy process, forcing the governments to consider alternative policy programs. Electoral manifestos also may function as instruments of democratic accountability: political parties in government may be confronted with their promises made at the time of elections. Although imperfect instruments in this regard, at least they give voters a signal about the party preferences and prospective government policies. Any type of data has merits and drawbacks and usually the strengths of one approach are the deficiencies of others and vice-versa (Volkens 2007). The only available strategy then is to select the type of data that suits the best the research question. Since the main concern of this thesis is to see whether political parties compete centripetally or not in the CEE elections, the party manifestos provide the most useful toolkit from theoretical point of view. First of all, party manifestos offer us reliable information about the ideological evolution of political parties from one election to another, about which expert surveys or mass surveys offer less or no information at all. Second, manifestos research data is available for every election from 1990 to 2002, for all countries that form the universe of cases of this thesis: the Czech Republic, Hungary, Poland, and Romania. By contrast, there are not enough mass or expert surveys to cover the relevant aspects for all elections in these countries or there are significant differences in terms of format of questions, scales etc that are detrimental to the validity of results. For these reasons, in order to address the main research question I rely on data obtained from decoding the party manifestos which the political parties issued between 1990 and 2002.

125

5.4.2. Extracting Data from Electoral Manifestos Analyzing the electoral manifestos requires a particular technique of content analysis. The text of electoral manifestos is first divided in (parts of) phrases or (parts of) sentences that have substantive policy content by their own and which are named the coding units. For instance, the sentence We shall increase the military expenditure by 30% by the end of 2010 has a substantive policy implication: it signals that respective party is in favour of a strong military, perhaps even in favour of militarism. A phrase such as Our party will cut the budget deficit that has affected our national economy in the last five years points to a policy position that comes in the line of classical economic orthodoxy. Also, a sentence may have no policy meaning or one or more policy meanings. Contrary to other techniques of content analysis, where you allocate a particular word or group of words to a specific category, in the case of electoral manifestos these coding units (or quasi-sentences) are assigned to a particular policy domain and policy category included into a predetermined coding scheme. The coding scheme contains seven substantial policy domains (External Relations, Freedom and Democracy, Political System, Economy, Welfare and Quality of Life, and Fabric of Society) divided in fifty-six policy categories detailed in the Appendix 3. By rule, every coding unit is assigned to only one policy domain and to only one policy category from the coding scheme. A special category of uncoded sentences, containing quasi-sentences without policy content, is set to include statements like Our party is the only that can bring economic prosperity. Also, in order to make inter-party (and even inter-country) comparability possible, the numbers of quasi-sentences from each policy category are standardized, taking the total number of quasi-sentences as a base. The entries from every policy domain and policy category are transformed in percentages. Thus, all the numbers on party positions reported in this thesis are percentages unless the text specifies otherwise. Obviously, the sum of percentages allocated to each policy category and the percentages of non-policy quasisentence equals 100%. As the raw text is transformed in raw data, the next analytical step is to devise a meaningful policy dimension for party competition in the CEE. Fortunately, using data

126

from party manifestos to draw up a map of political parties has had a long history in empirical research of party positions in the Western Europe (see Budge, Robertson and Hearl 1987, Klingemann et al 1994). Borrowing from their experience, I use a similar toolkit for party competition in CEE, which I detail below. Many of the previous empirical research have reported that the competition in Western Europe usually takes place on one political dimension which has been generally labelled as the left-right dimension (see for instance Klingemann et al 1994). Traditionally, this left-right scale has been included positions of political parties concerning the creation and distribution of wealth within the society. The leftist positions on economic issues have been traditionally associated with the state ownership and high redistribution of wealth, whereas the rightist positions have been linked to free market appeals and non-intervention of state in the economy. However, the political conflict between parties does not confine only to the fabric of economy since parties conflict about other values that are not limited to economic sphere. As some authors point out, especially after capitalist development in 1970s, another dimension of conflict between political parties has been the clash between rightauthoritarian values (morality in public sphere, pro-military attitudes, nationalism etc) and left-libertarian ideas (decolonization, internationalism, ecologism and so forth). One author suggests that the Western European competition political space has shifted in the last decades from a pure left-right policy space, delimited by socialist vs. capitalist politics, to an orthogonal policy space defined by left-libertarian vs. right-authoritarian politics (Kitschelt 1994, 31).74 If such contention about the change in the nature of political competition in Western Europe is accepted, then a single dimension of party competition should include policy issues which are not restricted to one policy domain (be it economic or social), but to include attitudes that reflect both economic and cultural attitudinal values. Using electoral manifestos as sources of data, a yardstick has been usually devised to position the West European political parties on a general left-right superdimension. The Table 5.3 shows how the left-right positions of political parties in Western Europe have
The theoretical concern of H. Kitschelt was the change in political strategies of Social Democratic parties in the Western Europe resulting from the change in the western societies determined by the capitalist development and globalization process. Since capitalism changed the attitudes and values of citizens, confronted with a new series of opportunities and costs, the social-democratic parties should adapt their political strategies and policy stances to face these new challenges (Kitschelt 1994, 280).
74

127

been derived using the decoded texts. The procedure assumes that some of the categories of the coding scheme belong traditionally to the left politics, whereas other categories are usually emphasized by the rightist parties. The equation which shows the position of one party on left-right dimension using the data obtained from electoral manifestos is: LRi = (Categories of Left emphases) (Categories of Right emphases)75 where LRi means the position on left-right dimension of political party i, and left and right categories are those identified in the Table 5.3. The equation tells us that if the final result has a negative sign, the respective party emphases rightist policy appeals in its political program (the rightist categories weight more than the leftist categories). Respectively, a positive sign means that the party stresses much more the leftist categories in the party manifestos. Given that the entries in each category are percentages of total coding units committed to respective policy category, the result of the equation take values in the interval [-100;100], where [-100] means the most rightist position and [100] the most leftist position. However, importing the above equation to derive party positions in the CEE on a single left-right dimension would raise serious theoretical and methodological problems. Some of the categories are not at all salient in the CEE politics (such as decolonialization, for instance), and the meanings of left and right might have substantially different policy content from country to country. For instance, Kitschelt et al (1999) argue that due to the negotiated transition, the economic issues have become less salient for party competition in Hungary as both camps agreed in 1989 upon the major economic transformations necessary to tackle the macro-economic problems. By contrast, in environments where negotiations were not possible, such as the Czech Republic, the disagreement between political parties about the economic reforms made the economic issues to impact more on the political process.

I have kept the original formula, although mathematically it would make much more sense to have (right) (left) positions, so that + means right position and - means left position. In my interpretation of the dimensions that I detail below, I have reversed the components of the formula in order to match the standard mathematical interpretation (minus means leftist position and plus means a rightist position).

75

128

Table 5.3: Left-right equation with data obtained from electoral manifestos Composition of left-right scale Left Emphases: Sum of % for Decolonialization Anti-Military Peace Internationalism Democracy _______ Regulate Capitalism Economic Planning Pro-Protectionism Controlled Economy Nationalization Pro-Social Services Pro-Education Pro-Labor Source: Laver and Budge (1993) took from Klingemann et al (1994). Right Emphases: Sum of % for Pro-Military Freedom, Human Rights Constitutionalism Effective Authority Free Enterprise Economic Incentives Anti-Protectionalism Economic Orthodoxy Anti-Social Services National Way of Life Traditional Morality Law and Order Social Harmony

To address these concerns, instead of employing the traditional left-right formula shown in Table 5.3, I make use of two separate competing dimensions. Since the main transformation of former communist countries has taken place in the economic sphere, the first dimension I analyze in the following chapters regards the positions of political parties on the economic left-right dimension. Such unprecedented economic transformation generated conflicts that severely affected the East-European societies and shaped the political competition even in peculiar countries like Hungary. As I show in the Chapter 7, the percentages of policy statements dedicated to economy or welfare are by far the highest among all policy domains of the coding scheme. The fact that political parties stress so much the economic or redistributive issues in their electoral appeals represents an indicator of the salience of the economy and welfare for the electoral party competition. However, political competition is unlikely to develop exclusively on economic and redistributive issues. In the CEE, social transformations accompanied economic reforms. The closed communist societies of CEE opened after 1989 and many challenges to the old cultural values or prejudices emerged. Ethnic groups, for instance, started to voice and asked for a political system in which they could preserve their social, cultural or linguistic identity. Others, such as religious groups, were more concerned with their religious 129

identities and moral values which they considered to be at stakes under the effects of globalization process. Thus, a second dimension of party competition combining conflicts between cosmopolite values (internationalist values, pro-European integration, positive views on multiculturalism, etc) and a traditional way of life (nationalism, traditional moral, and so forth) is likely to shape the strategies of political parties in the CEE.

Table 5.4: Economic Left-Right dimension in the CEE Economic Left-Right dimension Economic position Economic positions Right Left 401: Free Enterprise: Positive 403: Market Regulation: Positive 402: Incentives: Positive 404: Economic Planning: Positive 407: Protectionism: Negative 405: Corporatism: Positive 414: Economic Orthodoxy: Positive 406: Protectionism: Positive 505: Welfare State Expansion: Negative 412: Controlled Economy: Positive 702: Labor Group: Negative 413: Nationalization: Positive 704: Middle-Class and Professional 415: Marxist Analysis: Positive Groups: Positive ____ 4011: Privatization: Positive 503: Social Justice: Positive 4012: Controlled of Economy: Negative 504: Welfare State Expansion: Positive 4013: Property-Restitution: Positive 701: Labor Groups: Positive 5041: Private-Public Mix in Welfare: 4122: Mixed Economy: Positive Positive 4123: Publicly-Owned Industry: Positive 4124: Socialist property: Positive 4131: Property Restitution: Negative 4132: Privatization: Negative

Table 5.5: Cultural Left-Right components in CEE Cultural Left-Right dimension Cultural position Cultural positions Right Left 110: European Integration: Negative 108: European Integration: Positive 601: National Way of Life: Positive 602: National Way of Life: Negative 603: Traditional Morality: Positive 604: Traditional Morality: Negative ____ 608: Multiculturalism: Negative 607: Multiculturalism: Positive 1023: East European Countries: Negative 1013: East European Countries: Positive 6013: National Security: Positive 6071: Cultural Autonomy 6081: Multiculturalism Pro-Roma: 6072: Multiculturalism Pro-Roma: Negative Positive 7052: Minorities Abroad: Positive 7051: Minorities Inland: Positive

The Tables 5.4 and 5.5 show how I have structured the economic and cultural dimensions to match the party competition in the CEE elections. Similarly to the classical approach, the party placements of economic left-right can be obtained subtracting the 130

percentages of economic-left categories from the economic-right categories.76 The result of party is position on the economic left-right dimension is given by the equation:
LReconi = (per401 + per402 + per407 + per414 + per505 + per702 + per704 + per4011 + per4012 + per4013 + per5041) (per403 + per404 + per405 + per406 + per412 + per413 + per415 + per503 + per504 + per701 + per4122 + per4123 + per4124 + per4131 + per4132).
77

(5.1)

Similarly, the positions of party i on the cultural left-right dimension are given by the equation:
LRculti = (per110 + per601 + per603 + per608 + per1023 + per6013 + per6081 + per7052) (per108 + per602 + per604 + per607 + per1013 + per6071 + per6072 + per7051)

(5.2)

5.4.3. Data on Voters Policy Positions The second type of data I use is about voters policy positions. Since my main theoretical prediction is that, given the constraints I emphasized earlier in this chapter, the political parties approach the MVP in electoral competition, I need data about voter distribution at the time of elections. In this regard, I rely mainly on two sets of individual-level data which contain information about voters ideological distribution. The first set is formed by the Central and Eastern Eurobarometer (CEEB) series. The CEEB surveys contain individual-level data, collected yearly between 1990 and 1996, which covers a wide range of political, economic and social issues for almost all countries from Central and Eastern Europe.78 The population are individuals ageing 15 and over, selected through multistage national probability samples or national stratified quota samples.
Notice than now the negative scores mean leftist policy positions, whereas the positive scores denote rightist policy positions. 77 Per is the part of the name of the variable that comes with the original data set. The numbers are actually the identification variables of the coding scheme. 78 Some of the countries are not in all of the studies of this series because of the political or interethnic violence, such as Yugoslavia in 1991, or some areas from Moldova, Armenia and Georgia in 1992.
76

131

The second source is the Comparative Studies of Electoral Systems (CSES). The CSES represents a cross national collaborative project on election studies that aims at bridging researchers interested in electoral research from over fifty countries. The project includes individual level data about vote choice, party evaluations and a series of questions about various economic and political attitudes. The common methodology and questionnaire make this source of data being very suitable for comparative research. The Table 5.6 summarises the information about the individual level data I make use of in the next chapters, by country and year of elections. The last column provides the sample size for each data source. The CEEB and CSES cover all elections in the Czech Republic, Hungary, Poland and Romania, between 1990 and 2002, with three major exceptions. The first two are 1998 elections in the Czech Republic and 2000 elections in Romania, about which no information is available. Thus, I am unable to provide information about the median voter positions in these particular elections. However, as we shall see in the Chapter 7, the position of median voter on economic dimension is generally stable from one election to another and, although without having direct empirical evidence, we can infer with some necessary precautions where the median voter is located in 1998 Czech elections and 2000 Romanian elections. Another case not covered by CEEB and CSES is the Hungarian 1994 elections. The CEEB 1994 series does not provide information about the left-right self-placements of Hungarian voters and the CSES dataset does not include any elections prior to 1996. However, I supplement this shortcoming with data from a survey done by the Central European University in April 1994, three weeks before general elections in Hungary, which was based on a national random route sample of population of 18 year old and over.79 For two other cases, the 1993 Polish elections and the 1990 Romanian elections, the CEEB series includes no information about voter distribution on left-right dimension. Romania was not part of the first wave of the CEEB in 1990. In the case of 1993 elections in Poland, the data contains only missing values for the question regarding the selfplacements of voters on left-right dimension. However, this information is available for Romania (1991) and Poland (1992), which I use as proxies.

79

I thank Professor Gabor Toka who kindly agreed to share this data with me.

132

Table 5.6: Data information, sources and samples

Country Year of election 1992 Data Source Total population surveyed 924

The Czech Republic

1996

1998 2002 Hungary 1990

1994

1998

2002 Poland 1991

1992^

1997

2001 Romania 1991*

1992

1996

2000

Reif, Karlheinz, and George Cunningham. CENTRAL AND EASTERN EURO-BAROMETER 3: POLITICAL DISINTEGRATION, October November 1992 [Computer file]. ICPSR06106-v2. The Comparative Study of Electoral Systems (www.cses.org). CSES MODULE 1 FULL RELEASE [dataset]. Ann Arbor, MI: University of Michigan, Center for Political Studies [producer and distributor]. August 4, 2003. N/A The Comparative Study of Electoral Systems (www.cses.org). CSES MODULE 2 FULL RELEASE [dataset]. June 27, 2007. Reif, Karlheinz, and George Cunningham. CENTRAL AND EASTERN EURO-BAROMETER 1: PUBLIC OPINION IN CENTRAL AND EASTERN EUROPE, 1990 [Computer file]. ICPSR06104-v4. CEU (Kzp-Eurpai Egyetem), 1994, THE DEVELOPMENT OF PARTY SYSTEMS AND ELECTORAL ALIGNMENTS IN EAST CENTRAL EUROPE: HUNGARY, Spring 1994. SPSS data file. Budapest: Department of Political Science, Central European University The Comparative Study of Electoral Systems (www.cses.org). CSES MODULE 1 FULL RELEASE [dataset]. Ann Arbor, MI: University of Michigan, Center for Political Studies [producer and distributor]. August 4, 2003. The Comparative Study of Electoral Systems (www.cses.org). CSES MODULE 2 FULL RELEASE [dataset]. June 27, 2007. Reif, Karlheinz, and George Cunningham. CENTRAL AND EASTERN EURO-BAROMETER 2: CURRENT AFFAIRS AND THE MEDIA, SEPTEMBER-OCTOBER 1991 [Computer file]. Conducted by Gallup International (Europe), London. ICPSR06105-v2. Reif, Karlheinz, and George Cunningham. CENTRAL AND EASTERN EURO-BAROMETER 3: POLITICAL DISINTEGRATION, October November 1992 [Computer file]. ICPSR06106-v2. The Comparative Study of Electoral Systems (www.cses.org). CSES MODULE 1 FULL RELEASE [dataset]. Ann Arbor, MI: University of Michigan, Center for Political Studies [producer and distributor]. August 4, 2003. The Comparative Study of Electoral Systems (www.cses.org). CSES MODULE 2 FULL RELEASE [dataset]. June 27, 2007. Reif, Karlheinz, and George Cunningham. CENTRAL AND EASTERN EURO-BAROMETER 2: CURRENT AFFAIRS AND THE MEDIA, SEPTEMBER-OCTOBER 1991 [Computer file]. Conducted by Gallup International (Europe), London. ICPSR06105-v2. Reif, Karlheinz, and George Cunningham. CENTRAL AND EASTERN EURO-BAROMETER 3: POLITICAL DISINTEGRATION, October November 1992 [Computer file]. ICPSR06106-v2. The Comparative Study of Electoral Systems (www.cses.org). CSES MODULE 1 FULL RELEASE [dataset]. Ann Arbor, MI: University of Michigan, Center for Political Studies [producer and distributor]. August 4, 2003. N/A

1,229

N/A 948 989

1,200

1,525

1,200 1,000

999

2,003

1,794 1,000

1,000

1,175

N/A

^ Data on left-right self-placements is unavailable for 1993 Polish elections. 1992 is used as a proxy for that election. * Romania was not included in the first wave of CEEB (1990). 1991 is used as a proxy for that election.

133

The main questions I use regard the self-placements of voters on left-right dimension. The CEEB asks respondents to place themselves on a 10-point left-right scale, where 1 means left and 10 means right. The CSES uses an 11-point left-right scale, where 0 means left and 10 means right. The CEEB typical question about this issue is: In political matters, people talk of the "left" and the "right". How would you place your views on this scale? For CSES Module 1 (1996-2001), the question asking for left-right self-placements is: In politics people sometimes talk of left and right. Where would you place yourself on a scale from 0 to 10 where 0 means the left and 10 means the right? The same question is included in the CSES Module 2 (2001-2006). All data of this kind included in the next chapters concerns only the valid answers (persons who could place themselves on this dimension), if not mentioned otherwise. The usefulness of the left-right dimension for analysing the CEE politics has not received the unanimity of the scholars. Two arguments have questioned the utility of the left-right scale for analyzing the CEE politics. The first is that some of the authors regarded with scepticism the possibility that the CEE political systems would be shaped by political divisions similar to those from the Western Europe.80 The second argument concerned the substantive content of such dimension: is it about economic issues, cultural issues or includes a mix of them? However, regarding the first concern, studies on the dimensionality of political space in the Eastern Europe have shown that economic liberalism and social, political and ethnic liberalism shape the political choice of voters in the region (Whitefield 2002, 18889). Moreover, since political parties in the CEE use in their language words like left and right, it is inevitable that such language shapes to some extent the political choice of voters. On the other hand, voters prove that the left-right dimension is not an unknown notion for them as they are able to place themselves on such continuum and the missing values are generally low. In regard to the specific content, the left-right dimension may vary from country to country, raising methodological problems when using it in cross national comparisons. However, in this specific context, this potential problem can not affect the results because: 1) I do not make inter-country but rather intra-country comparisons, and 2) I make use, as mentioned above, of two different scales, one
80

See the discussion earlier in this chapter.

134

concerning economic and redistributive issues, the other referring to authoritarian or nationalistic versus liberal or cosmopolitan issues, which eliminates the first concern with this type of scale. The position of parties on economic left-right dimension will be paired with the distribution of voter self-placements on the left-right dimension. However, there is one limitation of the research that could not be avoided: there is no consistent data (crossnational and longitudinal) about voter positions on the cultural left-right dimension, so that I am unable to provide information about the position of median voter on this scale. Thus, I show in the next chapters that political parties converge ideologically on the cultural dimension of competition too, but I can not infer anything substantially about the convergence at the position of median voter. In the next chapters, I present the main findings of the thesis, based on data I have described here: political parties in the CEE compete centripetally from ideological point of view, both on economic and cultural dimensions. I start by briefly introducing in the next chapter the party systems of the four countries from CEE: the Czech Republic, Hungary, Poland, and Romania. The Chapter 7 presents the substantive analysis of data where the hypotheses presented earlier in this chapter are tested.

