Introduction

Social scientists have long been pessimistic about the influence of the basic choices in the institutional democratic design on the principle trajectory of democratisation, which was supposed to be dependent on the political cultural tradition, socio-economic factors and on the international situation. However, more recently there have been many important works supporting the notion that basic institutional choices in democratic design do matter in terms of democratic development in established democracies as well as democratising countries.

Most of the literature is concentrated on the debates between proponents and antagonists of the presidential form of government. There have been several attempts to summaries the argumentation on ‘perils’ of presidentialism: Linz (1990), Lijphart (1991), Stepan and Skach (1993), Power and Gasiorowski (1997), and Ishiyama and Velten (1998). Among the most frequently applied arguments are the following: expected instability due to the concurrent legitimisation of legislative and executive through popular election; authoritarian style of government because of concentration of power in hands of a popularly elected president; rigidity of the executive since the president is elected for a fixed term, and consequent lack of responsibility.

Simultaneously, there has been a growing notion of scepticism about anti-presidentialist argumentation. Many authors claim the relative desirability of a presidential system of government under particular circumstances, such as a heterogeneous society (e.g. Horowitz 1993). It is often mentioned that despite all the criticism, most new democratisers either initially chose a relatively strong presidency, or later switch from a parliamentary republic to a presidential one. The pressing need for prompt economic reforms, or conducive cultural traditions of a country are among the possible explanations.

Mainwaring (1993) has presented a revision of anti-presidentialist argumentation. He argues that presidential government can be effective in a two-party political environment, while the combination of presidentialism and multipartism exaggerates the problems associated with both of them.

The hypothesis of the disadvantages of presidential government for democratic consolidation has been tested in a number of empirical studies: Mainwaring (1993), Stepan and Skach (1993), Norris (1997), Power and Gasiorowski (1997), Gasiorowski and Power (1998). The results are diverse. This is, at first sight, surprising, but can be explained through the analysis of methodological assumptions and geographical preferences of these authors. Operating with a sample which includes OECD countries, the studies by Mainwaring and Norris support the hypotheses on the relative merit of parliamentary government, especially in combination with a proportional electoral system. The findings by Stepan and Skach produce almost the same suggestions, although they concentrate on the countries that have become independent after World War II. However, Gasiorowski and Power, whose sample consists of Third World democracies,2 find that neither hypothesis has much empirical support in their data. The study by Metcalf (1997) is also sceptical about anti-presidentialist argumentation.

Looking at the recent literature on the impact of institutional design on democratic consolidation, Norris (1997) has concluded: “The challenge for further research is to consider these issues in terms of the democratisation process”. The above controversies refer, largely, to the regional specifics of democratic consolidation and its specifics in the early stages of the democratisation process. Thus, the need to test these hypotheses in terms of the post-communist transformation is pressing.

This paper is an attempt to test the following hypothesis:

Basic choices in institutional design of new democracies have significant impact on the processes of democratic consolidation in the new post-communist democracies of Central and Eastern Europe (CEE).

The hypothesis is two-edged since it implies that (1) there are some institutional choices which have facilitated or hindered democratic consolidation in the region; (2) the impact of institutional choices is relatively powerful as an explanatory factor in comparison with non-institutional variables.

Geographical scope. The post-communist countries of Central and Eastern Europe present an ideal natural laboratory for empirical testing of the basic institutionalist propositions. These countries commenced democratisation at approximately the same time. They belong to a region, which has a lot of similarities but also a high degree of diversity in terms of cultural experience, economic situation, and social structure. Consequently, the countries of the region provide the researcher with data diverse in respect to the variables under study, but not so much different to be incomparable.

There is a number of case studies focusing on the political effects of particular institutional choices in the post-communist countries. As Frye (1998, p. 740) argued, “the task for scholars working on post-communist political institutions now is to move beyond rich political histories to directed studies that test and build theory”. This study includes 22 out of 27 post-communist countries of the former Soviet Union and its European allies: Albania, Armenia, Azerbaijan, Belarus, Bulgaria, Croatia, Czech Republic, Estonia, Georgia, Hungary, Kazakhstan, Latvia, Lithuania, Macedonia, Moldova, Poland, Romania, Russia, Slovakia, Slovenia, Ukraine, and Yugoslavia.3

Temporal frame. Since time is needed in order to judge upon direction and speed of the consolidation process, 1993 is the starting point for our study. At that time, most of the post-communist democracies had established some frameworks for democratic competition. 1998 was chosen as the end of the considered period in order to assess the effects of institutional choices in a medium-term perspective. During this six-year period, the countries under study have experienced general elections at least once, and the political systems demonstrated their ability to cope with the domestic and international challenges. A later date is not chosen due to the lack of available data.

Data collection and methods. The present study comprises a series of distinct but interrelated research tasks. First, we need to find appropriate operationalisations of the response, explanatory, and control variables. Simultaneously, a dedicated dataset will be developed.4 Then we shall proceed to the testing of hypothesised associations between institutional choices and the progress in democratic consolidation, including the analysis of possible joint effects. An especially challenging and methodologically difficult task will be to find empirical evidence in order to interpret the established associations in terms of causality. Finally, the relative importance of institutional choices in terms of democratisation will be tested alongside the rival non-institutional hypotheses.

Institutional Design and Democratic Consolidation: theoretical review and operationalisation of the basic terms

1.1. What is democratic consolidation?

There is a large and constantly increasing bulk of literature on various aspects of democratic consolidation. The diversity of its conceptualisations reflects the variety of the basic theoretical and methodological preferences in democratisation theory. Some authors consider democratic consolidation in terms of well-functioning of democratic institutions, others pay more attention to the behaviour of the principal political actors, while the third group draws its conclusions from the data on public attitudes and expectations. Consequently, one can propose distinguishing the following approaches in the contemporary literature on democratic consolidation: system,5 elitist, and culturalist. This classification corresponds with conceptualisation of democratic consolidation by Linz, Stepan and Gunther (1995), who distinguish structural, behavioural, and attitudinal dimensions6 , and that of Dawisha (1997).

System approach. According to a minimalist vision, democracy is a political system with free, secret, equal and general elections. However, elections in a democratic polity should serve as an effective means of policy determination. Therefore, the ‘two-turn-over test’ has been proposed in order to assess democratic consolidation. According to this criterion, democracy “may be viewed as consolidated if the party or group that takes power in the initial election at the time of the transition, loses a subsequent election, and turns over power to those election winners, and if those election winners then peacefully turn over power to winners of a latter election” (Huntington 1991, p. 266-67). The ‘two-turn-over test’ is demanding and has been criticised on these grounds. Post-war Italy or Japan, for instance, have passed this test only in the 1990s. The ‘one-turn-over test’ may be more feasible (Gasiorowski and Power, 1998).

Among more sophisticated democratic concepts, Dahl’s definition of democracy as “contestation open to participation” (Dahl, 1971, p. 5) is especially often employed for assessment of democratic consolidation. Many authors refer to Dahl’s concept of democracy as polyarchy, which imposes the following criteria on the democratic political system: it should be inclusive in terms of public participation; and exclusive in terms of representative and responsible decision-making.

However, ‘democracy’ does not refer exclusively to the structures of political participation; it also implies ‘democratic’ policy outcomes. Merkel (1999) has criticised Dahl’s approach due to the overlooked human-right dimension of democratic governance, and proposed discriminating between ‘effective’ and ‘defective’ democracies.7 A similar concept of ‘delegative democracy’ has been proposed by O’Donnell (1994).8

All these conceptualisations are based on similar assumptions that underlie the system approach and generate its strong and weak points. It is attractive in explaining the trajectory of political developments from the logic of the political system itself. The researcher can refer to a particular democratic theory and compare the current situation with the ideal picture.9 The research findings can be well-grounded in a broad theoretical context of the literature on democratic theory, and the researcher can formulate conceptions “in terms of systemic properties that facilitate comparative analysis” (Wheatherford 1992, p. 150). The data collection does not require time consuming and expensive methods. However, the strengths of the system approach can become its weak points, i.e. discretionary reference to a particular model of democracy and disregarding the role of latent social processes.