135

CHAPTER 6: AN OUTLINE OF THE CZECH, HUNGARIAN, POLISH AND ROMANIAN PARTY SYSTEMS

6.1. Introduction
Before discussing the substantial findings of the thesis, it is worth introducing the four national party systems where I am going to test the hypotheses I formulated in the previous chapter. The aim of the chapter is to outline the main political parties that have competed in national party systems from 1990 to 2002 (2000 for Romania), their ideological orientation and electoral results in parliamentary elections. Since the chapter is exploratory in its nature, the reader with knowledge about these national party systems is advised to move to the next chapter. Political parties and party competition have been new phenomena in the CEE. One of peculiarities of political democratization in the region has been the absence of any substantial role played by political parties in the break of communist regimes in the CEE (Lewis 2000, 19). Although many leaders of anti-communist movements involved politically after the end of the communism, the political parties as democratic institutions emerged only after the communist regimes liberalized and accepted to organize free and competitive elections. By contrast, the political parties played a significant role in democratization process of the South European countries (e.g. Spain, Portugal or Greece) or even in the Latin America where they successfully mobilized people against the authoritarian regimes and significantly shaped the institutional environment in the early period of transition. For this reason, the political parties in the CEE have been labelled as transitional and transitory phenomena, since they were considered produces of political transition, they were linked to the problems of transition and they were expected to vanish in transition or to undergo substantive changes when the political systems stabilized (Elster et al 1998, 132).

136

Developing as new political phenomena and initially without significant ties with the social groups at the beginning of transition (a statement which is proved by the numerous political parties competing in early elections), the political parties and party systems of CEE proved to be quite unstable political institutions. In some respects, this instability reflected the peculiarities of party emergence in the CEE. Some political scientists argued that the political parties in the region did not incorporate any organizations that were able to steer the parliamentary party from outside and parties usually were organized before the time of elections as political vehicles for political competition (von Beyme 1996, 122-23). This instability has been the most noticeable at the level of organizational continuity, where it could have seen a whole series of splits, mergers or ideological repositioning. The political parties in the CEE have emerged from three main sources. One source has been the communist period from which the former communist parties emerged by readapting ideologically to the new political context. Together with former communists, other political parties which previously had been linked to the communist regimes transformed themselves into genuine political parties.81 A second source of party genesis has been the pre-communist period that produced the recurrence of some historical political parties, such as the Christian Democratic Peoples Party and the Agrarian Party in Czechoslovakia, the Independent Smallholders Party in Hungary, or the National Peasant Party and the National Liberal Party in Romania. Although early scholarly works suggested that this could be an important source of party emergence, the importance of the historical parties has been very limited in transition period, except Romania where the two parties played an important role in the party system. Finally, the third source of party development has been the transition itself. New political parties came out during this period trying to obtain the voters support by using conflicts generated by the social, political and economic processes produced by transition. This source has represented the most significant pool for party emergence and party system development in the CEE, although the importance of the other two categories varies from country to country.

81

Especially the former marionette parties which had existed before 1989 without any possibility of challenging the leading role of the Communist Party, such as the PSL and the Union of Labour-UP in Poland.

137

6.2. Parties and Party System in the Czech Republic

The Czech party system is notorious for being considered one of the most stable party systems in the region. This political stability steams out not only from steadiness of the individual political parties, but also from predictable patterns of party competition and moderate polarization of party system (Klima 1998, 509; Evans and Whitefield 1998, 135; Olson 1998, 461-63).82 The party system has been well organized around the economicredistributive left-right division which separated the left parties, the unreformed Communist Party - KSM and the Social Democrats SSD, from the groups of right parties organized around the Conservative ODS. As well as other party systems from the region, the Czech political parties came out from the same three sources I have referred to above (Linek and Mansfeldova 2006). The KSM was the only communist party from the former satellite communist countries which did not reform its ideological appeals. Inside the party, the hardliners won the battle against the reformists in 1990 and the party remained confined to the Marxist-Leninist ideology. As a consequence, its electoral success has been generally low, but in the 2002 elections the party obtained almost twenty percent of the popular vote (see Table 6.1). The other party of the left, the social-democratic SSD with roots in the pre-communist party system of Czechoslovakia, has been the main party of the left. The party capitalized the public dissent with the ODS strict economic reforms and succeeded to displace the conservative ODS from power after 1998 elections, when it formed the government after signing a power sharing agreement with the ODS. The party maintained the main leading role in the government after 2002 elections, but it won fewer parliamentary seats than it had had in the previous legislature and it needed the support of KDU- SL and the Union of Freedom Democratic Union (US-DEU) to maintain in government. At the political centre there has been another historical party, the Christian Democratic KDU-SL, also with roots in 19th century, which continued to survive even during the communist regime, though its role was restricted to a marionette. It has always been rather small political party in transition, attracting limited electoral support, but which

82

Polarization is seen as the ideological distance between the parties of extremes.

138

has effectively played the role of pivotal party that has allowed it to join the government both in alliance with the ODS (in 1990s) and the SSD (after 2002). On the political right, we find political parties with significant social roots in the transition period. The most noticeable has been the Conservative ODS, which capitalized electorally the split of the Civic Forum in 1991. As Table 6.1 shows, the ODS has been the most successful party of the right, combining free market policy appeals with moderate scepticism about the European Union. The party earned much of the credits for the success of the Czech economic transition since this has been largely due to strict economic policies imposed by the ODS governments, under the premiership of Vclav Klaus. However the partys electoral fortune has been hit by economic problems, internal political instability and allegations of corruption, which made the party electoral score marginally decline between 1993 and 2002.
Table 6.1: The electoral party strength, the Czech Republic 1990 1992 1996 1998 2002 votes seats votes seats votes seats votes seats votes Seats % % % % % OF 49.5 127 KSM1 13.24 32 14.05 35 10.33 22 11.03 24 18.51 41 HSD-SMS 10.03 22 5.87 14 KDU-SL 2 8.42 19 6.28 15 8.08 18 9.00 20 21 SSD 4.11 0 6.53 16 26.44 61 32.31 74 30.21 70 SPR-RS/RMS3 1.00 0 5.98 14 8.01 18 3.90 0.97 0 ODS4 29.73 76 29.62 68 27.74 63 24.48 58 LSU 6.52 16 ODA 5.93 14 6.36 13 US-DEU2 8.60 19 10 Koalice2 14.28 31 others 13.70 0 19.11 0 11.16 0 7.42 0 11.04 0 Source: Linek and Mansfeldov (2006) Note: A hyphen in the table means that the party did not run in the elections or gained less than 1 per cent. 1) In 1992 KSM ran together with Democratic Left of the Czech-Slovak Federative Republic (Demokratick levice SFR ) in the coalition Left Block (Lev blok). 2) In 2002 KDU-SL and US-DEU ran together as the Coalition (Koalice). 3) In 1990 SPR-RS ran in coalition with the All Peoples Democratic Party (Velidov demokratick strana). In 2000 RMS succeeded SPR-RS. 4) In 1992 ODS ran in coalition with the Christian Democratic Party (KDS). Year of elections

Other smaller political parties have been the Liberal parties, which proved to be very unsuccessful electorally like in almost all other party systems from the region (von Beyme: 1996, 125). At the beginning of 1990s, the Civic Democratic Alliance (ODA) was 139

the main Liberal party, but its electoral score was extremely low and in the aftermath of 1996 election the party eventually failed to enter the parliament. The other Liberal party, Freedom/Democratic Union (US-DEU), took advantage of this opportunity and entered for the first time in the Parliament in 1998. The party maintained the parliamentary position after 2002 elections when it formed an electoral alliance with KDU- SL, under the label of Koalice (Coalition), but its electoral support is low and it is improbable to play a significant electoral role for a long period of time. Contrary to the former Communists which managed to increase steadily the electoral support, there is no significant political party on the extreme right in the Czech party system. Since the Czech Republic is ethnically homogeneous, there were no parties to compete on nationalistic issues. The EU scepticism, another typical issue for the far rightwing parties in Western Europe, has been already incorporated into the electoral appeals of ODS and KSM. Moreover, since the former communists usually appeal populistically against economic reforms, there has been little room for the development of significant extreme right-wing parties in the transition period.

6.3. Parties and Party System in Hungary

The Hungarian party system generally follows the same pattern as the Czech party system in terms of stability and consolidation of political parties (Evans and Whitefield: 1995, 1177). Remarkably at the Hungarian party system is that the main political parties, which had participated in the Round Table in 1989 when the opposition groups negotiated the political transition with the communist party, were still in the Parliament in 2006. The Hungarian party system also exhibits stable patterns of party competition, particularly in what concerns the government formation. The Socialist MSZP, the former communist party, usually has formed the government with their close allies, the left-liberals of SZDSZ, even though these two parties were at odds at the time of the Roundtable talks. On the other hand, the conservative parties, FIDESZ and MDF, are close allies and sometimes even run on the same electoral lists, as it happened in the 2002 elections.

140

The Hungarian party system reflects largely the political groupings that confronted in the 1989 Roundtable talks. Contrary to other countries from the region, the democratization process was negotiated between the Communist Party and the groups from the civil society which transformed later in political parties and run in the first free election of 1990. The Hungarian Democratic Forum (MDF) was established one year before the Roundtable and it led the anti-communist opposition in 1989. The party obtained a very good electoral result in the first free elections and the party leader, Jzsef Antal, became the prime minister forming the first non-communist government in coalition with FIDESZ and SZDSZ. However, the party lost the electoral supremacy because of the economic hardship that hit the Hungarian society during Jzsef Antal and Pter Boross (Antal successor as prime minister) governments, and it ended as the junior coalition partner of FIDESZ. Moreover, the party has been undergoing a substantial internal conflict, with many of the members, especially the MPs, favouring political unification with the other conservative party, FIDESZ. FIDESZ was founded in 1988, initially bringing together young students and intellectuals who shared the radical views about democracy and opposed the communist regime by advocating a radical political change. In many respects, but mainly for its political and ideological trajectory, FIDESZ has been one of the most interesting political parties in transition. After the 1994 elections, when the party obtained only 5% of the parliamentary seats, it underwent a radical ideological change, adopting more conservative positions on social and national issues. In this respect, the ideological shift of the party correlated with the decreasing electoral importance of MDF, so that after 1998 elections FIDESZ adopted conservative stances, developed closer links with the Church and addressed to a new, more conservative electorate (Enyedi 2005). At the other side of the political spectrum, the former communist party, MSZP, succeeded to reshape ideologically very soon after the shocking electoral result in the 1990 parliamentary election, when it had received only 8.55% of the parliamentary seats (see the Table 6.2). The party abandoned the Marxism-Leninism and developed as a typical west European Socialist party. Since the communist regime had allowed the introduction of economic liberalization well before 1989, the party accepted relatively easy the market economy and privatization of state economy in its political platforms after the end of

141

communist regimes. Capitalizing on economic grievances determined by economic reforms, MSZP returned to power after 1994 elections and continued the structural reform program, but it lost the 1998 elections to a coalition formed by FIDESZ and MDF. However, the party proved to be a very successful electoral vehicle and it returned to power in 2002 and succeeded to maintain it in the last elections (2006).
Table 6.2: The election results, Hungary 1990 1994 1998 2002 votes seats votes seats votes seats votes seats %a %a %a %a MSZP 10.89 8.55 32.99 54.15 32.92 34.72 42.05 46.11 SZDSZ 21.39 23.83 19.74 17.88 7.57 6.22 5.57 4.92 MDF 24.73 42.49 11.74 9.84 12.95b FKFPP/FKgP 11.73 11.4 8.82 6.74 13.15 12.44 FIDESZ 8,95 5.44 7.02 5.18 29.48 29.27 41.07 48.70c KDNP 6.46 5.44 7.03 5.7 2.31 0 Source: The Project on Political Transformation and the Electoral Process in Post-Communist Europe, University of Essex, http://www.essex.ac.uk/elections. a. The percentages of votes are only the percentages of the list votes. The single-member district votes were not included into these figures. b. Percentage of FIDESZ-MDF joint candidates. c. Percentage of FIDESZ-MDF joint candidates. Year of elections

Though taking opposite sides during the Roundtable talks, MSZP has constantly allied with SZDSZ ever since 1994. SZDSZ started its political history as the party of intellectuals in the Roundtable of 1989, when it was perhaps the fiercest opponent of the communist party. The party did very well in 1990 and in 1994 elections. After the Socialists transformed into a traditional European party and with the saliency of economic conflict decreasing, SZDSZ became politically close to MSZP. The alliance between these parties, agreed after 1994 elections, was facilitated by the fact that both parties began to attach more importance on cultural issues on which they took similarly left stances. Moreover, the alliance between MSZP and SZDSZ became feasible after FIDESZ turned after 1994 to more conservative positions with regard to social and political issues. Probably SZDSZ lost very much of the initial popular support because of this alliance, but it managed to play a significant political role in establishing the configuration of the Hungarian coalition governments.83
83

Grzymala-Busse (2001), for instance, argues that SZDSZ was punished electorally because it crossed the dividing line between the former communist parties and their political opponents. Grzymala-Busse suggests

142

6.4. Parties and Party System in Poland

Contrary to the Czech Republic and Hungary, Poland has had a highly fragmented and unstable party system, especially in what concerns the stability of party labels. If the parties of the left proved to be quite stable political entities, the parties of the right have constantly merged, split, and coalesced in politically ephemera party coalitions. However, even in this very unstable format, there has been a steady logic of how political competition has developed in Poland. The transition period has been marked by an intense ideological struggle between the former Communists and the heirs of the Solidarity, which had the social roots in the workers strikes at the end of 1970s and beginning of 1980s. In the first semi-free elections of June 1989, agreed after months of debates between the communist regime and the Solidarity opposition, the later obtained a stunning 96 out of 100 seats in the Senate, ending the rule of communists and signalling the fall of the communism in the Eastern Europe. The Communist Party, left with no outside support coming from USSR, accepted the defeat and passed the power to a new democratic government. The Communist Polish United Workers Party (PZPR) organized its last congress in January 1990, when the party transformed into the Social Democracy of Republic of Poland (SdRP). The new party did not run alone in the election of 1991, but it chose to form an alliance of the left with other social democratic and socialist parties under the label of Democratic Left Alliance (SLD). SLD has been the main political vehicle for the leftist parties from 1991 to 1999, when SLD transformed into a political party. The new party won the elections of 2001 in a coalition with the Labour Union (UP) and formed a government together with UP and the Polish Peasants Party (PSL), its traditional political allies. But the electoral strength of SLD decreased dramatically after some of the partys

there is an unbalance between the former communist parties and the parties that opposed them regarding the sensitivity to configuration of political coalitions. For Grzymala-Busse, the supporters of communist parties are allegedly less biased against parties that formed the opposition in 1989 than there are supporters of opposition towards the former communist parties. Thus, when political parties in the Eastern Europe cross the dividing line between former communist and non communist parties in the coalition formation process, parties of the right are severely punished by their political constituencies. Since SZDSZ has been the only party that crossed such ideological and political line in Hungary, it has been the one that severely lost votes in subsequent elections after 1994.

143

leaders had been accused of corruption which made the party lose eventually all the seats in the upper chamber of the parliament in 2005 elections (see Table 6.3).
Table 6.3: Electoral outcomes, Poland (Sejm only) 1991 1993 1997 2001 votes seats votes seats votes seats votes seats %a %a %a %a SLD 11.99 13.04 20.41 37.17 27.13 35.65 41.04 46.96a UD/UW 12.32 13.47 10.59 16.08 13.37 13.04 3.1 0 KPN 7.5 10 5.77 4.78 PSL 8.67 10.43 15.4 28.69 7.31 5.86 8.98 9.13 MN 1.18 1.52 0.44 0.65 0.39 0.43 0.36 0.43 UP 7.28 8.91 4.74 0 AWS 33.83 43.69 5.6 0 PO 12.68 14.13 LPR 7.87 8.26 SRP 0.08 0 10.2 11.52 PiS 9.5 9.57 Source: The Project on Political Transformation and the Electoral Process in Post-Communist Europe, University of Essex, http://www.essex.ac.uk/elections. a. In coalition with UP. Year of elections

The centre and the right, however, have been extremely fragmented political camps. The Liberal group included initially the Democratic Union (UD), a party with roots in the Solidarity, which was established in 1990 around the figure of Tadeusz Mazowiecki, the first democratic Prime Minister of Poland. Although the party obtained the best electoral score in 1991 elections, it got only 13.5 percent of the parliamentary seats, roughly the same score as the main competitor, SLD. In 1994, UD merged with another Liberal party (also with roots in the Solidarity), the Liberal Democratic Congress (KLD), and formed the Freedom Union (UW), but the new party failed to play a major role in the Polish party system. The Electoral Action Solidarity (AWS) was another political organization of the centre-right, organized in 1996 as a coalition of liberal, conservative or Christiandemocratic parties to compete effectively against SLD. AWS formed the government after 1997 election, but internal conflicts undermined the coalition and determined many parties to abandon it in favour of other political formations. AWS lost eventually all the electoral support and it did not enter into the Sejm after 2001 elections. By 2001, two other political parties emerged as the main centre-right alternatives. The new Law and Justice party (PiS) obtained a relatively low result in 2001, but it

144

succeeded to win the presidential and parliamentary elections in 2005, using a strange combination of socially conservative, pro-clerical, and rather centre-left economic appeals. The second party has been the Civic Platform (PO) which defines itself as a liberal party, advocating further liberalization of market economy. In its platforms, the party also aims at reforming the social and health systems and advocates for a fiscal reform through the introduction of a fifteen percent flat tax. The party came on top during the elections for the European Parliament, but only on the second position in the parliamentary elections of 2005 when it obtained 133 out of 460 seats in Sejm (with PiS on the top receiving 155). The Polish party system includes two main populist parties that emerged after 2001 election. The Self-Defence of the Republic of Poland (SRP) proposes a political program that combines elements of left populism with notable right extremist appeals. SRP proposes a state-funded industry and asks for larger government spendings on social programs. The party is also very unsympathetic towards accession to EU and fiercely opposes to foreign investments in Poland. Between 2001 and 2005, the SRP backed the social-democratic government in power, a political support which was very much needed by the minority coalition of SLD-UP-PSL. The second populist party is the League of Polish Families (LPR), which emphasizes the mainstream Catholic values, social conservatism and leftist economic policies. The party, created only in 2001, entered in the parliament in the same year obtaining 8% of the votes in the Sejm and maintained its result in 2005 elections. The party has been the only one who overtly opposed to accession of Poland to EU in 2005.

6.5. Parties and Party System in Romania

In terms of party families, there have been five major groups in the Romanian party system. The Social Democrats formed the largest political family, which included the Party of Social Democracy (PSD, former PDSR the Party of Social Democracy in Romania from 1992 to 2000), the Democratic Party (PD, since 1992), and the historical, smaller Romanian Social Democratic Party (PSDR). The first two parties emerged in 1991-92 when the National Salvation Front, the winner of the first election in 1990 with a comfortable 66 per

145

cent of the vote, split between reformists (PD) and conservatives (PDSR).84 Although both parties assumed social-democratic orientation, they have been in fierce opposition ever since. From 2005, PD gave up the social democratic orientation and struggled to be accepted into the European Peoples Party (EPP), though its ideological orientation remains an enigma to many of the party members and voters. To certain extent, the enthusiasm of the European Socialists to support PSD and the reluctant acceptance of the PD in the Socialist group determined this ideological reorientation of the party. Nevertheless, the absence of a major Romanian party affiliated to the EPP and the alleged support of this European party family played a significant role in determining the PD leadership to move formally to the right of political spectrum. Another major change in this party family occurred after 2000 elections when the PDSR and the PSDR merged into a unified Social Democratic Party (PSD) that governed between 2000 and 2004. This had no electoral purposes since the historical PSDR had at that time an insignificant share of votes in the opinion polls. However, it enabled PDSR to join the Socialist International after almost eight years of unsuccessful attempts. In the 2004 elections, PSD ran in coalition with the Humanist Party of Romania (PUR, after 2005 renamed as the Conservative Party), a small party with a fluctuating ideological position.85 Although the coalition obtained the largest number of parliamentary seats, it failed to form the government because of the veto powers of the new Romanian president, Traian Bsescu. The new president refused to appoint the leader of PSD as prime minister, but instead asked the D.A.86 leader, the Liberal Clin Popescu Triceanu, to form the new government. Threatened with new parliamentary elections, PUR and the Hungarian Alliance (UDMR) agreed to switch coalitions and joined D.A. in the new government that replaced the PSD in power in 2004.

The reformist group organized around the figure of the former prime minister and the official leader of National Salvation Front (FSN), Petre Roman, who was forced to step down with his government after miners protested violently in September 1991 against the reform of mining sector. The conservative group, led by former top communist party members, organized around the figure of president Iliescu, who declared himself against fast economic transition towards market economy. 85 The party started initially to point out that it is driven by a humanistic orientation, and declared itself as a centre party. After 2000 elections, the party entered in Parliament in coalition with PSD and almost immediately declared itself a social-liberal party that promotes middle class values and favours the economic development of this social class. After 2004 elections, the party changed its name to Conservative Party and started to emphasize values like tradition, family, and a sort of economic nationalism. 86 A coalition formed in 2004 by the Liberals and PD.

84

146

The second major party family has been the Christian Democrats (PNTCD), an important historical party that led the opposition parties to the first democratic change of government in 1996, after obtaining a poor electoral result in the founding 1990 elections. Just before the general elections in 1992, the party formed a heterogeneous coalition with the Liberals (PNL) and various other smaller parties and civic platforms under the label of the Democratic Convention of Romania (CDR). The Convention, together with PD, PSDR, and UDMR, formed the coalition government that ruled between 1996 and 2000. But the poor government performance of this coalition severely affected PNTCD, the largest party of the coalition, which lost the parliamentary representation after 2000 elections. The Liberals represent the third major family, which has been the most fragmented in Romanian politics, with no less than four liberal political parties between 1991 and 1996. The most important causes for this fragmentation were the internal disagreement about the junior status of PNL inside the CDR and the hopes of some leaders that an independent liberal party would be more appealing to voters. This disagreement made the party break into four smaller Liberal parties. Three of them joined the Democratic Convention, while the fourth did not enter in the Parliament after running alone in 1992 elections. After the 1996 elections, these four Liberal parties merged again under the label of the National Liberal Party (PNL), a name with historical resonance in the Romanian politics. Like the Christian Democrats, the poor performance of 1996-2000 coalition also hit PNL, so that in 2000 elections the party obtained a very low electoral score. Unlike PNTCD, it succeeded to enter the parliament on its own account though and the party returned to power after the 2004 elections when it established DA alliance with PD. The fourth group consists of nationalist parties, the Romanian National Unity Party (PUNR, from 1990 to 2000) and the Greater Romania Party (PRM, since 1992). Both parties proved quite successful in electoral terms, and they were able to join the coalition government led by PDSR between 1992 and 1996. Both parties remained reliable partners of PDSR during 1996-2000, but in 2000 elections the electoral growth of PRM brought it in opposition with the Social Democrats, while the PUNR failed to pass the electoral threshold. Finally, the fifth major group has been formed of representatives of minorities, and the dominant political party in this group has been the Democratic Alliance of Hungarians

147

in Romania (UDMR). In addition to the UDMR, another 12 to 18 minorities (depending on the election) received one parliamentary seat in the lower house of parliament (Chamber of Deputies) in the transition period.
Table 6.4: The electoral results, Romania (Chamber of Deputies only) 1990 1992 1996 2000 2004 Seats % Seats % Seats % Seats % Seats % PSM 13 3.8 FSN 266 66.9 PD 43 12.6 43 12.5 31 8.9 48 14.46 PDSR/PSD 117 34.3 91 26.5 138 40 113 34.0 PSDR 2 0.5 10 2.9 10 2.9 11 3.1 PER 8 2.0 4 1.1 5 1.4 0 0 FER 0 0 1 0.3 0 0 MER 12 3.0 0 0 0 0 PDAR 9 2.2 0 0 0 0 - ^^ - ^^ PUR/PC 0 0 6 1.7 19 5.7 PNL 29 7.3 0 0 25 7.2 30 8.7 64 19.3 PNLCD 3 0.8 6 1.7 PNLAT 11 3.2 0 0 PAC 13 3.8 0 0 PAR 3 0.8 0 0 PNTCD 12 3.0 41 12.0 82 23.9 0 0 PUNR 9^ 2.2 30 8.8 18 5.2 0 ^^ 0 ^^ PRM 16 4.7 19 5.5 84 24.3 48 14.5 UDMR 29 7.3 27 7.9 25 7.2 27 7.8 10 7.0 Minorities 12 2.7 13 3.8 15 4.3 18 5.2 18 5.4 Other parties 12 3.0 0 0 0 0 0 0 Independents 0 1 0.8 0 0 0 0 0 Source: Updated from tefan and Grecu (2004). As member of the Democratic Convention of Romania. Before the parliamentary elections of 1996, The Democratic Party and the Romanian Social Democrat Party formed an electoral alliance called the Social Democratic Union. In 2000, the Party of Social Democracy of Romania, the Romanian Social Democrat Party and the Humanist Party of Romania formed the electoral alliance called the Social Democrat Pole of Romania. As member of the Democratic Convention of Romania - 2000. ~ Parties reunited in the National Centrist Union. ^ In alliance with the Republican Party under the name the Alliance for the Unity of Romanians. ^^ In September 2000, Romanian National Unity Party merged with the Romanian National Party (that incorporated in 1998 the Agrarian Democratic Party of Romania) under the name: the National Alliance. Parties

The Romanian political parties have been weak political agents as anywhere else in the region. They have been the products of political leaders (even the large PSD is arguably the creation of former president Iliescu), have shown low programmatic coherence, attracted few party members and have been heavily dependent on state subsidies. Their weak relationship with society has resulted in high electoral volatility and instability in terms of party system format. As Table 6.4 shows, only two political parties that ran in the 148

first Romanian parliamentary elections of 1990 PNL and UDMR were still in the parliament in 2004. There has been, however, significant stability in terms of party system mechanics and patterns of inter-party competition, with the Liberals (PNL), Democrats (PD) and Christian-Democrats (PNTCD) opposing to the dominant Social-Democratic Party (PSD). While some political realignments took place, the pattern of competition in the party system generally remained stable. The major change took place after the 2000 elections and the defeat of the centre-right coalition government. The popular dissatisfaction with the performance of the incumbent coalition government mainly favoured the populist PRM, which received almost 30 per cent of the parliamentary seats (after redistribution), becoming the second largest party in the parliament. Confronted with the unexpected electoral growth of PRM, but also aware that coalescing with its former ally would give the party a bad reputation internationally at a time when the party actively sought the membership of the Socialist International, PSD moved closer to UDMR with which formed finally a parliamentary alliance with it. The 2004 elections brought no significant change in the Romanian party system. The format of the system remains unchanged, with the same parties receiving parliamentary representation as in the 2000 elections. In terms of the pattern of party competition, UDMR continued to play a pivotal role between the PSD and the right wing parties (PD and PNL), by joining the centre-right coalition government formed around the D.A. alliance. The small Humanist Party of Romania which was also vulnerable as it was able to enter the parliament in only alliance with the PSD decided to change its coalition partner, and dropped the alliance with PSD for joining the DA government in December 2004. The PRM, after an unsuccessful attempt to join the EPP with a Christian Democratic appeal in 2005, remained very isolated and with other populist parties appearing to receive popular support in public opinion polls, the future prospects of this party are quite pessimistic.

149

6.6. Concluding Remarks

The chapter has presented briefly the four party systems from the CEE: the Czech Republic, Hungary, Poland, and Romania. Although exploratory in its scope, the reader could observe why the initial assumption that the CEE politics is characterized by high political uncertainty is not far-fetched. The political parties in the region emerged, sometimes earned governmental status, and then vanished with a speed that is surprising for scholars accustomed more with the study of party competition in long-established democracies. Has this political uncertainty, together with other factors I emphasized in the previous chapter, had any political consequences? Answering to this question is dedicated the next chapter.

150

CHAPTER 7: PARTY COMPETITION IN THE CEE: IDEOLOGICAL CONVERGENCE AND CENTRIPETAL COMPETITION

7.1. Introduction
In the Chapters 3 and 4, I argue theoretically that the ideological convergence depends not only on the low level of information of parties and voters, but also on the capacity of the established parties to eliminate (or at least reduce) the probability of new entry in electoral competition. I also advance the argument that, to address the risks of new entries in political competition, political parties use strategies which are not necessary limited to electoral behaviour, as many rational choice authors argue. To be more precise, political parties use other instruments which are at their disposal. Designing institutions which prevent or make entry extremely costly is one of such instruments that keep small political parties out of electoral party competition. How established parties make use of existing institutions in order to prevent a state of perfect competition?87 Since competing through divergent policy positions is a risky strategy which may lead to unpredictable outcomes, designing institutions that raise the costs of entry should be the main strategy of political parties that wish to eliminate some of the small parties from political competition and prevent others to enter. Modern political democracies design two main political institutions that affect the political competition among parties. The first institution is the electoral threshold. The fragmentation of party system has been traditionally considered a menace to democratic governance because of the institutional deadlock produced by too many parliamentary parties competing for political power. In this respect, the party fragmentation of the Republic of Weimar and its political consequences provided the empirical support for the
87

In economics, perfect competition denotes a situation in which there are simultaneous many sellers and buyers on the market, so that, ceteris paribus, none of the existing actors can modify unilaterally the market price of commodities by changing the price of the commodity she sells on the existing market.

151

main argument against polarized multiparty systems.88 Thus, most of the modern democracies create electoral systems in which the parties are required to obtain a minimum number (or percentages) of votes in elections in order to be allowed to keep seats in the parliament. The electoral thresholds are institutions which passively protect the established parties from the undesirable (for them) situation of perfect competition. But the electoral thresholds are by no means the only institutions that keep new parties away from competing electorally. Along with electoral thresholds, political parties design other rules which maintain the competition inside a small group of political parties that use political resources coming from outside of party organization, typically from the state. The most well-known resources of this kind are the state financing for political parties and privileged access to media coverage during the electoral and non-electoral periods. As some authors remark, one of the peculiarities of modern democracies is the increasing financial support for political parties coming from the state budget. As Katz and Mair (1995) show in their conceptualization of a new type of party organization, the cartel party, increasing state funding has been a marked characteristic of the western democracies after 1970s, so that it currently represents an important source of financial revenues for parties. Though the proportion of state financing in the total budget of political parties differs from country to country, almost all political democracies make use of this institution to award the established political parties. State financing is mainly directed to political parties that won seats in the parliament, and in almost all cases the amount of money the political parties receive from the state depends on their electoral result. The state financing covers not only the operational costs of political parties between two electoral periods, but it usually contains extra allowances to pay for the expenses of parties in parliamentary elections. In some countries, parliamentary parties even receive a fixed sum for every MP to cover the daily expenses of parliamentary offices. By all standards, the amount of money the political parties receive from the state budget is impressive and, as the largest share of money is directed to the established parties, it represents a major factor in preventing new parties to enter in party competition.
In his seminal analysis of parties and party systems, Sartori (1976) argues that highly fragmented party systems exhibit also a high degree of ideological polarization, which endanger the democratic polity because they intrinsically develop a tendency towards a centrifugal party competition.
88

152

7.2. Institutional Barriers for Entry in the CEE

However, though the argument about the role of the institutional barriers in competition is theoretically straightforward, it is necessary to discern to what extent the political parties in the CEE set up institutional systems which allow them to avoid the competition with new parties. Is there any evidence that the political parties in region design institutions to block the entry of new parties in electoral competition, as the Hypothesis 1 suggests in the Chapter 5? All four political democracies I analyse in this thesis introduced new electoral laws under which the first free elections were organized in the CEE after more than four decades of non-competitive communist regimes. A major characteristic of the electoral laws in the four countries was that they allowed, to different degrees, a proportional representation of voter preferences, and at least three of the countries implemented similar variants of PR system. The Czech Republic, Poland and Romania used a very proportional electoral system during 1990s, which sometimes has been balanced by the dHondt or modified St. Lage formulas.89 The notable difference is Hungary which has designed a mixedcompensatory electoral system built on three tiers: 1) a national list (of minimum 58 eligible seats), 2) a group of 176 seats available for single-member districts with two rounds if no candidate receives an absolute majority of votes in the respective district, and 3) a regional list of maximum 152 seats, which can be presented by political parties that nominated candidates in at least 1/4th of the individual constituencies. As Table 7.1 shows below, the four countries from the CEE did not set up initially institutional barriers for entering in political competition. Romania and Poland designed in 1990 and 1991 electoral laws which did not asked political parties to pass any electoral threshold in order to obtain seats in the parliament. Not surprisingly, the number of parliamentary parties was very high in these two countries after the first free elections. Contrary, the Czech Republic and Hungary had electoral thresholds for individual parties,
89

PR with dHondt formula usually favors the big parties but it maintains a generally high level of proportionality. Poland used a PR system with modified St. Lague formula in 1991 and 2001 elections, which limit the disproportionality between big and small parties that characterizes dHondt formula. In 1993 and 2002 it used the dHondt formula for seat allocation.

153

but not for coalitions of parties. Thus, since the electoral laws were permissive, the probability of entry in the first elections was high (less perhaps in the Czech Republic and Hungary). What happens after the first elections is however a different picture. We do have established parties that struggle not only to win the elections but also to survive politically. They need to compete simultaneously for political power with other established political parties and with potential new parties. We see from the Table 7.1 that after the second round of elections (which comes about 1994-1996), the cost of entry in party competition increases in all four new democracies. The Czech Republic imposed an electoral threshold of 5% of total votes for a coalition of two parties in order to enter in parliament, plus additional 2% for any other party joining a coalition of two. In Hungary, the cost of entering was even higher. For 1994 elections, the new electoral law required an electoral threshold of 5% for individual parties (higher than in 1990) and 10% for a coalition of two parties. The Polish electoral law from 1993 required individual parties to obtain a minimum of 5% and it also set up an electoral threshold of 8% for coalitions. Romania was temporally lagged from this point of view as it required the parties to win a minimum of 3% for 1992 and 1996 elections (the same threshold of coalitions), but it notably raised up the electoral thresholds for parties and coalition in 2000 elections. When looking at the Table 7.1, there is indeed limited change in the electoral thresholds within the countries. After the second round of elections (around 1993-1994), the electoral thresholds stabilize, with the notable exception of Romania. Is this picture significant from the point of view of my hypotheses and how? There are two arguments that electoral laws play an important role in allowing parties to converge ideologically. The first argument is that all countries shifted upward the electoral thresholds after the first round of elections. There is of little importance if after the first elections there is significant variance or not, because the hypothesis and the theoretical reasoning argue that parties devise electoral laws (institutions) to prevent new entries. Once a relative high level is achieved, there is no need to raise them up any further. Moreover, increasing the thresholds above 5% would be very dangerous for established parties, particularly when taking into account the relative high electoral volatility in the region. Furthermore, I show bellow that after the first round of elections there is also a general

154

tendency (with few exceptions) of the political parties to shift in the direction of the median voter on economic dimension and in the direction of the mean of the scale, for cultural dimension.

Table 7.1: Electoral thresholds in the CEE

Country Czech Republic Year of election 1992 1996 1998 2002 Hungary 1990 1994 1998 2002 Poland 1991 1993 1997 2001 1990 1992 1996 2000 Electoral threshold for individual parties (%) 5.0 5.0 5.0 5.0 4.0 5.0 5.0 5.0 No electoral threshold 5.0 5.0 5.0 No electoral threshold 3.0 3.0 5.0 Electoral threshold for coalitions (%) None + 2.0 for each additional party (maximum 11%) + 2.0 for each additional party (maximum 11%) + 2.0 for each additional party (maximum 11%) None + 5.0 for each additional party (maximum 15%) + 5.0 for each additional party (maximum 15%) + 5.0 for each additional party (maximum 15%) None 8.0 8.0 8.0 None 3% 3% 8.0 + 1.0 for each another party after two (maximum though of 10%)

Romania

All data refers to Lower Chamber.

The second argument is that the electoral threshold represents only one type of institutions through which political parties may try to maintain the status-quo. The electoral formula (as I point above) and privileged access of scarce political resources are also important political barriers that supplement the political effects of electoral thresholds. The state financing is the second factor that augments the general tendency against new entries. It has become a widespread phenomenon in the western democracies and it represents an important political resource for political parties in established democracies. But state financing is a much more crucial political resource in the CEE. As I showed in the Chapter 5, the party membership in the region has been considerably lower than the average membership for consolidated western democracies. With low membership, the

155

political parties in the CEE can not obtain the necessary financial resources within the party organization and need to appeal to external opportunities. Like political parties from older democracies, the political parties in the CEE found this alternative source in state financing and private donations. As I already emphasized in Chapter 5, some scholars argue that private donations represent the largest share of money at the disposal of the political parties (Austin and Tjernstrom 2003). However, this source of money is not in the reach of all political parties. Minor parties are usually unappealing to private donors. On the other hand, large parties attract the most of the money from private donations because usually donors expect some returns from the party once it is in power. In this respect, the state funding is a more reliable source of income for political parties, and some authors consider it a critical source for political parties in the CEE (van Biezen and Kopecky 2001). But, nor the state funding is available to all political parties. State budget funds are directed mainly to parliamentary political parties, only a very limited amount of money (and usually exclusively in the electoral period) being available for nonparliamentary parties. Because money is not available between elections, nonparliamentary parties are largely out of the public debate since they can not plan and implement large scale political campaign, nor they can buy audio-visual time or hire professionals to advertise their political label and policy stances.90 In the Czech Republic, the law states that every parliamentary party receives an endowment from the state budget which is proportional to the number of parliamentary mandates the party holds. As a compromise with regional parties, state funding is available for political parties which obtain regional mandates too. For the annual public subsidies, the parliamentary political parties receive six millions Crowns for a result of 3% of parliamentary seats and additional 200,000 for each 0.1% up to 5%. A party which wins more than 5% of total parliamentary seats could obtain from the budget a maximum of ten millions Crowns. This annual public subsidy is supplemented by contributions that are intended to cover partly the expenses of electoral campaign. However, in order to qualify for these sums, a party has to receive at least 1% of the total votes (later on, modified to 3%, depending on the level of electoral threshold).

In some countries, as Romania for instance, political advertising is forbidden except during the electoral period, so that non-parliamentary parties are totally absent from public sphere between elections.

90

156

In Hungary, the tendency to favour the parliamentary parties in the detriment of non-parliamentary parties is also visible. Allowances from the state are available to political parties which receive at least 1% of all votes cast in parliamentary elections. The sum is distributed so that a quarter of the total amount is shared equally by the political parties holding seats in the parliament, whereas the remaining 3/4th of the sum is distributed among parties based on the number of votes that political parties or the candidates gained in the first round of elections. Poland has also a similar principle regarding the distribution of state funding for political parties. The private donations are limited to a maximum of fifteen minimum monthly wages from individuals and to a maximum of hundred minimum monthly wages for donations coming from firms. As a consequence, the state financing plays a very important role for Polish parties. The state financing comes through two main channels. The first channel is an annual subsidy which is available for political parties that won at least 3% (or 6% as a part of coalition) of the popular vote in parliamentary elections. The second channel is an amount of money, proportional to the number of mandates each party holds in the Sejm or Senate, intended to cover the campaign expenses. In Poland, special allowances for parliamentary caucuses and for individual MPs are also available, which cover the additional expenses (not salary though) occurring in their daily political activities (the office costs, for instance). In Romania, according to the law of party finance, the annual budget funds are the equivalent of maximum 0.04% of GDP and they are available for political parties that receive at least 2% of popular votes. In this total sum, there is a based subvention that amounts for 1/3rd of the total sum and which goes directly to the parliamentary parties. Furthermore, there is another separate subvention for individual parliamentary members that it is used to cover the costs of their territorial offices, staff and so forth. Only the amount that remains after the based subvention and the subvention for individual MPs are distributed to parliamentary parties is available for non-parliamentary parties. 91

91

The data about the party financing is taken from Ikstens, Smilov and Walecki (2001).

157

Table 7.2: State subsidies for parliamentary groups in Poland, Czech Republic and Hungary, 19951996 Country Poland Party Label Amount in USD 3,980 3,220 1,522 0,761 0,312 0,293 2,186/1,501 2,156/2,293 1,933/1,022 1,228/0,858 1,062/0,740 0,951/0,742 2,275 2,018 0,764 0,611 0,611 0,471

Democratic Left Alliance Polish Peasant Party Freedom Union Labour Union Confederation for Independent Poland Non-Party Reform Bloc Alliance of Free Democrats Hungary Hungarian Socialist Party Hungarian Democratic Forum Independent Smallholders Fidesz Christian Democratic Peoples Party Civic Democratic Party Czech Republic Social Democratic Party Communist Party Christian Democratic Union Association for the Republic Civic Democratic Alliance Source: Austin and Tjernstrom (2003) pg. 86

Table 7.3: Party financing, Czech Republic and Hungary Country Public Subsidies Private Fundraising 26,2 7.0 Other 26,0 37,2

Czech Republic 47,6 (1995-1996) Hungary (1990-1996) 55,8 Source: van Biezen (2004), pg. 710.