Elitist approach. Since the path-breaking work by Rustow (1970), there has been a growing number of attempts to explore democratisation in terms of the interplay of major actors and social groups on a national political stage. The elitist literature turns to the concept of democratic consolidation in order to account for the democratic achievement resulting from the elite settlement: “a consolidated democracy is a regime that meets all the procedural criteria of democracy and also in which all politically significant groups accept established political institutions and adhere to democratic rules of the game” (Higley and Gunther 1992, p. 3).10

The elitist approach has a significant (if not dominating) share in the literature on democratisation. Because the behaviour and attitudes of political elite derive from a unique situation, most of the studies employ the game theory paradigm and have a rather narrow geographical scope. The comparability of the data declines with increasing number of cases.

Culturalist approach. Evaluation of democratic consolidation can be also grounded in the data on dispersion of democratic attitudes and values among the population. Survey data has become the major source of empirical information. According to Weatherford (1992, p. 150): “The signal potential of survey methodology is its promise to illuminate the nexus between institutional context and individual psychology”.

The literature explores the following aspects of democratic attitudes:

The culturalist approach is also connected with some methodological obstacles. A number of possible interferences make the researcher cautious about implementation of the survey results as indicators of democratic consolidation.12 The crucial weakness of the survey approach is that the study begins with available data (in the case of secondary data analysis) or applicable methods of data collection, and then attempts to conform the theory to the given research opportunities.

Combinative concepts. Pridham (1994) suggests another difference between elitist and culturalist concepts of democratic consolidation. If the former is based on minimalist demands that the relevant political actors do not violate democratic rules of play because they do not see any plausible alternative (‘negative consolidation’); the latter concepts implies also acceptance of democratic values, normative legitimisation of the new political system (‘positive consolidation’). Merkel (1998) refers to this idea while elaborating a multidimensional dynamic model of democratic consolidation. He distinguishes four dimensions: (1) constitutional, (2) consolidation representative consolidation, (3) democratic behavioural consolidation of veto powers, (4) consolidation of civic culture. The principle direction of the democratic consolidation is from the former to the latter, but there is also reverse interaction of the distinguished dimensions.13 This type of combinative concepts can be criticised basically for the over-abundance of indicators if the purpose is construction of a quantitative measurement, while it can fit well for a comparative overview of democratisation processes in a set of case-studies.

Choosing appropriate operationalisation. Semantically, democratic consolidation can be understood as a process, as well as a particular outcome of this process (cf. Schedler 1997). In the first case, one can speak about ‘stages’ of democratic consolidation; in the second, about its ‘level’. As a process, democratic consolidation can be defined as a series of stages on the way from an authoritarian (non-democratic) regime toward a ‘well-established’ (i.e. consolidated) democracy.14 However, important questions would remain open: Why does a democratiser proceed from one stage in the consolidation process to another? Why should one expect that a political actor willing to change the rules of the game will not become powerful in the near future?

In short, we need to assess the embeddedness of the democratic ‘rules of game’ in a society, and current trends in democratic development. The first step is to evaluate the probability that anti-democratic political actors can receive some support from the public. As Offe (1997) argues, “the institutional reorganisation of postcommunist societies cannot...be regarded as successfully completed until this subsequent rooting of the new regime in the values and loyalties of the population has been achieved”. However, sustainability is only one aspect of democratic consolidation.

Another question is what kind of democracy is supported by the population. If it is a democracy of ‘Soviet type’ or ‘Fuehrerdemokratie’, one can speak about authoritarian rather than democratic consolidation. The political culture approach is effective but not self-sufficient. One should either control what is exactly meant by ‘democracy’, or use an independent indicator of democratic performance in the country. In both these cases, the researcher will aim at assessment of how ‘democratic’ is either the respondents’ idea of democracy, or the country of a respondent. Thus, the culturalist assessment should be complemented by a system evaluation.

Operationalisation of the response variable. We will employ a combination of political culture and system performance indicators in order to assess the level of democratic consolidation in the countries under study. Political culture data should indicate embeddedness of a new regime and options opened for its future development by the public expectations, while data on system performance should account for how democratic this regime is.

We should find rather sensitive indicators of democratic consolidation, since our goal is to examine the relationship between institutional choices and democratic consolidation in a rather short period of time, 1993-98. Moreover, our operationalisation should account for the progress in democratic consolidation rather than the level achieved.

As mentioned above, there are three groups of candidates for political culture indicators of democratic consolidation: (1) satisfaction with democracy; (2) normative attitudes toward democracy (its general support or rejection); (3) acceptance of democratic values. Theoretical literature suggests the following relationship between these types of political attitudes in the case of post-communist transformation: the durable satisfaction with the work of democratic institutional increases normative support for them, while involvement of the public in the networks of democratic institutions can contribute to diffusion of democratic values among the population (cf. Offe 1997).

Consequently, the data on democratic satisfaction can be interpreted as a cultural trend toward democratic consolidation, but the researcher has to control as to what is meant in responses to the questions about democratic satisfaction. As a non-culturalist assessment of democratic performance, we shall employ the Annual Survey of Freedom Country Scores by Freedom House,15 which are the most frequently used indicators of institutional democratic performance accounting for political freedoms, but also human rights preservation in a country.

To summarise: Our proposal on the index estimating the cultural trend toward democratic consolidation is to denominate the percentage of the respondents answering that they are satisfied or rather satisfied with how democracy works in their country16 by the assessment of democratic performance based on the Freedom House data. This index can be used for estimation as to which extent current political situation contributes to democratic consolidation in a given country.

1.2. Basic Choices in Institutional Design of a New Democracy.

Challenge for institutionalist debates. Since the Portuguese revolution of 1974, which is dated by Huntington (1991) as the starting point of the ‘third wave’ of democratisation, constitution making has become a ‘growth’ industry (Shugart 1998). This challenged social scientists to assess the impact of the basic institutional choices made in new democracies. The first works by ‘neo-institutionalists’ were based on a political economy approach17 or renewed attention to the classics of political thought: Weber, Michels, Schumpeter. Linz gave a new impulse to the debates about possible outcomes of the basic institutional choices through an influential essay published in 1984, and in subsequent works.

New impetus for the assessment of consequences of institutional choices has been given by the post-communist democratisation that provided social scientists with a large volume of empirical data for reconsideration of the institutionalist hypotheses. The intense debates have mainly focused on: (1) relative desirability of presidential and parliamentary systems, and (2) political and social consequences of the choices constructing an electoral system.18

Developing anti-presidential arguments. The increasing bulk of literature on the disadvantages of presidential government has been developed on the basis of two groups of arguments (cf. Shugart and Carey 1992). The first group focuses on the alleged ineffectiveness of presidential governments in terms of policy making, and possible conflicts between the executive and legislative, which are expected because:

The second group is primarily concerned with alleged dangers of a strong presidency for a new democratiser. Presidentialism is accused of being conducive for authoritarian tendencies due to:

Empirically, the notion of scepticism of the ability of presidential systems to attain democratic consolidation was supported by Stepan and Skach (1993). The authors focused on the countries which achieved independence after World War II. They found that none of those countries which initially chose a presidential system was continuously democratic between 1980 and 1989, while 15 out of 41 parliamentary democracies were so. Further empirical studies by Mainwaring (1993) and Norris (1997) supported the thesis on conduciveness of the parliamentary system for democratic consolidation. Hellman (1997) claimed that presidential governments are also less apt to urgent reforms in transitional societies.

Scepticism about anti-presidentialist arguments. Even this brief outline of the anti-presidentialist arguments indicates a logical inconsistency in the argumentation: Presidential governments are accused in possible policy deadlocks due to a stricter division of powers and, simultaneously, implicated in the affinity with the renewed authoritarism because of power concentration. A similar criticism can be directed toward parliamentary regimes since they lead either to long-term procedures of identifying the political agenda in the case of an uncertain parliamentary majority, or to power concentration in hands of the head of executive (Prime Minister), who is, at the same time, leader of a parliamentary majority. The power concentration and stalemates in policy determination are more likely to derive from the current party composition of governmental bodies than from constitutional provisions on legislative-executive relations (cf. Cheibub 1999).

A series of empirical studies has spread the notion of scepticism about anti-presidentialist argumentation. Horowitz (1993) observes that the Westminster model, which combines plural representation in parliament and a parliamentary system of government, can lead to disintegration of a divided society, while presidentialism might play an integrative function in such a case. The study by Metcalf (1997) draws attention to the experience of the Russian successor states after World War II. According to his report, three states which have chosen the parliamentary form of government (Estonia, Latvia, Lithuania), were not successful in democratic consolidation, while the only semi-presidential regime (Finland) has secured democracy. Gasiorowski and Power (1998) found that correlation between the choice of a form of government and successful democratisation (operationalised by the one-turn-over test) is very weak.

Refocusing on combinations of institutional choices. Mainwaring (1993) tried to overcome the impasse of the debates on the merits of presidentialism, by establishing a link between these debates and the discussion about the effects of the choice between proportional and majoritarian representation. He claimed that it is not the choice of a form of government or an electoral system, but rather their particular combinations, produce desirable or unwelcome effects.

Mainwaring refers to Duverger’s ‘threefold law’,20 and argues that party fragmentation in the case of proportional representation does not impede normal functioning of democratic institutions in a parliamentary system, since the parties have to engage in deliberations in order to produce a government coalition. Mainwaring sees the danger in the combination of proportional representation with the presidential form of government. In this case, parties do not need to establish stable coalitions, and a fragmented and ideologically polarised parliament tends to confront the presidential government. The situation may result in recourse to undemocratic means of conflict resolution.

However, theory and empirical evidence are not so conclusive. First, Duverger’s ‘threefold law’ has been questioned in a series of studies.21 Second, empirical testing of Mainwaring’s hypothesis produced similarly ambiguous results to the testing of anti-presidentialist arguments. Mainwaring (1993) and Norris (1997) found strong empirical support for the relative merit of parliamentary regimes, combined with proportional representation in terms of consolidating democracy. Studies by Gasiorowski and Power (1997, 1998) were not so conclusive. Recently, Cheibub (1999, p. 12) found in his study, which focused on a large sample of presidential regimes, that “the difference in transition probabilities between plurality and proportional systems [in the case of presidential governments] is...negligible”. The author also questioned the crucial assumption, which underlies Mainwaring’s thesis, that a divided government is conducive for breakdown of presidential regimes.

Refocusing of the debates on the link between society and politics. A possible alternative in the assessment of institutional choices is to re-consider their effects in terms of the link between society and politics they create, i.e. their contribution to the developing of representative mechanisms, civil society, and inclusion of different social groups and the public in the political process.

Parliamentary government and proportional representation are expected to facilitate development of a party system as the main link between society and politics; while introduction of presidential government and majoritarian representation may offer better options for a personalised politics with direct appeal to the people. The following consequences of institutional choices in terms of the social aspect of democratic consolidation and democratic institutionalisation are expected in the theory:

  1. parliamentary government and proportional representation are more conducive for development of influential political parties (see: Shugart and Haggart 1997);
  2. the parliamentary regime has “greater tendency to provide long party-government careers, which add loyalty and experience to political society” (Stepan and Skach 1993, p. 22), while presidential governments and majoritarian electoral systems give more opportunities for ‘outsiders’ to get in;22
  3. presidential systems and majoritarian representation create favourable conditions for personalisation of politics and demagoguery (Linz 1990);
  4. parliamentary governments have greater ability to respond to the impulses of the society through the representative mechanisms, while presidential governments undergo imperative guidance only during electoral campaigns (Hellman 1997).

However, Kiss (1996, p. 75) has questioned the thesis on the desirability of a strong parliament and influential disciplined political parties in terms of democratic institutionalisation in post-communist countries. Emphasis on party politics “may foster widespread alienation from political class which many have identified as a growing threat to democratic legitimacy in East and Central Europe”. Shugart and Haggart (1997) also support the argument that a combination of proportional representation with parliamentary government can dilute the link between the political class and the electorate.23

The debates on the effects of institutional choices in terms of the shaping the link between society and politics are challenging for a study that will empirically test the existence and relevance of the alleged association between parliamentary government and proportional electoral system, on the one hand, and strength of the party system, on the other.24

To summarise: We have distinguished between arguments in favour of or against certain institutional choices (in particular, those of a governmental system and electoral system) which examine their effects in terms of decision-making, and the arguments which refer to the shaping of the link between society and politics. While the second group of arguments looks more promising from the viewpoint of its explanatory ability, this paper cannot support the thesis of the existence and relevance of the link between governmental and electoral systems and democratic consolidation.

However, all versions of institutionalist argumentation have much in common. Working from this approach, one can expect that (1) the choice between presidentialism and parliamentarianism has significant effects in terms of democratic consolidation; (2) the basic choice in constructing an electoral system is the choice between proportional and majoritarian representation, while the former may either (2a) produce very negative effects in terms of democratic consolidation in combination with the presidential system of government, or (2b) exaggerate the effects of parliamentarianism in terms of framing the link between society and politics in the course of the democratisation process.

Consequently, we need operationalisations of our explanatory variables which will account for:

  1. relative autonomy of presidents (elected heads of state) from the parliament, and their authority in terms of policy determination;
  2. percentage of parliamentary seats distributed through proportional representation in competition among political parties.

Operationalisations

Presidentialism can be operationalised either as a classification of systems of government or as an interval measurement of the relative presidential strength. Most of the existing classifications of governmental systems are based on the evaluated abilities of presidents and parliaments to influence government formation. In Shugart and Carey’s classification (1992), the cabinet is appointed by the president (with possible legislative censure) and the president has the exclusive right to dismiss ministers (no vote of no confidence) in presidential systems; the president appoints the cabinet while either the president or parliament can dismiss ministers in president-parliamentary systems; ministers are appointed by the president (possibly, after parliamentary proposal or censure) and can be dismissed only by parliament in premier-presidential systems; and the president has very limited influence on the formation of the executive in parliamentary systems. This classification can be applied to the region under study as follows.25

However, there is no theoretical argument which is fully convincing that the legislative authority of presidents or their strength in relation to the judiciary are less significant than their influence on cabinet formation.

A reliable interval measurement is a challenging task. In particular, there is no evidence that presidential powers have unidimensional structure. Thus, we need an empirical insight into the given variety of presidential authorities before an appropriate operationalisation can be proposed.

One can distinguish two groups of provisions related to presidential power: (1) provisions contributing to the relative independence of presidents (elected heads of states) from the parliament; (2) authority of a president in terms of decision-making. Within the latter group, following the tradition of constitutional analysis, we distinguish between (2a) ability of a president to influence formation of a government; (2b) legislative authority of a president; (2c) his authority in the relations with the judicial branch of government; (2d) special presidential authority in relations with regional (local) governments, in the fields of foreign policy, military defence, and in cases of emergency (cf. Kutlesic). Empirical analysis26 indicates that the strength of presidents differs in respect to a set of constitutional provisions.27 We do not assume that the powers of a president has the above structure. Our argument is that the relative strength of a president depends on the variables included in these four categories. This assumption enables us to explore the structure of the presidential powers by means of factor analysis.