How important is the public financing for political parties in the CEE though? The definite answer to this question is difficult because empirical data is scarce and often unreliable. But the pieces of information we have about the structure of party financing point to dependence of parties of these funds. In a comparative study of party financing, Ikstens, Smilov and Walecki (2001) found that fourteen out of eighteen countries from the CEE provided political parties with state money. The Tables 7.2 and 7.3 above show that the political parties received impressive amount of money from the state in the Czech Republic, Hungary and Poland. The Table 7.3 partially reveals the size of state subsidies in

158

the official party budgets: about half of the official money the political parties spend in Hungary and the Czech Republic comes through public subsidies from the state. The data I present above shows that the parliamentary parties devise institutions which award insiders and discriminate outsiders. I point above to the role of state financing and electoral laws in blocking new entries, but these are not the only instruments available. Privileged media access for parliamentary parties constitutes also a powerful instrument through which non-parliamentary parties are discriminated in favour of those with parliamentary representation. Together with the effects of these institutions, there is also a psychological factor - the development of partisanship that plays a role in blocking new political parties to compete effectively in elections.92 With the time, established political parties build stronger ties with voters who may start identifying psychologically with some parties and vote for them based on this psychological factor rather than on ideological or policy appeals. In such context, new political parties will find even more difficult to gather electoral support, because they have to offset the unfavourable effects of institutional arrangement and the non-policy predispositions of voters. The conclusion about the imbalanced competition between parliamentary and nonparliamentary parties existing in the CEE does not include any normative statement. It surely does not imply that imperfect party competition is a negative (or, by contrary, a positive) political phenomenon, for instance. From the point of view of democratic theory, this unbalanced competition between parliamentary and non-parliamentary parties may look like a serious drawback. At the same time however, the imperfect political competition creates opportunities for what several authors have called institutionalization of party systems, meaning the stabilization of party system in terms of identity of elements (political parties) and patterns of party competition. The institutionalization of political parties and party systems has been considered an important factor of democratic consolidation because voters accept and attach importance on the existence political parties as the main agents of political representation (Mainwaring and Scully: 1995). Rather than attaching a normative implication, my argument should be understood in terms of political consequences: the imperfect political competition has certain political effects regarding ideological competition.
92

This argument about the role of partisanship has been suggested to me by Professor Stephen Whitefield.

159

7.3. On the Position of Median Voter in the CEE Elections

The existence of equilibrium in electoral competition and the viability of ideological convergence as the vote-maximization strategy depend not only on the preferences of selfinterested political leaders or on the institutional arrangements, but also on the distribution of voters preferences. Recalling the existent literature on spatial voting, the reader should acknowledge that the existence of policy preferences which are single-peaked is a sufficient condition for the existence of equilibrium in unidimensional party competition (see Black: 1958). Furthermore, the symmetric distribution of popular (voters) preferences is a favourable, though not sufficient, condition for ideological convergence. Given the magnitude of social and political transformations in the CEE, we would expect a high degree of polarization of voters. However, as the distribution of voters selfplacement on the left-right dimension shows (see Figures 7.1.A to 7.1.D), the polarization is generally low at the beginning of 1990s, with the noticeable exception of the Czech case, where we can observe a more dispersed distribution of ideological positions.93 By contrary, we observe that the polarization of voters is more significant at the later elections, especially again in the Czech Republic and Hungary. However, as we can see above, there is a noticeable single-peaked, concave distribution of ideological positions of voters, which follows the shape of the normal curve (only the distribution of Czech voters in 1992 is peculiar in this respect). The values of median and the mean of the distribution of voters are either identical or very close each other. From the formal point of view, the characteristics of distribution of voters in the CEE show that the median position is a Condorcet winner, compared to the other possible positions on the left-right dimension.94

All figures present the valid responses of respondents only. The missing values were excluded from the analysis. In the Czech Republic, the number of missing values varies from 3% to a maximum of 10% (1992). For Hungary, the highest number of missing values in 22.3% in 1998, but for the rest of elections this number is considerably smaller. The number of missing values is considerably higher in Poland, between 10 and 20% for most the elections and a 32.8% high in 1991. Romania has the largest number of missing values for leftright selfplacements, around one third of the respondents could not place themselves or refused to do it on this dimension (the peak values in 33% in 1991). 94 Recall that a Condorcet winner in spatial party competition is the point that defeats any other point from the feasible set in the pair wise competition.

93

160

Figure 7.1.A: The Left-Right selfplacement of voters, the Czech Republic 1992-2002

Left - Right Selfplacement

Czech Republic, 1992
30 30

Left-Right Selfplacement
Czech Republic, 1996
30

Left-Right Selfplacement
Czech Republic, 2002

20

20

20

10

10

10

Percent

0 Lef t 2 3 4 5 6 7 8 9 Right

Percent

Percent

0 00. LEFT 01. 02. 03. 04. 05. 06. 07. 08. 09. 10. RIGHT

0 0. LEFT 1 2 3 4 5 6 7 8 9 10. RIGHT

Figure 7.1.B: The Left-Right selfplacement of voters, Romania 1991-2002

Left - Right Selfplacement
Romania, 1991
40 30

Left - Right Selfplacement

Romania, 1992
40

Left-Right Selfplacement
Romania, 1996

30 20

30

20

20

10 10

10

Percent

Percent

0 Left 2 3 4 5 6 7 8 9 Right

0 Left 2 3 4 5 6 7 8 9 Right

Percent

0 00. LEFT 01. 02. 03. 04. 05. 06. 07. 08. 09. 10. RIGHT

161

Figure 7.1.C: The Left-Right selfplacement of voters, Hungary 1990-2002

Left - Right Selfplacement

Hungary, 1990
50

Left-right Selfplacement, Hungary 1994

40 30

40

30

20

20 10

Percent

10

0 1
0 Left 2 3 4 5 6 7 8 9 Right

Left-Right Selfplacement
Hungary, 1998
40

Left-Right Selfplacement
Hungary, 2002
30

30

20

20

10
10

Percent

0 00. LEFT 01. 02. 03. 04. 05. 06. 07. 08. 09. 10. RIGHT

Percent

0 0. LEFT 1 2 3 4 5 6 7 8 9 10. RIGHT

162

Figure 7.1.D: The Left-Right selfplacement of voters, Poland 1991-2002

Left - Right Selfplacement

Poland, 1991
40

Left - Right Selfplacement

Poland, 1992
30

30

20

20

10
10

Percent

0 Left 2 3 4 5 6 7 8 9 Right

Percent

0 Left 2 3 4 5 6 7 8 9 Right

Left-Right Selfplacement
Poland, 1997
40

Left-Right Selfplacement
Poland 2001
40

30

30

20

20

10

10

Percent

0 00. LEFT 01. 02. 03. 04. 05. 06. 07. 08. 09. 10. RIGHT

Percent

0 0. LEFT 1 2 3 4 5 6 7 8 9 10. RIGHT

163

Table 7.4: The Median Voter in CEE Elections Median Voter Median Voter Skewness Position Position (11-point scale) (200-point scale)95 Czech Republic 1992 6.0* 10 -.270 1996 5.0 -20 -.233 1998 N/A N/A N/A 2002 5.0 -20 .117 Hungary 1990 5.0* -10 .080 1994 6^ 0 .002 1998 5.0 -20 -.005 2002 5.0 -20 .027 Poland 1991 5.0* -10 .225 1992 5.0* -10 -.001 1997 5.0 -20 -.095 2001 5.0 -20 .166 Romania 1991 5.0* -10 .245 1992 5.0* -10 .014 1996 6.0 0 -.310 2000 N/A N/A N/A *Median voter placements on 10-point left-right scale included in CEEB data file. The centre of the scale is 5.5 compared to 6 on the 11-point left-right scale of CSES. All other entries are the mean placements on a 11point scale. ^ The data for Hungary 1994 (N = 951), three weeks before the elections, used a 7-point left-right scale. The original location of median voter is 4.0. Rescaled to match the 11-point scale I used in the case of the other elections, the new median position is 5.5. Country Year of elections

The position of median voter gives us further confidence that the self-interested, vote-maximizing political parties compete centripetally. Along with the position of the median voter, which is close to the mean point of the scale, the symmetry of distribution of voters is another favourable element of the centripetal competition. The column four from the Table 7.4 shows the values for the skewness, which is a measure of asymmetry of probability distribution (or probability density), that takes value zero for the normal distribution. All symmetrical distribution has the skewness value around the value zero and the value of this indicator increases or decreases with the asymmetry of the distribution.96 From the Table 7.4 we can see that the distribution of voters left-right selfplacements follows quite close the normal distribution, and the deviations are very small. Substantially, this indicator tells us that the left and right cohorts are quite balanced in the CEE elections, so that political parties in the region have electoral incentives to compete centripetally.

95 96

MVP has been rescaled for 200-point left-right scale to match the data on party manifestos (see Chapter 5). Skewness indicator takes values in the closed interval [-1;1].

164

7.4. Ideological Convergence and Centripetal Competition in Competitive Elections: Czech Republic, Hungary, Poland, and Romania in Comparative Perspective
I turn now to address the main empirical hypothesis of this thesis: in electoral competition, the political parties in the CEE compete centripetally around the position of median voter. Because of the limits imposed by the availability of data, I can fully test this hypothesis only for the economic dimension of competition. For cultural dimension, this approach is impossible, as noted in Chapter 5, because there is no longitudinally consistent empirical data about the distribution of voter preferences. Instead, I show that parties compete centripetally on the cultural dimension too, with parties agglomerating the centre of it. Although I suspect that the median voter is also centrally located, I refrain from making any reference to competition for the median voter on this cultural dimension. In the previous sections and chapters I emphasized several key factors that argue in favour of electoral convergence in the CEE: the predominance of leadership within the party organization, the peculiarities of CEE politics and party systems, and the last but not least, the institutional barriers (electoral thresholds, electoral laws, state financing and privileged access to media for parliamentary parties) that favour the established against potential new parties. I argued that the peculiarities of the CEE make the winning of offices to be the only viable political strategy for the political parties in the region because the political resources come from outside the party organization, particularly from the state. Thus, the party organizations in the region are subject to a more accentuated dominance by the leadership (the office holders) than in the western democracies. Since the winning the elections produces political advantages both for the party organization (collective rewards) and for the political leaders (private rewards), we expect to observe a party behaviour which maximizes the chances of winning the election. This maximization behaviour, which the formal analysis for multiparty competition in uncertain environments identifies it as ideological convergence around median or mean voter (see the Chapter 3 for a review), is thus the result of a combination of factors that include: 1) the leaders desire for political

165

power97, 2) the systemic factors that make the CEE politics to be idiosyncratic, and 3) the institutional design, specifically the electoral rules and party financing. The intuition we have about the electoral party competition tells us that in the first free national elections the ideological convergence is not a viable political strategy for two reasons. The first reason is the intensity of political conflict between the former communists and the democratic opposition that characterized the year of 1989. The later rejected any form of association with the former communists, emphasized the need of a total break with the communist past and attempted to obtain the support of people who evaluated negatively the communist regime at the time of the elections. At the same time, the former communists experienced a dramatic decline in their popularity and they were constrained to maintain the support of their extreme supporters, the only ones on whose support they could be certain of. The second reason for policy divergence at the time of the first free elections was the novelty of political parties for the CEE societies. As non-cooperative regimes, the communist countries were essentially one-party states and almost all the new political parties that challenged the Communists in the first free elections were virtually unknown political entities. A good part of the new political parties came into electoral competition without substantial political identities, at best relying on the leadership formed by wellknown figures of anticommunist dissidence. The political labels of these new parties were rather signs of leaders inventiveness than expressions of substantive ideological attachment. Thus, in order to differentiate from the numerous political parties that mushroomed just before the first competitive elections, the new political parties had to take on more extreme policy positions.98

97

This standard assumption of rational choice models is not empirically tested here. I rely, as a validity of this assumption, on testing the implication of the model which is based on this assumption: that parties converge ideologically in empirical situation. 98 The need of individualization is certainly a common feature for all political parties at the time of the first elections, but there are reasons to expect that it is particularly relevant for the parties of the right. As I argued before, former communists had political incentives to show that the ideological difference separating them from the opposition parties was small at the time of the first election in order to reduce the losses in membership and political support or at least to alienate a part of opposition supporters. At the same time, parties of the right needed to cut any similarities with former communist parties if they wanted to take advantage of the important anticommunist sentiment of the electorate, and thus they had to accentuate the ideological differences by taking more extremist positions.

166

After the first competitive elections, some of the parties that won public offices ceased to be unknown political organizations. Being already an established political party represents a significant competitive advantage against parties that are new. This political advantage functions in the same way in which incumbents have a competitive advantage against the opposition parties, for instance (Gelman and King 1990). Since they had already an established identity inside the party system, these parties were concerned with winning the later elections rather than purely with their political survival, and thus they could adopt a maximization behaviour which led them to centripetal political competition. Such type of electoral competition was also in part the result of the manipulation of electoral institutions by established political parties which set up rules and laws that favoured them in competition with other external parties. Also the psychological factor might have played a role: voters start identifying politically with established political parties. For new political parties in competition, these non-policy factors were much more difficult to offset by using only policy appeals. Summing up, political convergence is expected to appear after the first elections and political parties should move closer to this convergence point.

7.4.1. The Ideological Convergence on Economic Left-Right Dimension The magnitude of economic change in the CEE made the economy be the major dimension of party competition. The substantial economic transformations of 1990s shaped the political system and political parties have had to address the enormous economic difficulties that voters came across with during this time. Thus, there is no surprise to see the economic and redistributive issues scoring high within the party manifestos. Table 7.5 presents the aggregate scores of the saliency of all seven policy areas of competition that are included in the coding scheme: External Relations, Democracy and Freedom, Economy, Welfare and Quality of Life, Political System, Fabric of Society and Social Groups. Recall from the Chapter 5 that all entries are percentages of the total coding. The entries from the table are average scores for all countries, parties and elections during 1990 and 2002 (2000 for Romania). A first look on the Table 7.4 reveals the high saliency of the economic issues, captured here by Economy and Welfare and Quality of Life policy 167

domains, which score around 40% of the total coded units for all parties and elections (see the column five of the table). The means of all other domains are significantly smaller than 15% of the total coding units. Also, the distribution of the saliency scores of Economy domain is positively skewed, which suggests than some of the entries are significantly higher than the statistic mean.

Table 7.5: Saliency of Policy Domain in the CEE Domain N Minimum .72 .00 .00 .00 .00 .00 .00 Maximum 78.27 58.54 44.92 44.53 53.13 48.25 67.35 Mean 24.3448 19.3479 7.6288 8.7977 11.7988 11.5356 12.4049 Std. Deviation 12.2812 9.4694 5.9416 7.7218 9.6560 7.8635 10.3328 Skewness 1.327 .383 2.576 1.803 1.994 1.217 2.792

Economy 135 Welfare and 135 Quality of Life External 135 Relations Democracy 135 and Freedom Political 135 System Fabric of the 135 Society Social Groups 135 Source: Klingemann et al 2006.

The high saliency of economy and welfare for political competition maintains when turn to the individual countries. The Tables 7.6.A to 7.6.D show the comparative salience of different dimensions in the four countries, from which we can notice the importance of economic dimension for electoral party competition within the individual countries. The entries are the means and the standard deviations (in brackets) for all parties. The economy remains the major dimension of political competition and together with welfare cover about forty percent of the total policy issues that parties formulate in their electoral appeals. We observe that the importance of the economy slightly increases after the first election, presumably because of the rising economic problems that affected the CEE societies. Thus, the methodological decision of analyzing the party positions on the economic dimension as the principal axis of electoral party competition is backed not only by theoretical rationale but also by substantial empirical evidence. At the same time, the concerns that the economic dimension is not the principal axis of competition for political parties in some of the countries seems to have little empirical

168

support. In Hungary for instance, which has been considered a paradigmatic case in which the economy is less salient for differentiating among political parties, the economic issues score more than 45% of the total coded units, which suggest that the Hungarian political parties regard the economy as the major dimension of competition. The next analytical step is to show to which extent the political parties behave as vote-maximizers, formulating electoral appeals located around the equilibrium point the median voter. But before presenting the evolution of party ideologies in electoral competition, the reader should recall some methodological aspects which are crucial for a proper understanding of party strategies in the CEE. The imperfection of instruments with which we measure the party location makes the full convergence to be improbable empirically. The parsimony and elegance of the formal analysis of party competition can not be equalled by an empirical investigation of party behaviour. Also, the notion of equilibrium in spatial models is an abstract concept. It is reached not measuring empirical phenomena, but using elegant mathematical reasoning. In empirical situations, though, any small imperfection of instruments or error made by the researcher means that convergence in its theoretical meaning is unattainable. Instead, we should look better to the dynamics of party behaviour and observe the systemic tendencies in party competition. Coming back to party behaviour in electoral competition, the Figures 7.2.A to 7.2.D show the extent of party convergence on the economic left-right dimension. The Figures 7.2.A to 7.2.D plot the ideological evolution of political parties as defined by the equations 5.1 and 5.2 in the Chapter 5. Since the party systems of the CEE are notoriously fragmented, I plot here only the positions of the most important political parties from each party system. The dots on graphs represent the party positions on the economic left-right dimension at the time of election. The position of median voter in every election (except the elections where the data has been unavailable, such as the Czech Republic election in 1998 or the Romanian elections in 2000) is also marked on each of the graphs. As we can see from the graphs below, the position of median voter is very stable over time. Usually this position is somehow centre-left, but very close to the centre of the scale. A second observation is that the theoretical prediction of a centrifugal type of party competition at the time of the first election is accurate. The range of party position at the time of the first election goes from 51.2 points in the case of Romanian elections in 1990 to

169

Table 7.6A: Saliency of policy areas in the Czech Republic Policy Domain Economy 1990 15.91 (7.70) Welfare and 19.76 Quality of Life (6.01) External Relations 9.71 (4.76) Democracy and 13.73 Freedom (7.56) Political System 21.46 (18.96) Fabric of the 14.30 Society (7.52) Social Groups 3.86 (3.33 Source: Klingemann et al 2006. 1992 23.57 (7.88) 21.37 (5.95) 8.70 (5.90) 8.40 (4.54) 15.81 (12.70) 13.12 (5.27) 7.59 (3.94) 1996 21.23 (5.94) 23.23 (5.79) 7.38 (1.85) 5.63 (3.11) 9.69 (1.90) 13.23 (3.15) 11.14 (5.09) 1998 22.14 (10.70) 20.26 (6.70) 8.10 (6.17) 9.26 (3.49) 11.22 (3.50) 14.82 (6.41) 11.55 (4.03) 2002 23.38 (4.84) 23.17 (7.54) 9.49 (5.37) 8.14 (4.73) 11.73 (7.24) 14.01 (4.10) 7.48 (5.46)

Table 7.6B: Saliency of policy areas in Hungary Policy Domain Economy 1990 20.31 (8.93) Welfare and Quality 18.89 of Life (9.92) External Relations 8.67 (3.96) Democracy and 11.27 Freedom (7.88) Political System 9.06 (4.34) Fabric of the Society 11.57 (9.37) Social Groups 17.91 (19.54) Source: Klingemann et al 2006. 1994 38.36 (17.63) 18.91 (8.53) 5.00 (2.97) 2.97 (4.29) 10.45 (9.58) 6.68 (6.49) 16.93 (11.84) 1998 19.64 (5.08) 23.52 (10.86) 9.33 (4.33) 5.58 (2.69) 14.16 (6.10) 17.06 (5.73) 9.10 (4.40) 2002 20.95 (3.69) 25.81 (3.02) 3.13 (1.53) 5.02 (5.80) 15.52 (1.16) 9.73 (2.46) 15.66 (3.76)

170

Table 7.6C: Saliency of policy areas in Poland Policy Domain Economy 1991 24.36 (13.99) Welfare and Quality 12.80 of Life (8.85) External Relations 9.07 (7.29) Democracy and 11.12 Freedom (9.44) Political System 12.78 (13.09) Fabric of the Society 11.84 (12.00) Social Groups 7.44 (5.98) Source: Klingemann et al 2006. 1993 35.53 (18.84) 18.49 (4.69) 6.43 (6.44) 6.00 (4.26) 5.76 (3.98) 6.32 (5.46) 14.13 (12.51) 1997 19.04 (9.92) 22.95 (9.16) 4.84 (1.53) 3.58 (2.83) 14.84 (8.75) 12.68 (6.45) 19.69 (23.02) 2001 19.03 (6.38) 20.47 (11.27) 6.95 (6.28) 1.28 (1.51) 21.70 (10.12) 9.47 (7.82) 17.35 (15.97)

Table 7.6D: Saliency of Policy Areas in Romania Policy Domain Economy 1990 24.11 (16.48) Welfare and Quality 24.09 of Life (13.40) External Relations 5.49 (3.22) Democracy and 18.59 Freedom (10.51) Political System 5.04 (4.84) Fabric of the Society 7.97 (6.55) Social Groups 12.90 (5.09) Source: Klingemann et al 2006. 1992 28.23 (10.77) 13.11 (6.69) 12.09 (13.43) 8.86 (2.47) 6.48 (6.00) 17.48 (9.07) 12.47 (6.68) 1996 29.45 (9.85) 18.53 (10.59) 5.19 (3.16) 6.21 (7.09) 7.41 (3.06) 7.88 (3.16) 19.31 (6.87) 2000 29.97 (10.37) 13.31 (8.59) 6.65 (4.87) 5.49 (3.08) 17.44 (7.56) 7.92 (3.34) 17.38 (7.92)

171

23.74 points in the case of Hungary. Romania and the Czech Republic had the most polarized party systems at the time of the first parliamentary elections. In Romania, FSN led by the top figures of former Communist regimes won a comfortable majority in the 1990 election. Inspired by the prominent figure of president Ion Iliescu, the party advocated a reluctant, partial liberalization of the economy without giving up the pre-eminence of the state owned economic sector. By contrast, the opposition parties (the PNL and the PNTCD mostly) advocated a rapid economic transition to market economy and quasi-total privatization of state economy. In the Czech Republic, the high polarization occurred as the KSM did not give up the Marxist-Leninist ideology, refusing any kind of ideological transformation. In Hungary and Poland, there are the centre-right political parties that take more extreme positions in the first election, whereas the former communists came very close to the position of median voter. After the first elections, the ideological distance (the range) between the most extreme political parties decreases, but the trend is not linear. In the Czech Republic, the ideological distance on economic issues has constantly increased after 1996 election, more notably in 2002 elections. Hungary seems to be a special case since the peak ideological distance appeared not at the time of the first election, but in 1994, when it reached 38.57. Poland follows a downward trend as well, but the range stabilizes around the level of 20 points. Romania has also a downward non-linear trend and it records the lowest ideological distance in 2000. However, there is one problem related to the imperfect measurement of ideological distance I apply here. Since the range is by definition the distance between the extreme points, it does not take into consideration the dispersion of parties. As a consequence of the statistical indicator, the existence of outliers might increase artificially the ideological distance into a party system. Thus, in order to use meaningfully the range as an indicator of the ideological distance, we need to observe the presence of outliers. In this regard, we observe that the ideological distance in some elections increases as some political parties take temporarily more extreme ideological positions. In 1994 election in Hungary, for instance, if we were to eliminate the outlier SZDSZ, the ideological distance would decrease to about 20 (out of 38.5). In Poland, UP plays a similar role in 1997 election, increasing the range with about eleven points. In Romania, PNL and PNTCD, which run

172

Figures 7.2: Ideological convergence in the CEE

25 20 15 10 5 0 -5 -10 -15 -20 -25 -30

Figure 7.2A: The economic left-right placements of the Czech political parties
SSD KSM ODA ODS KDU-CLS SPR-RSC US MEDIAN 1992 1996 1998 2002

Left-righ position

Year of election

Source: Adapted after Klingemann et al 2006.