A relatively simple bifactor solution, explaining 49.4% of variance, has been found under the condition of minimal eigenvalue equalled to 3 by oblique rotation. This is a solution with two almost orthogonal factors (r = 0.06).

The first factor is most strongly associated with the abilities of a president to:

  1. influence formation of the cabinet: right of mandatory nomination of the Prime Minister (r = 0.79), rights to appoint Prime Minister without a vote of confidence in the parliament (r = 0.72), fire the Prime Minister (r = 0.75), appoint other ministers (r = 0.83);
  2. directly participate in the policy determination: take part in sessions of the government (r = 0.74), veto decisions of government (r = 0.88), issue decrees on the basis of existing legislation (r = 0.77), in the case of emergency (r = 0.54), or legislative degrees (r = 0.58), declare mobilisation (r = 0.80), and conclude international accords (r = 0.82);
  3. appoint regional executives (r = 0.78), army officers (r = 0.74), representatives of the state (0.83).

The president’s being popularly elected (r = 0.53), majority in parliament needed for his impeachment (r = 0.72), or in order to rid his veto on legislation (r = 0.66), and ability of the parliament to dismiss ministers by a vote of non-confidence (r = -0.69) are also strongly correlated with this factor, while the latter correlation is, of course, negative. Thus, this factor may be interpreted as the ability of the president to directly influence political agenda.

The second factor is correlated with the authorities of a president to:

  1. appoint the judges of the constitutional court (r = 0.64), supreme court (r = 0.87), and other courts (r = 0.63);
  2. dissolve the parliament under particular circumstances: if the latter lacks a responsible majority (r = 0.75) or support for the governmental actions (r = 0.63)28 .

The possible interpretation of this factor is the authority of the president to serve as arbiter between the legislative and executive and guarantee the independence of the judicative branch. Since the authority highly correlated with the first factor is usually associated with a ‘strong’ chief executive, while those correlated with the second factor refer to the concept of the head of state as an arbiter, we shall use in the text below the following appellations of these factors: ‘executive power’ and ‘arbiter authorities’, respectively. Table 1.2.2 indicates the scores the cases under study received on both factors.

All the cases can be divided into four groups along these lines: (1) presidents with strong executive powers and arbiter authority; (2) presidents with strong executive powers and relatively weak arbiter authority; (3) presidents with relatively weak executive powers and strong arbiter authority; (4) presidents with relatively weak executive powers and arbiter authority.

In the first group of cases, the president controls the governmental policy and serves as actual head of the executive. Simultaneously, he is assigned to play the role of the arbiter between the legislative and executive, i.e. to be judge in his own cases. This group can be labelled as super-presidential political systems. The second group includes what is usually called presidential or president-parliamentary republics with significant presidential authorities as the acting head of the executive. The third group is closer to the concept of semi-presidential republic as it was developed in the French Fifth Republic, i.e. president as arbiter with Prime Minister as the chief executive. The fourth group of cases represents a relatively pure parliamentary regime (see: table 1.2.3).

In the following, indices of presidential executive power and arbiter authority are weighed as the scores of factor analysis.29 We also employ classifications of regimes based on the presidential ability to influence government formation and the above factor scores.

Proportional representation in the parliament. The following table (1.2.4) represents the proportion of the seats in the Lower Houses of Central and Eastern European parliaments filled through party lists according to any form of proportional representation.30

To summarise: On the basis of the literature review, the principle institutional choices made in respect to presidency and electoral system are distinguished as the institutional choices which should be operationalised. Thus, the variables accounting for: executive power of presidents, presidential arbiter authority, and percentage of parliamentary seats distributed through proportional or proportional-like representation are proposed for operationalisation of the basic choices in democratic institutional design.

1.3. Scepticism about institutionalist explanations.

The attractiveness of institutional explanations is related to the vision that they can provide the policy-maker with more straightforward tools than the instruments that can be suggested by theories that connect democratisation success with non-institutional variables. One can hardly produce rapid changes in political culture or approach a drastic change in international situation, but one can propose and even succeed in radical change of political institutions.

A crucial, while not indisputable, assumption underlies this vision: the existence of an independent, enlightened and powerful policy-maker that is relatively independent of the ‘real world’ of particular interests, political culture, external influences, and institutional inertia. However, in this ‘real world’, people making policy (including institutional choices) are a part of the society, and their behaviour is influenced by a set of domestic and international factors. Consequently, the hypothesis of the crucial impact of institutional choices on success or failure in democratic consolidation has to be tested alongside the rival hypothesis that institutional choices as well as the progress in democratic consolidation are the effects of other internal and external factors in a given society.

Culturalist explanations. The pioneering work by Almond and Verba (1963) set the agenda for a culturalist approach in democratic studies. Working with data on five industrialised democracies, these authors found basic structures of democracy determined by the ‘civic culture’ of the country. In terms of democratisation, their theory was supported by Lipset (1994). He confronted the institutionalist explanations of democratic consolidation and emphasised the impact of political culture and, in particular, that of the colonial legacy of Catholicism and Britain, on the democratisation outcomes in the Third World.

In this study, traditional religious affiliation is controlled as an extraneous factor, operationalised as four variables with information on percentage of Catholic, Protestant, Orthodox and Muslim people among the religious percentage of the population.31 The Central and Eastern European Barometer 1992 is employed as the primary data source.

There are three interrelated hypotheses on the impact of the historical past on the prospects for democratic consolidation after the collapse of communism: (1) states based on strong national identities and having experience of independence before the communist rule are more likely to become consolidated in a peaceful way; (2) the countries with pre-communist authoritarian and nationalistic traditions have to face more obstacles in democratisation; while (3) the countries with previous democratic experience are more likely to be consolidated as democracies (see: Dawisha 1997). For the purpose of this study, variables assessing previous democratic experience will be analysed. The primary data are obtained from the Polity Data Archive.32 The positive indicators on the ‘institutionalised democracy’ will be counted in points of democratic experience.

Henderson and Robinson (1997) propose a classification of new post-communist democracies in terms of their historic and religious heritage. Their approach has been modified by Wallance and Haerpfer (1998) who also distinguish three regions with different paths of post-communist transformation. Wallance and Haerpfer provide the reader with a lot of evidence (primarily based on macroeconomic data and survey data from the New Democracies Barometer) that the transformation goes smoother in the West, with the biggest obstacles in the East, while the South path is in the middle, transformation there being hindered by the nation/state building and nationalistic populism. Constructing our ‘region’ variable, we have generally followed the approach of Wallance and Haerpfer.

Socio-economic explanations. This approach has its roots in Marxist theory, which indicates capitalist development as a pre-requisite for ‘bourgeois’ (liberal) democracy and small bourgeoisie (middle class) as its social ground. It was further developed by Lipset (1959). His research suggests that high level of average income, high degree of industrialisation, urbanisation and educational attainments would facilitate democratic consolidation. An empirical study by Gasiorowski and Power (1998) supports this theory and suggests the average level of educational attainments as a more powerful explanatory factor than average income. The indicators included in the present study as controlled variables are GDP per capita and GDP growth. The data are obtained from the OECD data base.33

Demos/ethos explanations. Another dimension of structural determinants of democratic consolidation refers to problems of ‘stateness’ in new democracies and impact of ethnic heterogeneity on the paths of democratic consolidation. Revealing this problem in regard to post-communist Central and Eastern Europe, Linz and Stepan (1996a) indicate that problems of ‘stateness’ (existing ‘demos’ as a pre-requisite of establishing democracy) as well as peripheral ethnic conflicts significantly hinder democratisation.