30 25 20 Left-right placements 15 10 5 0 -5 -10 -15 -20 -25

Figure 7.2B The economic left-right placements of the Hungarian political parties
MSZP FIDESZ SZDSZ MDF KDNP FKFPP MEDIAN 1990 1994 1998 Year of election 2002

Source: Adapted after Klingemann et al 2006.

173

Figure 7.2C The economic left-right placements of the Polish political parties
15 10 Left-right placement 5 0 -5 -10 -15 -20 -25 -30 1991 1993 1997 2001 Year of elections
SLD UD/UW KPN PSL MN UP AWS PO LPR SRP MEDIAN

Source: Adapted after Klingemann et al 2006.

Figure 7.2D The economic left-right placements of the Romanian political parties
40 30 20 10 0 -10 -20 -30 -40 1990 1992 1996 2000
Year of election
Source: Adapted after Klingemann et al 2006.

PD PSD PNL PRM UDMR PUNR PNTCD MEDIAN

Left-right placement

174

together in 1996 as the Democratic Convention from Romania (CDR), were also outliers that caused a significant increase of the ideological distance on economic dimension. The Czech Republic seems to be a special case in which a competition between leftist parties SSD and KSM increases the range particularly in 2002 elections, when the incumbent SSD confronted with the improved popularity of KSM. However, as in many other instances in social sciences, we face a noteworthy question: is the picture we see the sign of an ideological convergence and, therefore, does it reflects vote-maximization behaviour of political parties in the region? The Table 7.7 presents the distribution of ideological distance by country and election. Some of the data seem to be quite high in absolute values. However, absolute values are misleading if we do not relate them to the size of the scale. The last column of the Table 7.7 presents an indicator of ideological distance that takes into consideration the range related to the size of the dimension. The relative values are obtained by dividing the absolute values by 200, which represents the length of the dimension.99 The new measurement, the relative ideological distance from the last column, takes now values between 0 (absolute no distance) and 1 (the largest ideological distance, when extreme political parties are located at the extreme points of the dimension). After controlling for the size of dimension, we observe that the polarization of political parties in the CEE is very small. In none of the elections the ideological distance is higher than 0.30. In fact, the political parties in the CEE have occupied, in average, 13% of the ideological space available for party competition. Since the distribution of political parties is around the position of median voter, the conclusion is that competition in the region developed centripetally between 1990 and 2002 on economic left-right dimension, and the political parties in the region behave as vote-maximizers in electoral arena converging towards the MVP.

99

Recall that a party can take a position into the interval [-100; +100].

175

Table 7.7: The absolute and relative ideological distance in CEE, 1990-2002 (left-right economic dimension) Country Absolute Ideological Relative Ideological Election Distance Distance (Range/200) (Range) Czech Republic 1992 39.9 0.20 1996 18.47 0.09 1998 21.99 0.11 2002 35.04 0.18 Mean 28.85 0.14 Hungary 1990 23.74 0.12 1994 38.57 0.19 1998 15.5 0.08 2002 17.14 0.09 Mean 23.74 0.12 Poland 1991 29.37 0.15 1993 19.44 0.10 1997 20.15 0.10 2001 19.87 0.10 Mean 22.20 0.11 Romania 1990 51.2 0.26 1992 25.35 0.13 1996 32.3 0.16 2000 9.27 0.05 Mean 29.53 0.15

One conclusion we have from the evolution of political parties on economic dimension is the intense political dynamics and adaptation. The evolution towards the position of median voter is an evident phenomenon in the CEE politics, but a certain sentiment of chaos is inevitable when looking to the Figures 7.2.A-D. Contrary to what we know from the analysis of party manifestos of political parties in the western democracies, we observe in the CEE party systems a widespread phenomenon of leapfrogging. SZDSZ in Hungary is one such powerful example: it started as a moderate centre-right party on the economic dimension in 1990. The mediocre electoral result together with the competition from more radical MDF and FIDESZ made the party adopt a more radical economic position in 1994. However, the momentum was badly chosen after four years of harsh economic measures taken by the MDF-FIDESZ-SZDSZ government, and the party lost votes heavily. The Democratic Party (PD) in Romania is another example of adaptation and ideological leapfrogging. At the beginning of 1990s, the party was part of FSN which competed electorally on very leftist economic appeals and opposed almost totally to privatization of state sector industries. But the newly formed PD faced in 1992 a political

176

dilemma: to remain in the shadow of the electorally more powerful PDSR (PSD) or to assume a reformist position, perhaps taking advantage of the growing number of voters who considered radical leftist economic measures as obstacles for economic recovery and development? The party chose the reformist path and in 1992 became even more radical than its new allies, the PNTCD and the PNL. How the ideological distance between the political parties and the median voter differs from the point of view of party labels? Are the left parties significantly different in this respect than other party labels? One interesting finding is that party label does not appear to matter. The formal models we use in modelling electoral competition assume, for the sake of generality, that all political parties face similar obstacles and opportunities in competition. However, not always this assumption is correct in empirical analysis. Some political parties have larger membership and more structured internal political organizations. When competing for offices, leaders should keep an eye on the political support of their party members. From the graphs showing ideological party dynamics, it may seem that the leftist parties have smaller policy distance from the median voter, but this means only that they are advantaged in competition by the centre-left location of median voter. The Table 7.8 shows the mean absolute distance between party families and the MVP. The median voter position has been rescaled to match the 200-point scale that is used for decoding the party manifestos (see Table 7.4 above). Then I calculate the absolute ideological distance between party and median voter positions for all political parties (including smaller or ephemera political parties which were not presented in the graphs above) whose party manifestos are analyzed by Klingeman et al (2006). All cases from the Czech Republic 1998 elections and the Romanian 2000 elections were excluded from the results of the Table7.8 because there was no information about the position of median voter. We see that the leftist political parties, particularly the Communists and the Social democrats, have no significantly larger absolute mean distance from the MVP than other party families. By contrary, there are the Liberals and the Conservatives who are in average further away from MVP than the Social Democrats, although the later seem to be more homogenous ideological group. It is interesting to note that the Liberals score a mean

177

absolute distance which is significantly higher the mean absolute distance of all other party families. But the very large standard deviations for this party family suggest that score is largely the result of some outliers. Indeed, looking to individual Liberal parties, the mean of this party family is significantly inflated by two political parties. The Union of Real Politics (UPR) has an absolute ideological distance from the MVP of 77.65 in 1991 Polish elections. Also, the PNL has an ideological distance of 39.53 points from the MVP in 1990 Romanian elections. Without these effects of the two outliers, the absolute mean distance of the Liberal party family would be 17.28 (and SD of 16.00). But the smaller absolute ideological distances of Social democrats and Communists are largely the result of the median voter position in the CEE elections. Since median voter is positioned centre-left in most of the elections in the four countries between 1990 and 2002, the leftist parties needed smaller policy shifts to improve their electoral fortune. The Conservatives and the Liberals were disadvantaged electorally from this point of view but they compensated by moving toward the MVP almost as close as the leftist parties did.

Table 7.8: Mean absolute ideological distance between the L-R party family positions and the median voter Party family origin Ecologist Communist Social democratic Liberal Christian democrat Conservative Nationalist Agrarian Ethnic-regional Special issue Total Mean 5.5062 13.9724 10.2999 22.7871 13.1856 16.7596 8.6734 7.4445 13.0802 10.1984 13.0064 N 2 7 23 15 20 14 14 12 9 2 118 Std. Deviation .5434 10.5280 7.5001 17.8678 8.6122 9.3680 5.6101 5.5045 6.9981 6.5994 10.4065 Std. Error of mean .3842 3.9792 1.5639 4.6134 1.9258 2.5037 1.4994 1.5890 2.3327 4.6665 .9580

We also see from the evolution of political parties that the structure of competition matters. The ideological evolution of SSD for instance has been constrained by the existence and ideological appeals of KSM. As pointed above, the SZDSZ faced the

178

competition of other two major competitors at the economic right, so that the party had to shift its economic position towards a moderate left, benefiting from the more extreme position of MSZP at this part of the spectrum. The US, a party that had a policy position similar to the ODS in 1998 when it appeared in electoral competition, allied with the KDUCLS (a centre party but with notable leftist economic position generated by the Christian Democratic orientation) and speculated in 2002 the distance between the ODS and the median voter. The fragmentation of Polish centre-right and inexistence of any party to cover permanently this policy space generated the appearance of the right-liberals of PO very close to the position of median voter and of the nationalist LPR in 2002 elections, which replaced the defunct AWS and UD/UW.

7.4.2: The Ideological Convergence on Cultural Left-Right Dimension The saliency of policy areas (see again the Tables 7.6.A to 7.6.D) shows that the economic dimension does not cover the entire policy areas used by political parties in electoral competition. On the contrary, a good part of the manifestos content is allocated to institutional and cultural issues, such as around what principles the CEE political systems should be organized, how political democracy should function, what is the role of old traditions, identities and cultures in the new democratic societies, whether and to what extent minorities (especially linguistic and ethnic) should be allowed to express in the public life, whether the integration in the European Union is desirable or not and so forth. I argue that all these separate issues are part of a cultural dimension of political competition in which the left-libertarian values100 clash with the right-authoritarian values.101 This cultural dimension is salient in some of the Eastern European countries because of the historical and societal evolutions which made the creation of national states being a very late phenomenon. Moreover, the CEE is a geographical and political place where large ethnic and linguistic minorities live abroad or inland, where conflicts between majority and
100

For instance, internationalism, individualism, secularism, protection of minorities of all sorts etc. Values like international isolation, nationalism, organization of politics according to religious norms, or comunitarianism.
101

179

minority groups spark very often, and religiousness seems to be very present in social and political life. Many scholars of the CEE politics argue that this cultural dimension is, at least in some of the countries, more salient than the economic dimension either because of historical (so called cultural factors) or because of the specific path of economic and political transition in these countries (Evans and Whitefield 1993).102 And, because in some countries the social conflict may be much more intense as result of their social peculiarities, we might observe noticeable political divergence in electoral competition on the cultural dimension. However, this theoretical argument is not supported by substantial empirical evidence concerning the electoral behaviour of political parties. At least at this level, the electoral competition based on cultural issues does not offer voters a choice among different ideological platforms. By contrary, as the Figures 7.3A to 7.3D show, the degree of policy convergence is even higher on the cultural than on the economic dimension, even in countries in which the cultural issue are considered of having a larger potential for political conflict. Most of the parties, with very rare exceptions, compete centripetally on this political dimension. Table 7.9 shows below the absolute and relative ideological distance (range) among the CEE political parties on the cultural left-right dimension. The most polarized political systems on this dimension are Poland and Romania, but even in these cases the range of party positions is very short. The relative ideological distance, which has been obtained dividing the absolute ideological distance by the length of the scale, shows that the political parties in all four countries occupied significantly less than twenty percent of the ideological space. The mean ideological distance on the cultural dimension for all countries and all elections has actually been 12%, even lower than in the case of competition on economic left-right.

One such example is Kitschelt et al (1999) who argue that the degree of industrialization, bureaucratic development and rationalization of state system before the communist regime, plus the lack of negotiation of economic transition between communist regime and opposition groups in 1989 made that economic dimension be much more salient in Czech Republic than in Hungary where economic transition was negotiated before the formal end of the communism.

102

180

Figure 7.3: Ideological convergence on cultural Left-Right dimension

Figure 7.3A Party placements on cultural dimension, the Czech Republic 12 10 8 6 4 2 0 -2 -4 -6 -8 1992 1996 1998 2002 Year of elections
Source: H.D. Klingemann et al (2006)

Cultural left-rigth placements

CSSD KSBM ODA ODS KDU-CLS SPR-RSC US

Figure 7.3.B Party placemens on cultural dimension, Hungary Cultural left-right placements 30
MSZP

20
FIDESZ

10
SZDSZ

0
MDF

-10
KDNP

-20
FKFPP

-30 1990 1994 1998 2002 Year of elections

Source: H.D. Klingemann et al (2006)

181

Figure 7.3C Party placements on cultural dimension, Poland

70 Cultural left-right placements 60 50 40 30 20 10 0 -10 -20 -30 1991 1993 1997 2001 Year of elections
Source: H.D. Klingemann et al (2006)

SLD UD/UW KPN PSL MN UP AWS PO LPR SRP

Figure 7.3D Party placements on cultural dimension, Romania Cultural left-righ placements 25 20 15 10 5 0 -5 -10 -15 -20 -25 1990 1992 1996 2000 Year of elections
Source: H.D. Klingemann et al (2006)
PD PSD PNL PRM UDMR PUNR PNTCD

182

Table 7.9: The absolute and relative ideological distance in CEE, 1990-2002 (left-right cultural dimension) Country Election Absolute Ideological Relative Ideological Distance Distance (Range/200) (Range) Czech Republic 1992 7.3 0.04 1996 8.71 0.04 1998 16.91 0.08 2002 10.75 0.05 Mean 10.92 0.05 Hungary 1990 49.08 0.25 1994 13.02 0.07 1998 14.32 0.07 2002 7.18 0.04 Mean 20.9 0.10 Poland 1991 14.21 0.07 1993 20.39 0.10 1997 61.88 0.31 2001 38.83 0.19 Mean 33.83 0.17 Romania 1990 19.35 0.10 1992 42.56 0.21 1996 28.00 0.14 2000 23.38 0.12 Mean 28.32 0.14

Hungary and Romania have had similar policy progressions of political parties on the cultural dimension, with party system beings initially polarized but evolving towards a more consensual politics. The isolation and, ultimately, the electoral fiasco of extremist political parties, such as MIEP and FKFPP in Hungary or PUNR in Romania, clearly favoured such evolution towards centripetal competition. Contrary to Hungary and Romania, in the Czech Republic the trend seems to be towards a centrifugal competition, although at the beginning of transition political parties shared much more consensual positions on this dimension. Poland has a complete flat trend on this dimension, with only PSL (temporarily) and LPR deviating from the centre of the dimension. Alike the party competition on economic left-right dimension, the party organization does not play any obvious role on policy adaptation of political parties on cultural issues too. Since all but few political parties compete centripetally, we could hardly link this electoral behaviour to the internal consistency of party organization.

183

It is interesting to note that the extremist parties (parties that enter initially in competition taking extreme position on a dimension) have been punished electorally once when they shift towards more moderate positions. It appears they waste the support of the extreme voters, failing at the same time to convince the moderate voters of their attachment to new policy position. In Romania, the extremist PUNR failed to enter in the Parliament after the incumbent extremist party leadership had been replaced by the new moderate leaders on ethnic issues. Clearly, the party was electorally hurt by the competition with the other extremist party, the PRM, which happened to survive politically in spite of moderating the nationalistic language, but benefiting from very populist appeals the party leader C.V. Tudor campaigned on in the presidential elections of 2000. In Hungary, the authoritarian-populist party of FFKPP also lost parliamentary representation after moderating the appeals both on the cultural and economic dimensions, although it had obtained a fair electoral result in the 1998 elections. Another political party, the Hungarian Christian-Democratic KDNP, lost the parliamentary representation in 1998, after moderating the political platform on cultural issues. However, it is not obvious whether this change determined the electoral failure of the party or the sudden shift to more extreme position on economic dimension. In this respect, an interesting future case study for the future would be the evolution of the LPR in Poland. The structure of political competition though plays a much more significant role for electoral competition on this dimension. As argued in the Hypothesis 3, an empty political space should attract either new political parties or established parties. Probably the most notable example in this respect in the literature is FIDESZ (see Enyedi: 2005), although from my data the ideological shift is less spectacular. The party had in 1990 a leftist position on cultural dimension but it changed its policies towards a conservative platform after 1998, when the electoral support for the conservative MDF and KDNP had faded. Another such example is the conservative ODS which assumed a central position in 1992 and 1996 elections but it shifted to conservative position after 1998. The party was confronted with the competition of communists and took advantage of the political space freed by the shift towards the centre of KDU-CLS and by the vanished SPR-RSC. The PD in Romania offers us another example of ideological adaptation on cultural issue. We have seen than party reshaped its political platform on economic dimension after the break of the

184

FSN in spring of 1992, adopting a reformist, market-oriented platform. At the same time the party redesigned the cultural components of its political appeals and moved from the centre-right position to more cosmopolite and even libertarian policies.103 However, I need to point again that although competition is centripetal, I can not link this finding to the position of median voter. The lack of suitable time series data does not allow me to perform meaningful analysis in order to derive the position of median voter on cultural dimension. The available data covers only in part the policy issues I include under the cultural label here, so that any report based on these data would be partial and perhaps misleading. Therefore, although we see centripetal competition on the second dimension of competition and even though I suspect that the position of median voter should be somewhere near the centre of this dimension, I can not draw any definitive conclusion about the relationship between party goals, behaviour and the position of median voter. Further research with proper mass-level data is needed in this respect.

7.5. Discussion
The results from the previous sections offer a surprising picture of the party competition in the CEE. We have seen that the political parties compete centripetally on two major policy dimensions and the degree of policy collision is very high. Electoral policy overlap has never been foreseen at the level of party competition in the region. By contrary, scholars of democratic transition have been concerned for quite a long time with the possible detrimental effects of political and policy disagreements on the development of political democracy in the region. The economic difficulties and the social and political transformations occurring during 1990s were factors that determined political scientists to expect divergence rather than convergence in national elections. At best, it has been considered that some consent may be reached at the level of governmental policies, under

103

See for instance the open support of then the party leader and todays the president of Romania, Traian Bsescu, for policies in favour of the sexual minorities in 2004 elections.