Percentage of ethnic minorities but also relative size of the second largest ethnic group are often employed as indices of ethnic homogeneity. Both these approaches have their advantages. The first concentrates on the supposed weakness of the titular nation while the second assesses the strength of its most likely rival. Our proposal is to combine these measures and employ the ratio of the size of the largest minority to the size of the titular ethnic group.34

Explanations referring to external factors. There is also an increasing literature on international factors in democratisation. A more pessimistic viewpoint is presented by the theories of global order (and dependence theories, in particular). Their authors claim that the countries on the periphery of the world order are less likely to consolidate as democracies than countries in the semi-periphery; while the latter have to face more obstacles in democratic consolidation than the ‘old’ industrialised countries, which build the core of the world system (Bollen, 1983). The literature on the international regimes of democratisation presents a more optimistic point of view. These authors argue that there are growing networks and expanding systems of negative and positive sanctions established by the Western countries in order to promote democratisation. According to Gasiorowski and Power (1998), the likelihood of successful democratic consolidation is significantly higher in a democratic regional environment. External threats and direct involvement in international conflicts are other explanatory factors suggested for failure in democratisation. For the purpose of the present study, the value of international financial aid and a dummy variable with ‘1’ assigned to the countries with borders with the EU are employed as controlled variables for international factors in democratisation.

It is hardly feasible to establish an inclusive list of all possible alternative explanations. However, the above list contains the variables which allow us to control the effects suggested by the most likely rival hypotheses. All the above factors refer either to structural aspects of the societies under study, or to the external influence on the ongoing processes. Consequently, these two groups of controlled variables are distinguished in the following analysis. At this stage of our research the necessary operationalisations of the explanatory, response and controlled variables are offered, and we are able to make the next step toward the empirical data analysis, testing the bivariate associations.

Ii.

Impact of Institutional Choices on Democratic Consolidation

2.1. Effects of the Basic Institutional Choices in Terms of Democratic Consolidation.

As the first step in our analysis, the associations between the choices, and response variable (the index of the progress in democratic consolidation) and its constituents (Human rights scores, and satisfaction with democracy) have been assessed.

Presidentialism. As is shown in table 2.1.1, there are strong linear correlations between executive power of presidents and all three indicators of democratic consolidation. The correlations are significant at the 0.01 probability level.35 As has been predicted by anti-presidentialist arguments, executive power of presidents negatively correlates with advances in democratisation.36 The correlations with arbiter authority of presidents are weaker, but we found a positive correlation between the Index and arbiter authority significant at the 0.05 probability level, i.e. arbiter authority of presidents corresponds with better progress in democratic consolidation.

Comparison of mean score values of the indicators makes it even more apparent that presidential and president-parliamentary regimes cannot be associated with movement toward democratic consolidation in the region (Table 2.1.2). At the same time, the record of semi-presidential regimes appears to be outstanding even in comparison with parliamentary ones. The mean difference is significant at the 0.001 probability level.

Thus, empirical evidence contradicts the hypothesis that mixed forms of government have negative results in terms of democratic consolidation due to alleged increase of the tension between the head of state and parliament. The arbiter authority of presidents may be relevant for human rights preservation and democratic institutionalisation (see: respective correlation with and mean score of human rights index).37

The findings on associations of the indicators of democratic progress with the type of government formation support the tentative conclusions discussed above. Presidential and president-parliamentary governments have a less encouraging record in democratic consolidation than premier-presidential and parliamentary ones (Table 2.1.3). The difference of means of the index between countries with presidential and president-parliamentary governments, on the one hand, and the countries where presidents do not have the ability to dismiss cabinet members, on the other, is significant at the 0.001 probability level. These findings support Shugart and Carey’s hypothesis (1992) that among two possible mixed types of government formation, a president-parliamentary government distorts stable democratic development, while premier-presidential republics may be a better combination. Even compared to parliamentary regimes, the premier-presidential governments are associated with higher level of democratic satisfaction, while the countries with governments formed by the parliament have a better record on human rights preservation.38

Proportional representation. Proportional representation has strong linear correlations with all three indicators of progress in democratic consolidation. It is especially strongly correlated with preservation of human rights. The findings support the hypothesis on conduciveness of proportional representation for institutionalisation of a new democracy. One may hypothesise its contribution to the development of a party system and associated institutions of civil society. However, this link needs to be assessed through further investigation.

The difference between the countries with mixed and proportional representation is greater on the preservation of human rights than democratic satisfaction (Table 2.1.5). Moreover, countries with mixed electoral type have a slightly better record on democratic satisfaction. These results do not confirm the hypothesis on the perils of mixed parliamentary representation, which Golosov (1997) has proposed on the basis of Russian experience.

All the above findings fall squarely within the tradition of preference for parliamentary government and proportional representation. They conform the theoretical assumptions that both these institutional choices are conducive for democratic development, and correspond with the outcomes of some other comparative studies on the topic: Stepan and Skach (1993), Mainwaring (1993), and Norris (1997).

Joint effects of institutional variables. Effects of different combinations of institutional choices have become the main issue of debate in recent years. The proposed hypotheses are, thus, especially challenging for empirical testing, while the size of our set of cases puts some limitations on this approach.

39

The group of democratic designs with the best records in terms of democratic consolidation consists of combinations of semi-presidential and parliamentary regimes with proportional and mixed electoral systems. Combinations of proportional representation with semi-presidential and parliamentary regimes have similarly good results in respect to human rights preservation, but the former combination is associated with a higher level of democratic satisfaction (mean difference is 11.7%, which is significant at the 0.01 probability level).

Combination of parliamentary regime with mixed electoral system is associated with better democratic satisfaction and worse record on human rights score than parliamentary regimes combined with proportional electoral system. Combinations of semi-presidential and parliamentary regimes with proportional or mixed representation is associated with higher values on the scale of the progress in democratic consolidation (Table 2.1.7). Within this group, mixed electoral systems correspond with a higher level of democratic satisfaction, while semi-presidential regimes are associated with better preservation of human rights.40

Despite the very low scores related to the Russian combination of presidential government with the mixed electoral system, mixed electoral systems are, on average, associated with better outcomes than majoritarian one within the countries with presidential governments. This does not fit with Golosov’s (1997) hypothesis. The proportional component in combination with presidential and president-parliamentary types of government formation corresponds with a better record on human rights preservation. A possible explanation refers to the alleged effect of proportional representation in terms of party system strength. Strong political parties may serve as a balance to presidential predominance.

The record of combination of mixed and proportional electoral systems with premier-presidential and parliamentary types of government formation is outstanding. The democratisation model of an influential parliament with proportional representation is eventually called the ‘party state’. This model emphasises the importance of political parties as a medium between society and politics. Is it still enviable for new democratisers?

Multivariate model. Another way to assess the relative effects of the explanatory variables, but also their joint explanatory power, is to introduce them in multivariate analysis. Executive power of presidents, their arbiter authority, and proportional representation are subsequently introduced in a multivariate model explaining the variation in the index of progress in democratic consolidation (Table 2.1.10).

This model explains 40.8% of variance. Executive power of presidents appears as the most powerful explanatory variable; the explanatory power of presidential arbiter authority is weaker, however, being significant at the 0.05 probability level. The correlation with presidential executive power is negative, while the hypothesis of conduciveness of presidential authority as an arbiter for democratic consolidation has found new empirical support. Previously hypothesised as relatively important, the effect of proportional representation appears not to be independent but rather due to the association of proportional representation with certain political regimes.

Robustness over time. Implementing Pearson linear correlation tests for analysis of the pooled time series data, we assume the following model of the correlation between the response variable (Y) and explanatory variable (X):

where n is number of cross-section (country), and t is a time point (see: Sayrs 1989), i.e. robustness of the data over time and space. One way to test this assumption is to run the linear correlation test separately for time and spatial divisions of the data set.

Correlations between executive power of presidents and factor score are consistent over time. However, one can observe an increasing strength of correlation between index of presidential executive powers and human rights score. This finding supports the thesis on negative impact of presidential executive strength on successful democratic institutionalisation.

The arbiter authority of presidents correlates with the indicators of democratic consolidation less consistently over time (Table 2.1.12). A possible explanation is that presidential arbiter authority is more relevant in the early stages of democratisation; while later, development of democratic institutions and, in particular, political parties and parliamentary procedures may substitute the importance of this arbitrage.

Correlations of proportional representation with the index and human rights scores are relatively consistent over time. This supports the previously made observations on the correspondence of proportional representation with the progress in democratic consolidation.

Robustness over space (regions). The considerable differences between the associations of institutional choices and indicators of democratic consolidation in the three sub-regions (see: table 2.1.14) reveal the relevance of regional specifics. Proportional representation has the strongest associations in the Eastern part of the region. A possible explanation is the relevance of strong political parties as counterweights against authoritarian tendencies in presidential regimes of the Eastern part of the region.

Direction of dependency (time lags). Statistical data analysis is more apt for testing associations, while providing empirical support for interpretation of the established links in terms of causality is a difficult task. However, there is a temporal variable in our data set, which may help us test, if not causality, then time sequences in the change of the variables under study. The technique of time lags is employed for this purpose.

Table 2.1.15 presents correlations of institutional choices and indicators of democratic consolidation of presidents with time lags of two years. All conclusions we can draw from this table are tentative. However, one may find support for the thesis that arbiter authority of presidents leads to positive effects in terms of democratic consolidation (since it is better correlated with the progress in democratic consolidation) after two years, than with the progress two years before or contemporarily. In the case of the correlation of executive power of presidents and democratic satisfaction, the established time sequence is the opposite. Presidential executive strength rather follow democratic satisfaction than leads to it. The latter may give impetus to re-assess the findings of Frye (1999) who claimed a link between party fragmentation and increasing presidential powers, in order to establish the structure of associations between these three factors: democratic satisfaction of the public, party system properties, and presidential executive strength.

Summarising these findings: Countries with semi-presidential or parliamentary regimes, premier-presidential or parliamentary formation of government, and mixed or proportional electoral systems have a better record in democratic consolidation than the others in our set of cases. Within this group, semi-presidentialism and parliamentary types of government formation are associated with better preservation of human rights, while mixed parliamentary representation corresponds with a higher level of democratic satisfaction.

The above associations with executive power of presidents and proportional representation are relatively consistent over time, while the arbiter authority of presidents is more relevant in the early stages. Significant differences are established on the interregional scale. Analysis based on the time lag technique suggests tentative conclusions that arbiter authority leads to better results in democratic consolidation, while executive power of presidents rather follows the level of democratic satisfaction with a negative correlation. A reciprocal interaction of institutional choices and indicators of democratic consolidation may be a probable inference.

2.2. Relative importance of institutional choices and problem of path dependency.

Association of democratic consolidation with controlled variables. In order to assess the possible interference of the controlled variables in the above estimated associations between institutional choices and the progress in democratic consolidation, a series of tests of linear correlation has been made. As one can infer from Table 2.2.1, structural factors of GDP per capita, existing borders with the EU and previous democratic experience may be the most powerful explanations among the newly introduced variables. The level of ethnic heterogeneity and GDP change have weaker correlations, significant at 0.05 probability level. International aid per capita and size of the population do not significantly correlate with indicators of democratic consolidation.

Religious heritage seems to be a very powerful explanatory factor, but one may have certain doubts as to whether these correlations are not due to the association of distribution by religion and the sub-regions. In order to control homostadescidity of the correlations over space, the linear regression analysis has been made for the data set split by sub-region (Table 2.2.2).

Aid per capita and size of population are insignificant within the sub-regions as well as for the whole set of cases. GDP change is a factor whose significance in the whole set of cases may be related to the association with regional effects on democratic consolidation. Indicators of religious heritage also lose their significance within Western and Southern sub-regions (except the correlation of percentage of Orthodox people with democratic satisfaction in the South).41 Thus, all these variables are excluded from the following development of a multivariate explanatory model.42

Multivariate models. The institutional choices, whose associations with the progress in democratic consolidation are found in the previous analysis, are introduced in the multivariate model for the Index of the progress in democratic consolidation (see: Table 2.2.3), as well as the controlled variables of the existence of borders with the European Union, previous democratic experience, ethnic heterogeneity, and GDP per capita. The developed multivariate model is rather strong from the point of view of its explanatory power (57.1% of explained variance).

Previous democratic experience is the most powerful explanatory factor, with a positive impact on the progress in democratic consolidation. This finding supports the hypothesis on the relevance of previous experience with democracy for the success of the following democratisation attempts, and confirms the path dependency explanations. Parts of the region where democracy and rule of law already have some tradition are more likely to succeed in democratic consolidation.

However, institutional variables related to presidential powers also appear important. Presidential executive power strongly corresponds with the advances of democratic consolidation, as does arbiter authority, but the directions of the correlations are different. The stronger executive power of the president is associated with less progress in democratic consolidation, while arbiter authority corresponds with onward movements in this process. This supports our previous findings.

The same explanatory variables are introduced, while developing a multivariate regression model for Human Rights score (see: table 2.2.4). This model is very strong, it accounts for 78.8% of variance. Two institutional variables (executive power of presidents and proportional representation) appear as the most powerful explanatory factors in the model. This may support the previous claim that parliamentarianism and proportional representation are conducive for preservation of human rights and institutionalisation of a new democracy.43

The presented multivariate model for democratic satisfaction is the least satisfactory in comparison with the model explaining the Index and Human Rights score (Table 2.2.5). Nevertheless, it accounts for 15% of variance, which is significant at the 0.05 probability level. The existence of borders with the European Union appear to be the only significant explanatory factor. This evinces the need for other explanations for democratic satisfaction itself.

The multivariate models supports our main hypothesis that institutional choices do matter in terms of democratisation, while other factors (borders with the EU and previous democratic experience), which may be tentatively associated with the problem of path dependency, have a significant impact too.

Path analysis. Regression analysis with introduction of time lags in the previous section helped to make some observations on the temporal order of the associated changes in the response and explanatory variables, which may, with some reservations, be interpreted as indicators of causality. Another method of exploration of causal relations between the variables is path analysis, which seems to be a more suitable method in this case of introducing controlled variables (Walsh 1989).

Under the basic assumptions of institutional determinism, institutions have impact on democratic consolidation, while structural factors influence both progress in democratic consolidation and choices of institutional design, i.e. they can have direct as well as indirect effect on democratic consolidation (Diagram 2.2.1). We tentatively accept these assumptions for the purpose of the following path analysis.

This model has been assessed on the basis of existing data by (1) running a multivariate linear regression model for the index with introduction of all seven above-indicated variables as independent; and (2) running multivariate models for executive power and arbiter authorities with inclusion of four structural variables as independent.44 The result of the path analysis is shown in the Diagram 2.2.2 (only path effects significant at 0.05 probability level are indicated).45

According to this model, there are three factors with a significant direct effect on progress in democratic consolidation. Among them, the executive power of presidents and their arbiter authority are institutional variables, while previous democratic experience is a structural one. However, the powers of the chief executive vested in presidents are strongly determined by structural factors, especially the existence of borders with the EU and previous democratic experience. The arbiter authority of presidents appears more independent from the chosen structural factors.

The model which may be inferred from the path dependency theory is a simpler one, since it does not imply institutional choices as possible interfering variables. Diagram 2.2.3 presents the results of testing this model.

Both models are strong; they account for 63.1 and 55.1% respectively of variance in the index. However, the second, ‘structuralist’ model is the simpler one, since four instead of seven explanatory variables are introduced (compare the model with three institutional explanatory variables which explains 50% of variance under pairwise case exclusion).