185

the compelling influence of the international institutions (the EU or IMF) on the national governments during the 1990s. How can we then explain the results? I argued that policy convergence in the CEE has been possible as a result of several factors. The first, and perhaps the most important, factor has been the level of uncertainty that characterizes the politics in the region. Politics has been an uncertain business in the region. The electoral fortunes of political parties were changing vastly from one election to another. The examples are abundant in this regard. Poland is also notorious for the instability of political parties. The PNTCD was initially the main party of centre-right and in 1996 it gave the first non-communist Romanian prime minister after 1945. Four years later, the party failed to enter in the parliament and never recovered. Also, MDF led the opposition parties to a stunning victory against the communists in 1990 elections in Hungary; only eight years later the party was nothing more than a junior partner of FIDESZ. Today the party is about to vanish from the Hungarian party system. The second factor is largely the result of the political uncertainty within the national party systems. The political parties need political resources in order to survive politically. But the peculiarities of CEE politics offer political parties only resources that come from outside: donations from individuals or private companies, but mainly spoils from public offices. Political parties are thus constrained in the long run to compete for winning the elections instead of pursuing specific policy goals. A third factor I emphasized is the institutional settings. The political parties from the parliamentary party system face two types of uncertainty: one which was about the party-voter links, the other about the number and identity of competing actors. Their typical response has been to eliminate the second source of uncertainty by devising political institutions which confine the electoral competition to a relative small number of relevant parties. All these combined factors produce the degree of political convergence I have presented earlier in this chapter. This type of electoral competition is likely to have several consequences, at the theoretical and practical levels, which I discuss them in the next, concluding chapter.

186

CHAPTER 8: CONCLUSIONS: IMPLICATIONS OF ELECTORAL COMPETITION IN THE CEE

8.1. Introduction
In the thesis I have argued that self-interested political parties compete in the CEE parliamentary elections by using colliding policy strategies. Political parties formulate extremely similar ideological stances to appeal to voters under the constrains of uncertainty and other systemic peculiarities of the CEE politics. Moreover, the political parties in the region have been in a constant process of political adjustment, proving their readiness to shift ideologically to a more favourable political position. Methodologically, I based my analytical approach on the findings of formal models of party competition, reviewed extensively in the Chapter 3. The scholars of rational choice have formulated numerous and often conflicting models of party competition which enhanced our theoretical knowledge about the dynamics of party behaviour in electoral arena. For instance, the traditional formal models consider political parties as vote hunters which alter their political ideologies in order to maximize the electoral outcome. However, extensive empirical surveys of party behaviour and party policies in the Western Europe raised doubts about the validity of these early models as instruments for the analysis of party strategies in the real life politics. Political parties in the Western Europe compete in confined ideological spaces and in many political situations they display non-winning political behaviours. As a consequence, a second generation of models has been developed on the assumption that the political parties are genuinely interested in policy outcomes. At the same time, there have been attempts to integrate the two distinctive traditions in formal modelling of party competition (Strm 1990, Strom and Mller 1999, Mller and Strm 1999). The argument is that political parties have simultaneously multiple political objectives, which include winning votes, winning offices and introducing public policies. Thus, a comprehensive research on party behaviour has to take into consideration all these 187

different, yet equally important, political objectives. However, from all these three traditions, the later is the most undeveloped. It has perhaps some empirical support, but the theoretical argument has still to be strengthened. Furthermore, by including all sorts of political motivation, the theory becomes less falsifiable and thus, less helpful for understanding the political behaviour. Throughout this thesis, I based my argument on two assumptions that integrate into the traditional approach. The first assumption is that the political parties are interested in winning as many votes as possible and consequently, they are ready to shift their policy positions in elections to maximize this goal. The second assumption is that voters make their choices according to proximity principle. Both of these assumptions have not been tested empirically in the thesis. However, to the critics who question the validity of these assumptions for the CEE politics, the best argument in supporting these assumptions consists in the capacity of the vote-maximization models to explain a widespread phenomenon in the CEE politics: the party policy convergence on the economic and cultural dimensions of electoral competition. On the contrary, the directional voting and policy-seeking models have opposite implications: political parties deviate from the median or mean positions in competition. Since these models can not offer an adequate explanation of why political parties in the CEE compete centripetally, in spite of some institutions which are traditionally considered to stimulate the policy divergence (multiparty system, PR or similar electoral systems etc), I argue that a model of party competition based on traditional assumptions is more suitable for the analysis of party behaviour in election in the CEE. The thesis also emphasises the crucial role the uncertainty plays in the political process in the region. Fifty years after Anthony Downs published his seminal work it is surprising to observe that probably the most important contribution of the author in political science has been so neglected in the study of political parties. Two parts out of four that form the structure of An Economic Theory of Democracy analyse the influence of information and uncertainty on the political process. Yet, this key legacy of Anthony Downs has received far less scholarly attention, both theoretically and empirically, that it would have deserved given the crucial role of uncertainty for the general political process.

188

The study of information process is important because arguably the politics is one of the most uncertain social environments. Even more influenced by uncertainty is the electoral competition where political parties need to tackle the political uncertainty that steams out of four major sources: 1) the interparty competition, namely the number and identity of party labels, 2) the party-voter linkages, that is the nature of voters political preferences, 3) how the other competitors respond to partys political strategy, and 4) which are the salient issues for political competition. The study of party responses to these sources of uncertainty is crucial for understanding the political process. In this respect, the thesis argues that political parties in the region address the first source of uncertainty by limiting the political resources of other competitors. Moreover, the second and third sources are directly responsible of the convergent pattern of political competition in the region. The fourth source of uncertainty is tackled by formulating broad political appeals which, when aggregated, consign the political parties in the region at the centre of political dimension. The policy convergence and centripetal competition in the CEE are themselves political phenomena that are worth to be acknowledged and explained scientifically. But at the same time, it is likely that they had, have or will have serious political implications for our societies and these implications are not necessarily related strictly to the electoral political competition. The society as a whole is going to be affected in several important ways.

8.2. The Implications of Centripetal Competition for the CEE Democracies

One major debate in the modern democratic theory has been whether the democratic regime is primarily based upon political cooperation or political conflict. During the early stages of modern democratic theory, the political cooperation and harmony were considered to be prerequisites of the true republican government.104 The major challenge of the constitutional bodies was to eliminate the possibility of the conflict inside the republican government. Obviously, this utopian mission had at the foundation the argument that the
104

See for instance the Federalist Papers.

189

leadership and the institutions of republican government had to pursue a common political goal and since only one common goal could exist, the conflict was regarded as menace for the true republicanism. By contrast, the scholars of pluralist democracy, Schumpeter (1950) and Dahl (1972, 1989) particularly, argued that the conflict (or at least dissent) and competition lies at the heart of political democracy. According to this pluralist vision of political democracy, the democratic government is possible because it allows the existence of conflict between different groups of leaders, social groups or even different institutions, and it offers mechanisms of controlling this conflict to reasonable level of intensity. This debate between conflict and cooperation as foundations of political democracy is not just abstract discussion without any visible social consequences. By contrary, it is potentially of great interest for democratization process in the CEE. Coming from a political regime in which opposing behaviours were severely punished, the countries of the region set up norms and rules that institutionalized the conflict in the new democratic societies. Free elections, freedom of association into political parties, competition for public offices are all institutions that allow and moderate the conflict in modern democracies. But at this point, the new democracies of CEE had to address a question that is crucial for the survival of political democracy: How can we allow the political conflict and, at the same time, how can we maintain it into reasonable levels of intensity? There is not a novelty that the societies of CEE countries have underwent systemic transformations during the last sixteen years. The citizens of the region have experienced unprecedented economic hardships and poverty has affected large groups from the population. The transformation of the East-European societies has had impact on the old traditions, existing for generations, too. Thus, in many respects the potential for intense political conflict which could engender a non-democratic upheaval has been very high. Yet, the democracy and democratic practices survived and consolidated not only in the four countries I deal with in the thesis, but in almost all former satellite communist countries in the CEE. How can this be explained? Obviously, factors affecting the democratic survival and consolidation are legion. Many of them are systemic, others are contextual. But one potentially important factor, the type of political competition, has been either neglected or not sufficiently addressed in the literature of democratic transition. At a very general level, the centripetal competition is a

190

factor of political democratization because political parties take moderate policy positions and the potential for political conflict is less acute. The argument is not new and has been repeatedly advanced in the context of two versus multiparty competition. But what is particularly of scientific interest here is not the ordinary relationship between the centripetal competition and the democratic survival. My argument concentrates on the entire causal chain that produces centripetal competition and, ultimately, ensures endurance of democracy in the CEE. The behaviour of parties as vote maximizers in electoral competition has produced centripetal competition which may have played a role in the survival of political democracy. The crucial element in this causal chain has been the impossibility of the extremist parties to appeal successfully to extremist voters for longer periods of time. On the long run, the extreme policy positions have not been politically rewarding and the political parties pressured to survive politically have had to moderate the policy appeals in their attempts of winning the offices needed for political survival. Once they moderated their political appeals, extremist parties became politically unreliable and vanished into the competition for the median voter. The argument is counter-intuitive because it implies that an individualistic behaviour of political parties and political leadership produces an environment favourable to upholding political democracy, even in circumstances in which the odds seemed to be initially very inauspicious. To certain extent, it contradicts the common belief that individualistic behaviour is socially disruptive and it also questions the idea that political parties need to be only agents of policy representation as a prerequisite for democratic consolidation. However, this argument has to be further tested before arguing confidently that the centripetal competition between self-interested political parties has produced a democratic public good in the CEE. For the moment, this argument is rather a hypothesis given that although the theoretical reasoning is thorough, empirical evidence is insufficient at this stage. We have for the moment evidence from four successful cases of democratization in the CEE, but the number is too small and the selection is biased to draw strong conclusions in this regard.

191

8.3. The Implication of Results for Political Competition

A second implication of the centripetal competition is the nature of political linkages between political parties and voters. If the political parties offer voters a very limited range of policy alternatives, how is the political competition likely to evolve? Several alternatives could be the answer to this question. First, borrowing an answer from microeconomics, a rational choice hardliner would argue that as in microeconomics, the equilibrium is only temporarily reached. This means that presumably we would see in the future smaller or larger deviations by some of the parties from the median voter position. They will probably prefer to deviate in some elections from the centre in order to gain the support of more extreme voters or to identify harder with particular policy positions. However, on the long time, I expect parties to continue competing for the median voter, as long as this position is going to be rewarding electorally. The effects of political uncertainty are relevant too. With the time, the political parties and voters will strengthen the political tie and the electoral volatility is likely to decrease. The increase in political certainty of political parties should affect their political strategies in ways that are detrimental to political convergence. The formal models of multiparty competition point at the existence of multiple equilibria when political parties compete under perfect information, of which party convergence at median voter is only one of them. If the political parties improve their level of political information, they may choose strategies that lead to non-convergent pattern of electoral competition. The second alternative is a competition based on valence issues. Very often we see in the region parties and politicians claiming that only they are capable of implementing effectively a socially desirable policy or that the success of a policy depends on whether they would be elected or not. In this sense, the competition is not between opposite policies but between qualities of politicians. Who is more credible, who is more capable or at least who is perceived as being more capable are matters that become crucial under this type of party competition. Under valence competition, the characteristics of leadership and the voters party attachments turn out to be particular important for the result of political competition.

192

The third alternative would be a competition based on leaders personal abilities or what Max Weber called charisma. In order to prove that that party is the most suitable to undertake a certain political policy, the political parties may use the personalities of leaders as promise of their success. Yet, such alternative suffers from two major problems. For political parties, one shortcoming is that the charismatic leaders are not often available and the success of such strategy is very much context dependent. After all, we know about charismatic leaders only after they have had political success and many potential charismatic leaders (in the right place, at the right time) are only political caricatures of in more quiet times. Moreover, the second major drawback of a competition based on charisma is that it undermines the foundations of political democracy. The political democracy is not the result of exceptional leaders but of exceptional institutions. Under the leadership of strong personalities who find legitimacy for their political power not in democratic procedures but in personal abilities, the political democracy may be at stakes. The political parties as democratic institutions may be overlooked in the political process because charismatic leaders may find easier to address directly to voters. Poland and Romania, countries which hold institutional settings that are favourable to personality politics - strong presidents elected directly by voters - are particularly auspicious to this type of party competition. Combined with the fragility of political parties in the region, the electoral competition based on leaders charisma is a serious potential danger for quality of democracy if not for democracy itself. The fourth alternative is the development of the partisan affiliations in which the party choice is conditioned by voters psychological affinity rather than by objective policy distance. In elections, a voter may vote for a party not because their policy positions are very close each other, but because of her past political behaviour. Yet, though party affinity does not necessarily involve policy affinity, the development of the partisanship has to be somehow linked theoretically with policy positions of party and voter. It is implausible that a voter can develop partisan affiliation to a political party that systematically proposes policies in contradiction with the voters ideal policy positions. The development of partisan affiliation plays a role in the process of decreasing the political uncertainty too. Building closer ties with the voters, the political parties enjoy a more secure political environment. The increase in the level of information should produce

193

more polarized party systems in the region because political parties would need to present policy programs which do not alienate their traditional voters. At the same time, partisanship is likely to play a decisive role in a political competition with valence issues, as I have mentioned above. A fifth potential evolution of the party competition is the issue-based voting. Although political parties could be very close each other on broad political dimensions, such as the cultural or economic left-right scales I used in this thesis, they also could take distinctive positions on particular issues that come and go on the public agenda. Many of these issues are very relevant for the public and thus the potential for conflict and divergence inside the party systems is quite high. Issue-based competition is far less structured than a competition based on ideology and political values though. Yet, this kind of political competition may be more relevant for the societies of CEE if the political parties fail to present in elections with clear-cut ideological positions or if long-standing political affinities are not going to develop within the electorate. At the same time, the development of partisanship may be linked to this issue-based competition rather than to positions on the broad ideological dimensions, although it certainly requires a high level of political sophistication from voters, an assumption which is often debatable in the modern democracies.

8.4. The Implication for Political Representation

As I argued in the introductory chapter, our normative expectations require political parties to reflect in political competition the interests and conflicts that are manifest within the society. The programs of political parties should address those issues that are high on public agenda and they should also give voters a choice between different political alternatives. The reflection of constituencys interests and endorsement of the policy alternatives form the process of substantial policy representation. But the existence of centripetal competition and the small ideological distance between the political parties in the CEE raise a noticeable difficulty for the substantial

194

policy representation. The political parties offer a very limited range of policy alternatives and a good part of the electorate is left aside of this political process. The major implication of such phenomenon is that many of the voters could be alienated and political parties may lack legitimacy as political agents of the democratic government. On the long span, either the political democracy will be at risk if agents as political parties are becoming less credible or other institutions will start performing this function of political representation at the expenses of political parties. As we have seen in the previous chapter, the centripetal competition and policy collision are pervasive political phenomena in all four countries, with Romania and Poland being at the top of the group. As result, we would expect to see voters feeling unrepresented by the political parties in their countries. Also, the alienation of voters should be higher in countries where the party ideological convergence is more noticeable, such as Romania and Poland. Indeed, the Table 8.1 shows that the perceived level of political representation is higher in Hungary and the Czech Republic than in Romania or Poland. The differences among countries are quite significant, from about +50% in the Czech Republic and Hungary to -10% in Romania and about -20% in Poland. However, at the same time it is surprising to see that people still feel represented to large extent in spite of the political discourse of parties at the time of elections. How can this be interpreted?

Table 8.1. Political representation in the CEE Country and year of elections Is there a party that represents respondents views? Total

Yes Czech Republic (2002) 56.2% Hungary (2002) 71.6% Poland (2001) 33.9% Romania (2004) 39.5% Source: CSES, Module 2, 2001-2006

No 15.9% 26.2% 50.7% 48.2%

Missing 27.8% 2.2% 15.4% 12.3

100.0% 100.0% 100.0% 100.0%

One answer may reside in the process of voters projection or rationalization. Recall from the Chapter 3 that one argument against the proximity voting model is that voters may bring the favourite party closer to their ideal policy position than it actually is. This argument, usually called projection or rationalization of party positions, may have an influence on the fact that voters feel represented in a political environment that objectively

195

is detrimental to substantial political representation. If the voters rationalize the distance between the preferred parties and their ideal policy positions, then they may regard political parties as responsive political agents.

Table 8.2: Rationalization and political representation Country Party Supporters L-R placements Others L-R placements 3.40 (2.64) 6.68 (2.58) 5.08 (2.96) 3.74 (2.34) 5.35 (2.39) 5.91 (2.88) 3.89 (3.11) 4.07 (3.19) 4.01 (3.17) Party Position (Supporters only) 8.28 (1.88) 1.85 (2.18) 2.86 (1.72) 8.55 (1.33) .57 (1.05) 1.24 (2.23) 7.60 (2.13) 5.35 (2.70) 5.62 (2.81) Party Position (Nonsupporters only) 8.39 (1.93) 1.68 (1.81) 2.64 (1.64) 8.22 (1.90) .83 (1.99) 1.42 (2.27) 6.29 (2.43) 4.27 (2.02) 4.36 (2.33)

Hungary

FIDESZ 7.28 (2.19) MSZP 2.65 (2.12) The Czech CSSD 3.52 (1.59) Republic ODS 7.49 (1.66) KSCM 1.55 (1.40) Poland SLD-UP 2.08 (1.97) PO 6.67 (2.27) PSL 4.96 (2.54) S (Self Defense) 5.12 (2.80) Source: CSES, Module 2, 2001-2006

The Table 8.2 displays the positions of the political parties on the left-right dimension as perceived by voters this time. The column three presents the mean positions of voters who declared they felt represented by the respective political party. The forth column shows the mean location of the voters who declared they were represented by other political party. The last two columns are the partys mean positions as perceived by these two categories of voters. The Table 8.2 includes only the parties that scored relatively high (more than fifty persons). Romania is not included because there was no individual-level data for 2000 elections in the CSES Module 2. As we can see from the data, rationalization does not occur in the CEE elections, at least for the above cases. If rationalization had been a pervasive phenomenon in the CEE elections, we would have observed persistent differences between the supporters and nonsupporters mean scores of political parties. If the supporters had brought the preferred parties closer to their ideal positions, we would have seen systematic differences between the mean locations as perceived by supporters and non-supporters. By contrary, we see from the table above that there is a striking agreement between the two categories of voters about the positions of political parties on the left-right scale. The mean party position and the mean position of supporters are very similar indeed, with parties perceived positions

196

being more extreme than the positions of their constituencies. At the same time, the ratings of political parties by non-supporters are very similar to the evaluations of party supporters. How can we then explain that a large part of population feel represented by political parties? Obviously, I can not explore in depth this question here but several possible explanations may be relevant for our cases. One explanation could be that voters do not take into consideration only the positions expressed by political parties in election when they estimate the party locations. Thus, the sentiment of being represented by a political party may relate to political phenomena which are outside of electoral competition. Also, another explanation could be that psychological attachments to political parties matter. The argument is that people with partisan attachments to parties may feel represented even though in elections political parties propose strikingly similar ideologies. The third possible argument regards the individual data I use to assess the level of political representation. The CSES series are post-elections surveys done in about a month time after the end of electoral process. This proximity of elections may have an influence on the degree to which individuals feel represented by political parties. Since electoral process is still fresh in the memory of voters, they might report feeling represented by the party they voted for in election. Their reply to the question might not signify a substantial policy representation, but the expression of a political choice made in elections. In any case, there is an intriguing discrepancy between the relative high level of political representation and the high level policy proximity of parties in elections. Although further analysis is required, it is improbable that substantive policy representation could be the cause of high figures concerning the level of representation of voters presented above.

8.5. Final Remarks:

As I show above, the findings of this research give birth to questions that have huge potential for further scientific research in politics. Has central competition played any role in democratic consolidation? Apart from the general trend of centripetal competition, why some countries in the region (the Czech Republic and Hungary) have more dispersed party

197

systems than others (Poland and Romania)? Does the partisan affiliation play any role in it? Has this limited polarization of party systems any impact on the way political democracy is evaluated by the citizens of the CEE countries? What about the impact on the way the political parties are perceived in the CEE? Also, the two assumptions which remained untested here, the proximity model of voting and the parties as vote-seekers, need closer inspection in future research. For instance, Todosievi (2005) finds that the directional model of voting outperforms the proximity model in Hungary and, quite surprisingly, the vote is better predicted by the positions on the left-right dimension than on other specific policy issues. This result conflicts with the data on the party positions on the left-right dimension I report in the Chapter 7, where the Hungarian parties are located very close to the median and mean voters. Obviously, closer analysis is needed in this regard.

198

APPENDIX

Appendix 1: Formal proof of Wittmans alternative competitive model (Chapter 3, Section 3.5.4)

Consider an election that concerns the policy regarding income distribution in which three voters (X, Y, and Z) need to select a policy from four alternative income distribution platforms (A, B, C, and D). Voters are interested solely in the personal income they obtain after a platform is selected. Individual preferences are shown in the table below.

Platforms A B C D
Source: Wittman: 1973, pg. 496.