Thus, both structural and institutional explanations may be effective. The latter results conform to the predictions of path dependency theory. However, institutional variables appear to be strong interfering factors. Among them, the executive powers of presidents are more dependent on non-institutional factors, while presidential arbiter authority seems to be a more affordable institutional choice, associated with positive outcomes in terms of democratic consolidation (except the cases where it has been combined with the strong position of the president as the acting chief executive, i.e. super-presidential regimes in the Eastern part of the region under study).

Conclusion

The main task of this study was to test the hypothesis on relevance of the choices in institutional democratic design for the progress (or failure) in democratic consolidation, and provide the reader with an assessment of the outcomes of certain institutional choices. Democratic consolidation has been operationalised as an index accounting for democratic satisfaction of the population (a political culture variable supposed to have positive impact on stabilisation and increasing social embeddedness of the new regime) and preservation of human rights (a variable controlling the democratic system performance in a country). Operationalising institutional variables, two dimensions of presidential authorities, proportional component of parliamentary representation, and number of general and parliamentary elections have been distinguished as the most promising explanatory factors. Different domestic and international factors have been controlled as possible extraneous variables. Methods of mean comparison, bivariate and multivariate regression analysis, and path analysis have been applied for the further data analysis.

Three theses are proposed on the basis of the previous analysis:

In particular, presidential power as the chief executive is one of the most explanatorily powerful institutional variables, negatively associated with progress in democratic consolidation. However, the arbiter authority of presidents is associated with a better record in terms of democratic consolidation, especially in the first years of the studied period. The proportional component in parliamentary representation has a positive association with advances in democratic consolidation, while this may be due to the association of this component of the electoral system with the weaker executive powers of presidents. A possible explanation of the outstanding record of the combinations of parliamentary and semi-presidential governments with proportional representation is its facilitating effect on the development of a responsible party system, and, consequently, establishing a strong link between society and politics.

Semi-presidential regimes appear to be more suitable for a new democratiser than purely parliamentary ones. A president who has been given some arbiter authority but denied the direct influence on government formation seems to be a well-functioning model in the region. Presidential arbitrage may serve as a counterweight to the ‘party state’ in political systems, where the parliament is responsible for government formation.

This paper supports the relevance of the institutionalist approach to the study of post-communist democratisation. Some of our findings do not conform to the existing institutionalist hypotheses, while others fit them well. Nevertheless, institutional variables are established as powerful explanatory factors for the progress in democratic consolidation. At the same time, we assume that the institutionalist debates may benefit from shifting their main focus from a policy-making perspective towards more attention to the link between society and politics. Thus, an assessment of the effects of institutional choices in terms of how they contribute to the development of a responsible party system and civil society is challenging for further investigation.

Endnotes

1. This article was originally published Democracies, Markets, Institutions: Global Tendencies in Local Contexts. Warsaw, IFiS Publishers, 2002. Many thanks to the Centre for Social Studies for their cooperation in the re-publication of this article.

2. More precisely, their sample was made up of attempts toward democratisation, including Latin American cases but excluding small countries with population less than one million.

3. This does not include Bosnia-Herzegovina because of the state problems there (see: Henderson and Robinson 1997). Four newly independent states of Central Asia are also excluded from our study since little democratisation effort was undertaken there, while other countries had some democratic experience, at least, in terms of open elite competition for electoral support.

4. Surveys on democratic attitudes, election results, statistics on socio-economic developments, and international studies of democratic performance (Human Rights scores by the Freedom House, Polity Archive etc.) have been employed as the sources of data.

5. The approach which concentrates on the well-functioning of democratic institutions, is not called 'institutionalist' but 'system' in order to avoid misleading associations with institutional explanations of determinants of democratic consolidation and emphasise the crucial point of this approach, the relative importance of system properties of a democratic polity as an institutional complex in comparison with the democratic outlook of particular institutions.

6. "- Structural: ... It posits that no significant reserve domains of power should exist that preclude important public policies from being determined by laws, procedures, and institutions that have been sanctioned by the new democratic process. - Attitudinal: When a strong majority of public opinion acknowledges that the regime's democratic procedures and institutions are appropriate and legitimate, and where support for antisystem alternatives is quite low or isolated from the prodemocratic forces. - Behavioural: When no significant national, social, economic, political, or institutional actor spends significant resources attempting to achieve its objectives by challenging the regime's institutions or rules with appeals for a military coup or revolutionary activities, and when prodemocratic forces abide by its rules and do not engage in semiloyal politics." (p. 79). However, the authors concentrate mainly on the elitist aspect of democratic consolidation in their study.

7. According to Merkel, an effective democracy has three dimensions: (1) universal suffrage, (2) 'effective monopoly on government by democratically-legitimated representatives', and (3) 'liberal constitutionalism and rule of law' (Merkel, 1999, p. 7). Consequently, Merkel distinguishes between three types of 'defective democracy': (1) 'exclusive democracy', where some significant social groups are denied the rights to effective political participation; (2) 'domain democracy', where some spheres of politics are excluded from democratic control and such veto-players as military or former authoritarian leaders have significant roles; and (3) 'illiberal democracy', where effective social control of state power and protection of human rights are absent.

8. In O'Donnell's 'delegative democracy', (1) president stays 'above' parties, (2) power of parliament and judicature is minimised, (3) policy is set by presidential administration, and (4) president is considered to have full responsibility for developments in the country (cf. Moiseev 1998).

9. Alternatively, the researcher can employ or avoid direct reference to a particular democratic model and exploit the situation in the social sciences when such concepts as 'democratic elections', 'multipartism', or even 'human rights' seem to be more precise than 'democracy' itself.

10. Burtan, Gunther and Higley have elaborated two criteria of elite consolidation: (1) structural integration ("the relative inclusiveness of formal and informal networks of communication and influence among elite persons, groups, and factions"); (2) value consensus ("the relative agreement among elites on formal and informal rules and codes of political conduct and on legitimacy of existing political institutions") (Higley and Gunther 1992, p. 10).

11. Challenging as a new methodological perspective but ambiguous from the viewpoint of its theoretical grounding, operationalisation of democratic consolidation has been proposed by A. Schedler (1997). The authors defines democratic consolidations expected stability, while the latter can be assessed through public opinion surveys and interviews with representatives of elite. Unfortunately, there has not been done much in empirical assessment of properties of this operationalisation. Schedler himself refers to a work by McClintock, who had conducted several interviews in elite in Peru. The latter study suggested rather high level of democratic consolidation in the country just before the crises. Schedler's explanations on this shortcoming of his theory are rather vague.

12. For example, the large differences in dispersion of democratic values between different social strata can hinder democratic consolidation (Waldron-Moore 1999); one should control the average level of acceptance of democratic values as well as their average acceptance. Another example is the differences of experiences with authoritarian regimes in post-communist countries. They can affect the responses to the survey questions, and diminish the ability of the data to be compared (Evans and Whitefield 1995).

13. There are many other combinative concepts. For example, Dawisha (1997, p. 44) enumerates four aspects of democratic consolidation: "the turn-over test, low public support for anti-system party and groups, high public commitment to the fundamental values and procedural norms of democratic politics, and elite consensus about the desirability of institutionalising and legitimising democratic norms and values". He criticises the applicability of the two turn-over test to the study of post-communist cases and suggests that a proper index of democratic consolidation can be developed as a combination of the latter three criteria.

14. This concept first appeared in the influential essay by Rustow (1970), who claimed that the consciousness of rival political groups and their inability to destroy each other are necessary conditions for successful transition from authoritarian rule to democracy. Democratic consolidation can be considered as the last phase of the regime change, i.e. normalisation of the new regime (Kaase 1994). This approach was modified in respect to the third wave of democratisation (including post-communist cases) by Linz and Stepan (1996a).