A 1/3 1/2 0 1/4

Voter Allocation of income to voter B 1/3 1/2 3/4 1/4

C 1/3 0 1/4 1/2

Such a distribution of preferences leads to an intransitive social preference, in the sense that no platform wins a majority of votes in pairwise context. Because intransitivity is a characteristic of such society, the incumbent party knows that the opposition is able to propose another platform that will defeat its own most preferred platform in an electoral contest. However, Wittman (1973, 496) argues that the incumbent has a strategy of choosing a platform that will force the opposition to offer the incumbent as much as possible. Imagine an example of five voters (X, Y, A, B, and C) who vote into elections. This time, A, B, and C are voters and should not be confused with the platforms of above 199

example. From these five voters, only X and Y can advance platforms in election, and for the sake of the argument let X be the incumbent party and Y the opposition party. If the incumbent party X advances, lets say, a platform like [X(1/2-2e), Y(0), A(1/3+e), B(0), C(1/3+e)], where e is a very small number and the expressions in the brackets show the share of income each voter receives, then Y can reply with another platform like [X(1/3), Y(2/3-d), A(0), B(d), C(0)], where d is a very small number). By assumption that each voter is interested in its own revenues, X, Y and B should prefer the platform of Y to the incumbents platform. Advancing successive platforms, Wittman argues that parties (in our case X and Y) will keep almost all the income for them, providing the voters with only a small share of total income.

200

Appendix 2: Formal proof of Davis and Hinich (1968) (see the Chapter 3, Section 3.6)

Consider a voter whose most preferred positions in multidimensional policy space are a vector X = {x1,,xn}. This voter prefers party to party if and only if ||x-||A < ||x-||A where: ||x-||A2 = (x-)A(x-) (2) and A is a positive definite n x n matrix. ||x-||A2 is the utility loss that an individual feels when the policy is implemented instead of his most preferred policy. Now let , which is the mean position, be the position of one party. According to (1), a voter votes for instead of if and only if: ||x-||A2 < ||x-||A2 By (2), this is true only if: 2(x-)A(-) < ||-||A2 instead of , we have: P(y < ||-||A2), y = 2(x-)A(-) where (4) (3) Given P(||x-||A2 < ||x-||A2) being the probability of one individual voting for (1)

Consider that the policy space is a multinormal distribution of preferences, which makes that mean position is also the median position. Thus, P(y<0) = P (y>0) = 1/2 Since (||-||A2>0)105 and for any fix A, then: P(y<||-||A2|A) > 1/2, thus P(y<||-||A2) = EA[P(y<||-||A2)A] > 1/2, which implies that: P(||x-||A2 < ||x-||A2) > 1/2. Q.E.D.

105

Since

201

Appendix 3: Party Manifestos Coding Scheme

Domain 1: External Relations per101 Foreign Special Relationships: Positive Favourable mentions of particular countries with which the manifesto country has a special relationship. For example, in the British case: former colonies; in the German case: East Germany; in the Swedish case: the rest of Scandinavia; the need for cooperation with and/or aid to such countries. per102 Foreign Special Relationships: Negative Negative mentions of particular countries with which the manifesto country has a special relationship; otherwise as 101, but negative. per103 Anti-Imperialism: Positive Negative references to exerting strong influence (political, military or commercial) over other states; negative references to controlling other countries as if they were part of an empire; favourable mentions of decolonisation; favourable references to greater selfgovernment and independence for colonies; negative references to the imperial behaviour of the manifesto and/or other countries. per104 Military: Positive Need to maintain or increase military expenditure; modernising armed forces and improvement in military strength; rearmament and self-defence; need to keep military treaty obligations; need to secure adequate manpower in the military. per105 Military: Negative Favourable mentions of decreasing military expenditures; disarmament; evils of war; promises to reduce conscription, otherwise as 104, but negative. per106 Peace: Positive Peace as a general goal; declarations of belief in peace and peaceful means of solving crises; desirability of countries joining in negotiations with hostile countries. per107 Internationalism: Positive Need for international cooperation; cooperation with specific countries other than those coded in 101; need for aid to developing countries; need for world planning of resources; need for international courts; support for any international goal or world state; support for UN.

202

per108

European Integration: Positive Favourable mentions of European Community in general; desirability of expanding the European Community and/or of increasing its competence; desirability of the manifesto country joining (or remaining a member).

per109

Internationalism: Negative Favourable mentions of national independence and sovereignty as opposed to internationalism; otherwise as 107, but negative.

per110

European Integration: Negative Hostile mentions of the European Community; opposition to specific European policies which are preferred by European authorities; otherwise as 108, but negative. Domain 2: Freedom and Democracy

per201

Freedom and Human Rights: Positive Favourable mentions of importance of personal freedom and civil rights; freedom from bureaucratic control; freedom of speech; freedom from coercion in the political and economic spheres; individualism in the manifesto country and in other countries.

per202

Democracy: Positive Favourable mentions of democracy as a method or goal in national and other organisations; involvement of all citizens in decisionmaking, as well as generalised support for the manifesto countrys democracy.

per203

Constitutionalism: Positive Support for specific aspects of the constitution; use of constitutionalism as an argument for policy as well as general approval of the constitutional way of doing things.

per204

Constitutionalism: Negative Opposition to the constitution in general or to specific aspects; otherwise as 203, but negative. Domain 3: Political System

per301

Decentralisation: Positive Support for federalism or devolution; more regional autonomy for policy or economy; support for keeping up local and regional customs and symbols; favourable mentions of special consideration for local areas; deference to local expertise.

per302

Centralisation: Positive Opposition to political decision-making at lower political levels; support for more centralisation in political and administrative

203

procedures; otherwise as 301, but negative. per303 Governmental and Administrative Efficiency: Positive Need for efficiency and economy in government and administration; cutting down civil service; improving governmental procedures; general appeal to make the process of government and administration cheaper and more effective. per304 Political Corruption: Negative Need to eliminate corruption, and associated abuse, in political and public life. per305 Political Authority: Positive Favourable mentions of strong government, including government stability; manifesto partys competence to govern and/or other partys lack of such competence. Domain 4: Economy per401 Free Enterprise: Positive Favourable mentions of free enterprise capitalism; superiority of individual enterprise over state and control systems; favourable mentions of private property rights, personal enterprise and initiative; need for unhampered individual enterprises. per402 Incentives: Positive Need for wage and tax policies to induce enterprise; encouragement to start enterprises; need for financial and other incentives such as subsidies. per403 Market Regulation: Positive Need for regulations designed to make private enterprises work better; actions against monopolies and trusts, and in defence of consumer and small business; encouraging economic competition; social market economy. per404 Economic Planning: Positive Favourable mentions of long-standing economic planning of a consultative or indicative nature, need for government to create such a plan. per405 Corporatism: Positive Favourable mentions of the need for the collaboration of employers and trade union organisations in overall economic planning and direction through the medium of tripartite bodies of government, employers, and trade unions. This category was not used for Austria up to 1979, for New Zealand up to 1981, and for Sweden up to 1988.

204

per406

Protectionism: Positive Favourable mentions of extension or maintenance of tariffs to protect internal markets; other domestic economic protectionism such as quota restrictions.

per407 per408

Protectionism: Negative Support for the concept of free trade; otherwise as 406, but negative. Economic Goals Statements of intent to pursue any economic goals not covered by other categories in domain 4. This category is created to catch an overall interest of parties in economics and, therefore, covers a variety of economic goals.

per409

Keynesian Demand Management: Positive Demand-oriented economic policy; economic policy devoted to the reduction of depressions and/or to increase private demand through increasing public demand and/or through increasing social expenditures.

per410

Productivity: Positive Need to encourage or facilitate greater production; need to take measures to aid this; appeal for greater production and importance of productivity to the economy; increasing foreign trade; the paradigm of growth.

per411

Technology and Infrastructure: Positive Importance of modernisation of industry and methods of transport and communication; importance of science and technological developments in industry; need for training and research. This does not imply education in general (see category 506).

per412

Controlled Economy: Positive General need for direct government control of economy; control over prices, wages, rents, etc; state intervention into the economic system.

per413

Nationalisation: Positive Favourable mentions of government ownership, partial or complete, including government ownership of land.

per414

Economic Orthodoxy: Positive Need for traditional economic orthodoxy, e.g. reduction of budget deficits, retrenchment in crisis, thrift and savings; support for traditional economic institutions such as stock market and banking system; support for strong currency.

per415

Marxist Analysis: Positive Positive references (typically but not necessary by communist

205

parties) to the specific use of Marxist-Leninist terminology and analysis of situations which are otherwise uncodable. This category was not used for Austria 1945-1979, for Australia, Japan and the United States up to 1980; for Belgium, Ireland, The Netherlands and New Zealand up to 1981; for Italy and Britain up to 1983; for Denmark, Luxembourg and Israel up to 1984; for Canada, France and Sweden up to 1988. per416 Anti-Growth Economy: Positive Favourable mentions of anti-growth politics and steady state economy; ecologism; Green politics; sustainable development. This category was not used for Austria 1945-1979, for Australia, Japan and the United States up to 1980; for Belgium, Ireland, The Netherlands and New Zealand up to 1981; for Italy and Britain up to 1983; for Denmark, Luxembourg and Israel up to 1984; for Canada, France and Sweden up to 1988; and for Norway up to 1989. Test codings, however, have shown that parties before the beginning of the 1990s hardly ever advocated anti-growth policies. Domain 5: Welfare and Quality of Life per501 Environmental Protection: Positive Preservation of countryside, forests, etc.; general preservation of natural resources against selfish interests; proper use of national parks; soil banks, etc; environmental improvement. per502 Culture: Positive Need to provide cultural and leisure facilities including arts and sport; need to spend money on museums, art galleries etc.; need to encourage worthwhile leisure activities and cultural mass media. per503 Social Justice: Positive Concept of equality; need for fair treatment of all people; special protection for underprivileged; need for fair distribution of resources; removal of class barriers; end of discrimination such as racial or sexual discrimination, etc. per504 Welfare State Expansion: Positive Favourable mentions of need to introduce, maintain or expand any social service or social security scheme; support for social services such as health service or social housing. This category excludes education. per505 Welfare State Limitation: Positive Limiting expenditure on social services or social security; otherwise as 504, but negative.

206

per506

Education Expansion: Positive Need to expand and/or improve educational provision at all levels. This excludes technical training which is coded under 411.

per507

Education Limitation: Positive Limiting expenditure on education; otherwise as 506, but negative. Domain 6: Fabric of Society

per601

National Way of Life: Positive Appeals to patriotism and/or nationalism; suspension of some freedoms in order to protect the state against subversion; support for established national ideas.

per602

National Way of Life: Negative Against patriotism and/or nationalism; opposition to the existing national state; otherwise as 601, but negative.

per603

Traditional Morality: Positive Favourable mentions of traditional moral values; prohibition, censorship and suppression of immorality and unseemly behaviour; maintenance and stability of family; religion.

per604

Traditional Morality: Negative Opposition to traditional moral values; support for divorce, abortion etc.; otherwise as 603, but negative.

per605

Law and Order: Positive Enforcement of all laws; actions against crime; support and resources for police; tougher attitudes in courts.

per606

Social Harmony: Positive Appeal for national effort and solidarity; need for society to see itself as united; appeal for public spiritedness; decrying anti-social attitudes in times of crisis; support for the public interest.

per607

Multiculturalism: Positive Cultural diversity, communalism, cultural plurality and pillarisation; preservation of autonomy of religious, linguistic heritages within the country including special educational provisions.

per608

Multiculturalism: Negative Enforcement or encouragement of cultural integration; otherwise as 607, but negative. Domain 7: Social Groups

per701

Labour Groups: Positive Favourable references to labour groups, working class, unemployed; support for trade unions; good treatment of manual and other

207

employees. per702 per703 Labour Groups: Negative Abuse of power of trade unions; otherwise as 701, but negative. Farmers: Positive Support for agriculture and farmers; any policy aimed specifically at benefiting these. per704 Middle Class and Professional Groups: Positive Favourable references to middle class, professional groups, such as physicians or lawyers; old and new middle class. per705 Underprivileged Minority Groups: Positive Favourable references to underprivileged minorities who are defined neither in economic nor in demographic terms, e.g. the handicapped, homosexuals, immigrants, etc. per706 Non-economic Demographic Groups: Positive Favourable mentions of, or need for, assistance to women, old people, young people, linguistic groups, etc; special interest groups of all kinds. Data Entries Per1011 per7062 per1011 Subcategories used for CEE countries. For comparisons between OECD and CEE countries, subcategories can be aggregated into one of the 56 standard categories used in all countries. Russia/USSR/CIS: Positive Favourable mentions of Russia, the USSR, the CMEA bloc or the Community of Independent States. per 1012 Western States: Positive Favourable mentions of Western states, including the USA and Germany. Eastern European Countries: Positive Favourable mentions of Eastern European countries in general. Baltic States: Positive Favourable mentions of the Baltic states, including other states bordering the Baltic Sea. Nordic Council: Positive Favourable mentions of the Nordic Council. SFR Yugoslavia: Positive Favourable mentions of countries formerly belonging to SFR Yugoslavia including special relationships with Montenegro, Macedonia, Slovenia, Croatia and Bosnia-Hercegovina. Russia/USSR/CIS: Negative Negative mentions of Russia, the USSR or the Community of

per1013 per1014

per1015 per1016

per1021

208

Independent States. per1022 Western States: Negative Negative mentions of Western states, including the USA and Germany. East European Countries: Negative Negative mentions of Eastern European countries in general. Baltic States: Negative Negative references to the Baltic states. Nordic Council: Negative Negative references to the Nordic Council. SFR Yugoslavia: Negative Negative mentions of countries formerly belonging to SFR Yugoslavia including negative references to Montenegro, Macedonia, Slovenia, Croatia and Bosnia-Hercegovina. Russian Army: Negative Need to withdraw the Russian army from the territory of the manifesto country; need to receive reparations for the damage caused by the Russian army or other Soviet institutions. Independence: Positive Favourable mentions of the independence and sovereignty of the manifesto country. Rights of Nations: Positive Favourable mentions of freedom, rights and interests of nations. Transition to Democracy General references to the transition process of one-party states to pluralist democracy. Restrictive Citizenship: Positive Favourable mentions of restrictions in citizenship; restrictions in enfranchisement with respect to (ethnic) groups. Lax Citizenship: Positive Favourable mentions of lax citizenship and election laws; no or few restrictions in enfranchisement. Presidential Regime: Positive Support for current presidential regime; statements in favour of a powerful presidency. Republic: Positive Support for the republican form of government as opposed to monarchy. Checks and Balances: Positive Support for checks and balances and separation of powers, and specifically for limiting the powers of the presidency by increasing

per1023 per1024 per1025 per1026

per1031

per1032

per1033 per2021

per2022

per2023

per2031

per2032

per2033

209

legislative/judicial powers, or transferring some executive powers to the legislature or judiciary. per2041 Monarchy: Positive Support for a monarchy, including conceptions of constitutional monarchy. Republican Powers: Positive Favourable mentions of stronger republican powers. Public Situation: Negative Negative references to the situation in public life after the founding elections. Communist: Positive Co-operation with former authorities/communists in the transition period; pro-communist involvement in the transition process; and 'let sleeping dogs lie' in dealing with the nomenclature. Communist: Negative Against communist involvement in democratic government; weeding out the collaborators from governmental service; need for political coalition except communist parties. Rehabilitation and Compensation: Positive References to civic rehabilitation of politically persecuted people in the communist era; references to juridical compensation concerning communist expropriations; moral compensation. Political Coalitions: Positive Positive references to the need of broader political coalition; need for co-operation at the political level; necessity of collaboration among all political forces. Privatisation: Positive Favourable references to privatisation. Control of Economy: Negative Negative references to the general need for direct governmental control of the economy. Property-Restitution: Positive Favourable references to the physical restitution of property to previous owners. Privatisation Vouchers: Positive Favourable references to privatisation vouchers. Social Ownership: Positive Favourable references to the creation or preservation of co-operative or non-state social ownership within a market economy. Mixed Economy: Positive Favourable references to mixed ownership within a market

per3011 per3051

per3052

per3053

per3054

per3055

per4011 per4012

per4013

per4014 per4121

per4122

210

economy. per4123 per4124 Publicly-Owned Industry: Positive Positive references to the concept of publicly-owned industries. Socialist Property: Positive Positive references to socialist property, including public and cooperative property; negative references to privatisation. Property-Restitution: Negative Negative references to the physical restitution of property to previous owners. Privatisation: Negative Negative references to the privatisation system; need to change the privatisation system. Private-Public Mix in Culture: Positive Necessity of private provisions due to economic constraints; private funding in addition to public activity. Private-Public Mix in Social Justice: Positive Necessity of private initiatives due to economic constraints. Private-Public Mix in Welfare: Positive Necessity of private welfare provisions due to economic constraints; desirability of competition in welfare service provisions; private funding in addition to public activity. Private-Public Mix in Education: Positive Necessity of private education due to economic constraints; desirability of competition in education. The Karabakh Issue: Positive Positive references to the unity of Karabakh and Armenia or the recognition of the independent Republic of Karabakh; rendering assistance to Karabakh. Rebuilding the USSR: Positive Favourable mentions of the reunification of all republics and nations living on the former territory of the USSR into a new common (democratic) state or into a common economic space whereby the new union would be the guarantor of the manifesto country's sovereignty; negative references to the dissolution of the USSR and the respective treaties. National Security: Positive Support for or need to maintain national security in all spheres of social life; policies devoted to this goal. Cyprus Issue All references concerning the division of Cyprus in a Greek and a Turkish part. General Crisis

per4131

per4132

per5021

per5031 per5041

per5061

per6011

per6012

per6013

per6014

per6061

211

Identification of a general crisis in the country. per6071 per6072 per6081 per7051 Cultural Autonomy: Positive Favourable mentions of cultural autonomy. Multiculturalism pro Roma: Positive Favourable mentions of cultural autonomy of Roma. Multiculturalism pro Roma: Negative Negative mentions of cultural autonomy of Roma. Minorities Inland: Positive References to manifesto country minorities in foreign countries; positive references to manifesto country minorities. Minorities Abroad: Positive References to ethnic minorities living in the manifesto country such as Latvians living in Estonia. War Participants: Positive Favourable mentions of, or need for, assistance to people taking part in the war on the territory of ex-Yugoslavia. Refugees: Positive Favourable mentions of, or need for, assistance to people who left their homes because of the war (for instance, on the territory of exYugoslavia) or were forcibly displaced. Percentage of uncoded (quasi-) sentences Total number of quasi-sentences

per7052

per7061

per7062

Peruncod Total

212

BIBLIOGRAPHY

Adams, James, Samuel Merrill III, Bernard Grofman. 2005. A Unified Theory of Party Competition: A Cross-National Analysis Integrating Spatial and Behavioural Factors. Cambridge: Cambridge University Press. Adams, James, Samuel Merrill III. 1999. Modelling Party Strategies and Policy Representation in Multiparty Elections: Why Are Strategies So Extreme? The American Journal of Political Science, 43: 235-55. Adams, James. 2001. A Theory of Spatial Competition with Biased Voters: Party Policies Viewed Temporally and Comparatively. The British Journal of Political Science, No. 31, 121-58 Agh, Attila. 1998. The Politics of Central Europe. London: Sage. Alt, James, Kenneth A. Shepsle. 1990. Perspectives on Positive Political Economy. Cambridge: Cambridge University Press. Alvarez, R. Michael, Jonathan Nagler. 1995. Economics, Issues and the Perot Candidacy: Voter Choice in 1992 Presidential Elections. American Journal of Political Science 39: 714-44. Aristotle. 1981. The Politics. New York: Penguin Classics. Arrow, Kenneth. 1951. Social Choice and Individual Values. New York: Wiley. Austin, Reginald and Maja Tjernstrom (eds). 2003. Funding of Political Parties and Electoral Campaign. International Institute for Democracy and Electoral Assistance (IDEA). Austin-Smith, David; Jeffrey Banks. 1988. Elections, Coalitions and Legislative Outcomes. The American Political Science Review, 82: 405-22. Bartolini, Stefano. 2000. The Political Mobilization of European Left, 1860-1980: The Class Cleavage. Cambridge: Cambridge University Press. Berelson, Bernard R., Paul F. Lazarsfeld, William N. McPhee. 1954. Voting: A Study of Opinion Formation in Presidential Campaign. Chicago: University of Chicago Press. Beyme, Klaus von. 1996. Transition to Democracy in Eastern Europe. New York: St. Martins Press. Biezen, Ingrid van, and Peter Kopecky. 2001. On the Predominance of State Money: Reassessing Party Financing in the New Democracies of Southern and Eastern Europe. Perspectives on European Politics and Society, 2(3): 401-429. Biezen, Ingrid van. 2004. Political Parties as Public Utilities. Party Politics 10: 6, 70122.

213

Black, Duncan. 1958. The Theory of Committees and Elections. Cambridge: Cambridge University Press. Budge, Ian, David Robertson, Derek Hearl (eds). 1987. Ideology, Strategy, and Party Change: Spatial Analysis of Post-War Election Programmes in 19 Democracies. Cambridge: Cambridge University Press. Budge, Ian, Ivor Crewe, and Dennis Farlie. 1976. Party Identification and Beyond. New York: Wiley. Campbell, Angus, Phillip Converse, Warren E. Miller, Donald E. Stokes. 1960. The American Voter. New York: John Wiley & Son. Castles, Francis, Peter Mair. 1984. Left-Right Political Scales: Some 'Expert' Judgments. European Journal of Political Research, No. 12, 73-88. Cox, Gary. 1999. The Empirical Content of Rational Choice Theory: A Reply to Green and Shapiro. The Journal of Theoretical Politics, 11(2): 147-69. Dahl, Robert. 1972. Polyarchy. Participation and Opposition. New Haven: Yale University Press. Dahl, Robert. 1989. Democracy and Its Critics. New Haven and London: Yale University Press. Dalton, Russell J, Martin P. Wattenberg (eds.). 2000. Parties without Partisans: Political Change in Advanced Industrial Democracies. Oxford: Oxford University Press. Dalton, Russell J. 2000. The Decline of Party Identification. In Dalton, Russell J, Martin P. Wattenberg (eds.). 2000. Parties without Partisans: Political Change in Advanced Industrial Democracies. Oxford: Oxford University Press. Dalton, Russell J. 2002 [3rd ed]. Citizen Politics: Public Opinion and Political Parties in Advanced Industrial Democracies. New York: Chatham House Publishers. Davis, Otto A., Melvin J. Hinich. 1968. On the Power and Importance of Mean Preference in a Mathematical Model of Democratic Choice. Public Choice, Vol. 5 (Fall), 59-73. de Palma, Andr, Gap-Seon Hong, Jacque-Franois Thisse. 1990. Equilibria in Multiparty Competition under Uncertainty. Social Choice and Welfare, Vol. 7, 247-259. De Swaan, Abram. 1973. Coalition Theories and Cabinet Formation: A Study of Formal Theories of Coalition Formation Applied to Nine European Parliaments after 1918. Amsterdam: Elsevier. Dodd, Lawrence C. 1976. Coalitions in Parliamentary Government. Princeton: Princeton University Press. Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper & Row. Duverger, Maurice. 1954. Political Parties, Their Organization and Activity in the Modern State. New York: Wiley. Ekiert, Grzegorz. 1991. Democratization Process in Eastern Europe: A Theoretical Reconsideration. British Journal of Political Science, Vol. 21, No. 3 (July), 285-313.

214

Elster, Jon, Claus Offe, Ulrich F. Preuss. 1998. Institutional Design in Post-Communist Societies: Rebuilding the Ship at Sea. Cambridge: Cambridge University Press. Elster, Jon. 1992. Nuts and Bolts for the Social Sciences. Cambridge: Cambridge University Press. Enelow, James, Melvin J. Hinich. 1984. Probabilistic Voting and the Importance of Centrist Ideologies in Democratic Elections. Journal of Politics, Vol. 46, No. 2 (May), 459-78. Enelow, James, Melvin J. Hinich. 1989. A General Probabilistic Theory of Elections. Public Choice, Vol. 61, No. 2 (May), 101-13. Enyedi, Zsolt. 2005. The Role of Agency in Cleavage Formation. European Journal of Political Research, Vol. 44, No. 5, 697-720. Erikson, Robert, David Romero. 1990. Candidate Equilibrium and the Behavioral Model of Vote. The American Political Science Review, 84: 1103-26. Evans, Geoffrey, Stephen Whitefield. 1993. Identifying the Bases of Party Competition in Eastern Europe. British Journal of Political Science, Vol. 23, No. 4, 521-548. Evans, Geoffrey, Stephen Whitefield. 1995. Social and Ideological Cleavage Formation in Post-Communist Hungary. Europe-Asia Studies, Vol. 47, No. 7, 1177-1204. Evans, Geoffrey, Stephen Whitefield. 1998. The Structuring of Political Cleavages in Post-Communist Societies: The Case of Czech Republic and Slovakia. Political Studies, Vol. 46, 115-139. Franklin, Charles and John E. Jackson. 1983. The Dynamics of Party Identification. The American Political Science Review 77: 4, 957-973. Franklin, Mark, Tom Mackie, and Henry Valen, eds. 1992. Electoral Change. New York: Cambridge University Press. Franklin, Mark. 1985. The Decline of Class Voting in Britain. Oxford UK: Oxford University Press. Geer, John; Mark E. Shere. 1992. "Party Competition and the Prisoner's Dilemma: An Argument for the Direct Primary. The Journal of Politics, Vol. 54, No.3, pp. 741-61. Gelman, Andrew, and Gary King. 1990. Estimating Incumbency Advantage without Bias. American Journal of Political Science 34: 4, 1142-64. Ginsburg, B. 1976. Elections and Public Policy. American Political Science Review, No. 70, 41-49. Green, Donald, Ian Shapiro. 1994. Pathologies of Rational Choice: A Critique of Applications in Political Science. New Haven and London: Yale University Press. Greenberg, Joseph; Kenneth Shepsle. 1987. The effect of Electoral Rewards in Multiparty Competition with Entry. The American Political Science Review, Vol. 81, No. 2 (June, 1987), pp. 525-38. Grofman, Bernard. 1985. The Neglected Role of Status-Quo in Models of Issue Voting. Journal of Politics 47: 230-7.

215

Grzymala-Busse, Anna. 2001. Coalition Formation and the Regime Divide in East Central Europe. Comparative Politics, 34 (October): 85-104. Hardin, Garrett. 1968. The Tragedy of the Commons. Science, No. 162, 1243-1248. Hechter, Michael. 1997. Sociological Rational Choice Theory. Annual Review of Sociology, Vol. 23: 191-214. Hibbs, D. 1979. Inflation, Unemployment and Left-Wing Political Parties: A Reanalysis, The American Political Science Review, No. 73, 185-90. Hinich, Melvin, John O. Ledyard, Peter C. Ordeshook. 1973. A Theory of Electoral Equilibrium: A Spatial Analysis Based on the Theory of Games. The Journal of Politics, Vol. 35, No. 1 (Feb), 154-193. Hinich, Melvin, Peter C. Ordeshook. 1970. Plurality Maximization vs. Vote Maximization: A Spatial Analysis with Variable Participation in The American Political Science Review, Vol. 64, No. 3 (Sept. 1970), pp. 772-91. Hotelling, Harold. 1929. Stability in Competition. The Economic Journal, Vol.39, Issue 153 (March), 41-57. Huber, John, Ronald Inglehart. 1995. Expert Interpretations of Party Space and Party Locations in 42 Democracies. Party Politics, No. 1, 73-111. Ikstens, Janis, Daniel Smilov, Marcin Walecki (2001) Party and Campaign Funding in Eastern Europe: A Study of 18 Member Countries of ACEEEO, paper presented at ACEEEO annual conference Transparent Election Campaign Financing in the 21st Century, Oct. 13-17, Brijuni, Croatia. Inglehart, Ronald. 1977. The Silent Revolution: Changing Values and Political Styles Among Western Politics. Princeton: Princeton University Press. Inglehart, Ronald. 1990. Culture Shift in Advanced Industrial Society. Princeton: Princeton University Press. Jennings, M. Kent, and Richard Niemi. 1973. The Character of Political Adolescence. Princeton N.J.: Princeton University Press. Katz, Richard S., Peter Mair, 1995. Changing Models of Party Organization and Party Democracy: The Emergence of the Cartel Party. Party Politics, Vol. 1, 5-28. Kedar, Orit. 2002. Balancing the Seesaw: Rationality and Manu Dependence in Voter Behaviour. Paper presented at the annual meeting of Midwest Political Science Association, Chicago IL, April 25-28. Kiser, Edgar and Michael Hechter. 1998. The Debate on Historical Sociology: Rational Choice Theory and Its Critics. The American Journal of Sociology, Vol. 104, No. 3 (Nov.), pp. 785-816. Kitschelt, Herbert, Zdenka Mansfeldova, Radoslaw Markowski, Gabor Toka. 1999. PostCommunist Party Systems: Competition, Representation and Inter-Party Cooperation. Cambridge: Cambridge University Press. Kitschelt, Herbert. 1994. The Transformation of European Social Democracy. Cambridge: Cambridge University Press. 216

Kitschelt, Herbert. 1995. The Formation of Party Cleavages in Post-Communist Democracies. Party Politics, Vol. 1, No. 4: 447-72. Klima, Michal. 1998. Consolidation and Stabilization of the Party System in Czech Republic. Political Studies, Vol. 46, 492-510. Klingemann, H.D, Andrea Volkens, Judith Bara, Ian Budge, and Michael McDonald. 2006. Mapping Policy Preferences II: Comparing 24 OECD and 24 CEE Countries, 19902003. Oxford: Oxford University Press. Klingemann, Hans-Dieter, Richard I. Hofferbert, Ian Budge (with Hans Keman, Torbjon Bergman, Franois Ptry and Kaare Strm). 1994. Parties, Policies and Democracy. Boulder: Westview Press. Kopecky, Petr. 1995. Developing Party Organizations in East-Central Europe: What Type of Party is Likely to Emerge? Party Politics, 1: 4, 515-534. Krmer, Jrgen, Hans Rattinger. 1997. The Proximity and the Directional Theory of Issue Voting: Comparative Results for USA and Germany. European Journal of Political Research, Volume 32, Issue 5 (Aug.), 1-29. Laver, Michael, Ian Budge (eds). 1993. Party Policy and Coalition Government in Western Europe. London: Macmillan. Laver, Michael, John Garry. 2000. Estimating Policy Positions from Political Texts. American Journal of Political Science, Vol. 44, No. 3, 619-34. Laver, Michael, Kenneth Benoit, John Garry. 2003. Extracting Policy Positions from Political Texts Using Words as Data. The American Political Science Review, Vol. 97, No. 2 (May), 311-31. Laver, Michael, W. Ben Hunt. 1993. Policy and Party Competition. New York & London: Routledge. Laver, Michael. 1997. Private Desires, Political Action: An Invitation to the Politics of Rational Choice. London: Sage. Lewis, Paul G. 2000. Political Parties in Post-Communist Eastern Europe. London and New York: Routledge. Lewis-Beck, Michael and Martin Paldam. 2000. Economic Voting: An Introduction. Electoral Studies, 19: 113-22. Lewis-Beck, Michael and Mary Stegmaier. 2000. Economic Determinants of Electoral Outcomes. Annual Review of Political Science 3: 183-219. Lewis-Beck, Michael. 1988. Economics and Elections: The Major Western Democracies. Michigan: Michigan University Press. Lin, Tze-Min, James M. Enelow, Han Dorussen. 1999. Equilibrium in Multicandidate Probabilistic Spatial Voting. Public Choice, Vol. 98, 59-82. Linek, Luk, Zdenka Mansfeldova. 2006. Efficient Representation Endangered? Czech Party System and the European Union, in Paul G. Lewis, Zdenka Mansfeldova (eds) The European Union and Party Politics in East Central Europe. London: Palgrave McMillan. 217

Linz, Juan J. 1994. Presidential or Parliamentary Democracy: Does It Make a Difference? In J. Linz and A. Valenzuela (eds) The Failure of Presidential Democracy. Baltimore: John Hopkins University Press. Linz, Juan J. 1996. The Perils of Presidentialism. In L. Diamond and M.F. Platter (eds) The Global Resurgence of Democracy. 2nd ed. Baltimore: John Hopkins University Press. Lipset, S.M., Stein Rokkan. 1967a. Party Systems and Voter Alignments. Cross-National Perspectives. New York: The Free Press. Lipset, S.M., Stein Rokkan. 1967b. Cleavage Structure, Party Systems and Voter Alignments: An Introduction in Saymour M. Lipset and Stein Rokkan (eds) Party Systems and Voter Alignments. Cross-National Perspectives. New York: The Free Press. Macdonald, Stuart Elaine, George Rabinowitz, Ola Listhaug. 1995. Political Sophistication and Model of Issue Voting. British Journal of Political Science, Vol. 25, No. 4, 453-83. Macdonald, Stuart Elaine, George Rabinowitz, Ola Listhaug. 1998. On Attempting to Rehabilitate the Proximity Model: Sometimes the Patient Just Cant Be Helped. The Journal of Politics, Vol. 60, No. 3 (Aug), 653-90. Macdonald, Stuart Elaine, George Rabinowitz, Ola Listhaug. 2001. Sophistry versus Science: On Further Efforts to Rehabilitate the Proximity Model. The Journal of Politics, Vol. 63. No. 2 (May), 482-500. Madison, James. 1987. Federalist X, in Alexander Hamilton, James Madison, John Jay Federalist Papers. London: Penguin Classics. Mainwaring, Scott, Timothy R. Scully (eds). 1995. Building Democratic Institutions: Party Systems in Latin America. Stanford: Stanford University Press. Mair, Peter, Ingrid van Biezen. 2001. Party Membership in Twenty European Democracies, 1980-2000. Party Politics, Vol. 7, No. 1, 5-21. Mair, Peter, Richard S. Katz. 1998. Party Organization, Party Democracy, and the Emergence of Cartel Party, in Peter Mair Party System Change - Approaches and Interpretations. Oxford: Oxford University Press. Mair, Peter, Wolfgang C. Muller, Fritz Plasser. 2004. Political Parties in Western Europe: Party Responses to Electoral Markets. London: Sage. Mair, Peter. 1998. Party System Change - Approaches and Interpretations. Oxford: Oxford University Press. Marks, Gary, Marco Steenbergen (eds). 2004. European Integration and Political Conflict. Cambridge: Cambridge University Press. Marschak, Jacob. 1954. Towards an Economic Theory of Organization and Information, in R.M. Thrall, C.H. Comb, and R.L. Davis. Decision Processes. New York: John Wiley & Sons, Inc., 188-189.

218

McKelvey, Richard D. 1975. Policy Related Voting and Electoral Equilibrium. Econometrica, 43: 815-44. Merrill, Samuel, and Bernard Grofman. 1999. An Unified Theory of Voting: Directional and Proximity Spatial Models. Cambridge: Cambridge University Press. Merrill, Samuel. 1995. Discriminating between the Directional and Proximity Spatial Models of Electoral Competition. Electoral Studies, Vol. 14, 273-87. Michels, Robert. 1959 [1915]. Political Parties: A Sociological Study of the Oligarchical Tendencies of Modern Democracy. New York: Dover Publications, Inc. Miller, Warren. 1991. Party Identification, Realignment, and Party Voting: Back to the Basics. The American Political Science Review, 85: 557-68. Miroiu, Adrian. 2006. Fundamentele politicii I: Preferin e i alegeri colective [The Bases of Politics: Preferences and Collective Choices]. Iai: Polirom. Miroiu, Adrian. 2007. Fundamentele politicii II: Ra ionalitate i Ac iune Colectiv [The Bases of Politics: Rationality and Collective Action]. Iai: Polirom. Muller, Wolfgang; Kaare Strom. 1999. Policy, Office or Votes. How Political Parties in Western Europe Make Hard Decisions. Cambridge: Cambridge University Press. ODonnell, Guillermo. 1995. Do Economists Know Better? Journal of Democracy, Vol. 6, No. 1 (January), 23-8. Olson, David. 1998. Party Formation and Party System Consolidation in the New Democracies of Central Europe. Political Studies, Vol. 46, 432-464. Olson, Mancur. 1971. The Logic of Collective Action: Public Goods and the Theory of Groups. Cambridge Mss: Harvard University Press. Ordeshook, Peter C. 1971. Pareto Optimality in Electoral Competition. The American Political Science Review, 65:4, 1141-45. Page, B. 1978. Choice and Ecoes in Presidential Elections: Rational Man in Electoral Democracy. Chicago: University of Chicago Press. Palfrey, R Thomas. 1984. Spatial Equilibrium with Entry. The Review of Economic Studies, Vol. 51, No. 1 (Jan), 139-156. Pedersen, Morgens N. 1979. The Dynamics of European Party Systems: Changing Patterns of Electoral Volatility. European Journal of Political Research, Vol. 7, No. 1, 1-26. Plattner, Marc F., Larry Diamond. 1995. Democracys Future. Journal of Democracy, Vol. 6, No. 1 (January), 3-6. Plott, Charles R. 1967. A Notion of Equilibrium and Its Possibility under Majority Rule. The American Economic Review, Vol. 57, No. 4, 787-806. Rabinowitz, George, Stuart MacDonald. 1989. A Directional Theory of Issue Voting. The American Political Science Review, Vol. 83, No.1, 93-121. Riker, William. 1962. The Theory of Political Coalitions. New Haven: Yale University Press. 219

Riker, William. 1980. Implication from the Disquilibrium of Majority Rule for the Study of Institutions. American Political Science Review, Vol. 74, No. 2 (June), 432-46. Riker, William. 1990. Political Science and Rational Choice. In James E. Alt, Kenneth Shepsle Perspectives on Positive Political Economy. Cambridge: Cambridge University Press, 163-81. Rose, Richard, Neil Munro, Tom Mackie. 1998. Elections in Central and Eastern Europe since 1990. University of Strathclyde Studies in Public Policy 300. Rose, Richard. 1997. Where Are Postcommunist Countries Going? Journal of Democracy, Vol. 8, No. 3 (July), 92-108. Samuelson, P. A. 1948. Consumption Theory in Terms of Revealed Preference. Economica 15: 243-253. Samuelson, Paul. (1938). A Note on the Pure Theory of Consumers' Behaviour. Economica 5: 61-71. Sartori, Giovanni. 1976. Parties and Party Systems. New York: Cambridge University Press. Schlesinger, Joseph. 1984. On the Theory of Party Organization. The Journal of Politics, Vol. 46, No. 2 (May, 1984), 369-400. Schumpeter, Joseph. 1950. Capitalism, Socialism and Democracy. New York: Harper & Row. Sen, Amartya K. 1970. Collective Choice and Social Welfare. Oxford: Oxford University Press. Smithies, A. 1941. Optimum Location in Spatial Competition. The Journal of Political Economy, Vol. 49, No. 3, 423-39. tefan, Lauren iu, Rzvan Grecu (2004) Parliamentary Elites in Post-Communist Romania. Paper presented at EURELITE regional conference, Vilnius, May, 2004. Stokes, Donald. 1963. Spatial Models of Party Competition. The American Political Science Review, Vol. 57, Issue 2 (June, 1963), pp.368-77. Strom, Kaare. 1990. A Behavioral Theory of Competitive Political Parties. American Journal of Political Science, Vol. 34, Issue 2 (May, 1990), pp. 565-98. Strom, Kaare; Wolfgang Muller. 1999. Political Parties and Hard Choices in Wolfgang Muller and Kaare Strom Policy, Office or Votes. How Political Parties in Western Europe Make Hard Decisions. Cambridge: Cambridge University Press, pp. 1-35. Szczerbiak, Aleks. 2001. Party Structure and Organizational Development in PostCommunist Poland. Journal of Communist Studies and Transition Politics 17: 2, 94130. Todosievi, Bojan. 2005. Issues and Party Preferences in Hungary: A Comparison of Directional and Proximity Models. Party Politics 11: 1, 109-26. Volkens, Andrea. 2007. Strengths and Weaknesses of Approaches to Measuring Policy Positions of Parties. Electoral Studies, 1: 26, 90-107.

220

Webb, Paul, David M. Farrell, Ian Holliday (eds). 2003. Political Parties in Advanced Industrial Democracies. Oxford: Oxford University Press. Westholm, Anders. 1997. Distance versus Direction: The Illusory Defeat of the Proximity Theory of Electoral Choice. The American Political Science Review, Vol. 91, No. 4 (Dec.), 865-83. White, Stephen, Richard Rose, and Ian McAllister. 1997. How Russia Votes. Chatham, NJ: Chatham House. Whitefield, Stephen, Milada Anna Vachudova, Marco Steenbergen, Robert Rohrschneider, Gary Marks, Paul Loveless, Liesbet Hooghe. (2006). Do Experts Surveys Produce Consistent Estimates of Party Stances on European Integration? Comparing Expert Surveys in The Difficult Case of Central and Eastern Europe, Electoral Studies, forthcoming. Whitefield, Stephen. 2002. Political Cleavages and Post-Communist Politics. Annual Review of Political Science 5: 181-200. Wittman, Donald. 1973. Parties and Utility Maximizers. The American Political Science Review, Vol. 67, No. 2, pp. 490-98. Wittman, Donald. 1983. Candidate Motivation: A Synthesis of Alternative Theories. The American Political Science Review, Vol. 77, No.1, pp.142-57.

221