15. Since the scores for political rights and civil liberties reflect different dimensions of political system performance they should be rather multiplied than added to each other. In order not to overweigh the factor, the square root is to be extracted from the product of the multiplication.

16. The data on satisfaction with democracy will be obtained from the Central and Eastern Eurobarometer surveys.

17. For a critical review see: Hall and Taylor 1996

18. These issues of institutional design have become central, since the democratisers faced a pressing need to balance, on the one hand, necessary decisiveness required for the urgent reforms with division of powers and, on the other hand, governability with representativeness of major political institutions.

19. For a critical revision of the argument see: Cheibub 1999

20. "(1) Proportional representation tends to lead to the formation of many independent parties, ... (2) the two-ballot majority system tends to lead to the formation of many parties that are allied with each other, ... (3) the plurality rule tends to produce a two-party system" (Duverger 1986).

21. The authors refer to Mills' venerable idea that party systems tend to a two-major-coalition format since the crucial goal of the party competition is to obtain a majority. Due to a lot of evidence contradicting the laws' predictions, authors incline to either reformulate Duverger's idea in a probabilistic concept and consider the electoral formula as one of many factors influencing party system formation (see: Riker 1986), or reject the hypothesis.

22. This thesis is supported by Suarez (1982) who examined a large data set on heads of state and government in Western democracies between 1940 and 1976, and found that heads of presidential governments have far less ministerial experience than their counterparts.

23. The lack of openness of the political elite and its relative rigidity has been seen as one of the major problems of developed representative democracies based on a parliamentary government system and proportional representation (see: Glotz 1997).

24. The meaning of the 'strength' of a party system and its possible consequences also need to be clarified in that context.

25. This is another place in the present study where we have to complain about the lack of available data, especially from the early years of post-communist transformation and the Eastern part of the region under study. Due to the lack of respective data, we are unable to introduce in our analysis the cases of Armenia, Azerbaijan, Belarus, Georgia, Kazakhastan, Moldova before their current constitutions had been enforced in 1994, 1995, 1996, 1994, 1994, and 1993, respectively. Table 1.2.3 presents the list of the cases included in the analysis of presidential powers.

26. The data set on presidential powers is based on the original analysis of respective constitutions. It is recognised that there may be some errors but assumed that these errors will not strongly influence the results of the analysis.

27. In particular: (1) relative independence from the legislative differs in the way the presidents are elected (popularly or by a body consisting mainly out of members of parliament), term in office, majority in the parliament required for the presidential impeachment, mandatory need in a judicial approval for the impeachment, ability of the parliament to amend the constitution without a plebiscite; (2a) direct influence on the executive depends on their rights of obligatory nomination of the prime minister, appointment of the prime minister without vote of confidence in the parliament, dismissal of the prime minister, and appointment of other members of the cabinet; veto right on the governmental decisions and participation in the sessions of the cabinet; (2b) ability of presidents interfere in the legislative is determined by the provisions concerning the right of a president to appoint some members of Parliament, decide upon dissolution of Parliament if no responsible majority, or no support for governmental actions, have veto power, as well as the majority in Parliament required to rid the veto, presidential ability to call referenda, issue decrees on the basis of existing legislation, in case of emergency, and legislative decrees; (2c) since all presidents are given the right to pardon, denied the ability to directly interfere in the judicial process on the side of the prosecution, their strength in relation to the judicial is determined by the provisions which empower them to influence the personal composition of this branch of government; (2d) all of them are Commanders-in-Chief of the armed forces and have to fulfil some politically significant ceremonial duties; the differences appear in their rights to appoint regional executives, appoint and dismiss army officers, order mobilisation, conclude international treaties, appoint representatives of the state, and declare state of emergency. However, this list is not exclusive and counts for the authorities which can be compared on the basis of constitutional provisions and their following development in the legislation.

28. All the above-mentioned correlations are significant at 0.01 probability level (two-tailed).

29. Application of factor scores as a measurement may result in the difficulties associated with robustness and interpretability of the assigned values. Thus, developing more reliable scales will be challenging for further investigations.

30. Single transferable vote, or such complex proportional-like systems as the Hungarian one may allow non-party candidates to get into parliament. However, these cases are rather exceptional and the major electoral competition takes place along the lines of party representation. Thus, we have regarded these systems as a kind of proportional representation.

31. The religious part of the population, rather than the whole population, is applied as denominator because we are interested in the relative strength of religious traditions.

32. Available-on-line: http://k-gleditsch.socsci.gla.ac.uk/Polity.html.

33. The situation with availability of socio-economic data described by Henderson and Robinson (1997), still lasts: "no single authoritative source of comparable data exists. Many international organisations publish varying figures, having used different bases of computation, and all are reliant on the raw data complied within the countries themselves" (p. 279). The authors especially criticise the reliability of the data on GDP decrease in the first years of democratisation (the decrease can refer to the ceased production of low-quality goods without public demand, or conversion of the military industry) and the data on unemployment.

34. Unfortunately, the contemporary data on the ethnic composition of particular countries are not obtainable, and we have to rely on the data from the last censuses made in 1989-91. These data do not account for the changes caused by migration waves in course of transformation.

35. Application of the significance test, a concept related to the probability of making errors in generalising sample results on a population, may be criticised since the analysed data set nearly exhausts the population under study. However, the significance test appears as a suitable tool in terms of our research design, if we assume a probabilistic vision of the actual developments, and consider the 'actual Eastern Europe' (picture of the region suggested by our indicators) as a 'sample' randomly drawn from probable developments in the region. Consequently, the significance test may be instrumental for testing these hypotheses, and it will be employed as an additional supporting criterion for identifying main features of the processes.

36. Correlation with human rights scores has positive value since the higher scores on this scale are given in the cases with less advanced human rights preservation. The reader is asked to pay attention to this particularity in the following text in order not to be mistaken by the frequently different signs before R coefficient. If the signs are different, this means that the phenomena correspond in a similar way with explanatory factors.

37. As one example, we can mention the experience of Slovakia, which moved from an almost purely parliamentary regime toward popular elections of the president and a slight increase of his powers after several years of ambivalent experience in the development of the link between civil society and politics, and authoritarian tendencies under the parliamentary government.

38. A possible explanation is that parliamentary governments may (1) have broader support in the legislative and, thus, more possibilities to implement necessary reforms, (2) imply the inevitability of coalition negotiations among political parties, and, thus, result in a stronger and more responsible party system; while the premier-presidential type of government formation may be more associated with the direct appeal to the public from both competitors (head of state and parliament) and, consequently, better conform to current public expectations in the situations where a parliamentary government could be more rigid.

39. Here and in the following, 'n' means number of cases.

40. The combination of semi-presidential regime and mixed electoral representation gave excellent results on all three scales of our response variables in Slovenia.

41. The strong correspondence of Protestant and Orthodox (both positive) and Muslim (negative) heritage with preservation of human rights in the Eastern part of the region under study may reflect regional differences in democratic institutionalisation within this sub-region, between its Northwest and Southeast parts.

42. Regional inconsistency in the correlation of democratic experience and level of GDP (these factors do not have significant correlations with indicators of democratic consolidation in the East) has its explanation rather in the lower variation of these factors and lesser number of cases in this part of the region under study.

43. Among controlled variables, the existence of borders with the EU and GDP per capita are the strongest explanations.

44. The model for proportional representation has not been assessed since the effect of proportional representation on index was insignificant in the multivariate model for the index.

45. Pairwise case exclusion has been applied; e2 means percentage of variance which is not explained by the introduced independent variables, i.e. may be accounted for possible extraneous factors. Explained percentage of variance (r2) can be easily calculated according to the formula: r2 = 1 - e2.