Abstract
Since gaining its independence in 1990, Ukraine has experienced two revolutions and a war in the east of the country. Meanwhile, it has sampled all three major electoral systems (majoritarian, proportional, and mixed) and has created dozens of major parties and political blocs. Such a volatile context has consequences for such crucial political issues as accountability and circulation of the political class. At the same time, it allows researchers to address questions concerning the sources of change and stability in the politics of countries in transition. In particular, the case of Ukraine allows us to pinpoint the effects of popular upheavals and changing institutional arrangements on the stability of a political scene and the population of politicians. Data limitations of previous studies have prevented such a necessarily detailed temporal accounting. We analyze it for all Ukrainian MP hopefuls during the 1990–2014 period using the East European Parliamentarian and Candidate data (EAST PaC). The issue of political stability was approached by examining affiliation switching and dropping out in all consecutive elections to the Verkhovna Rada. The results reveal that revolutionary events had relatively minor immediate effects on the circulation of the political class, while individual strategies were highly dependent on the electoral rules working in conjunction with other traits of the political system.
Introduction
In the course of the twenty-five years of its post-Soviet history, Ukraine has experienced two major political upheavals (Orange Revolution, Euromaidan), a pro-independence movement (which, in contrast to the revolutions, was relatively consensual), three major electoral systems—majoritarian, proportional representation (PR), and mixed voting rules. All this was accompanied by high flux on the party scene. It is compelling to pose questions of how all these political developments are interrelated. In this article, we aim to analyze the micro-foundations of a volatile party system by looking into the tendencies and presumed incentives in party switching (and, conversely, party attachment). In particular, we look at how electoral rules and the context of revolutionary events have shaped politician–party ties in Ukraine.
In many young democracies, especially in Central and Eastern Europe (CEE), discussion of the effects of electoral rules is ongoing. 1 The arguments in this dispute often echo the distinction between the accountability provided by majoritarian systems and the scope of democratic representation and power sharing granted by proportional voting systems. 2 In both cases, the issue is addressed at the level of a party system, with the party as a primary unit of analysis. In recent studies of party systems, more prominence has been given to the processes underlying party politics while mostly abandoning the basic assumption of treating the party as a unitary actor. 3 Parties are thus “endogenized” and internal organizational processes moved to the spotlight of political analysis. 4 Especially in cases of “floating party systems,” with weak voter identities and low institutionalization, it is the individual politician that is treated as the more valid unit of analysis. 5 The argument in this respect is that volatility of political affiliation leads to a shakier party system as it blurs boundaries and diminishes accountability. 6 One of the key topics in the field is the impact of the electoral system. 7 The influence of rules for the popular vote goes well beyond its immediate results, as such influence is observed also in legislative behaviour. 8 It is also not inherent only to voters’ decisions, as it is the recruitment and selection processes that produce a “supply” of political candidates, consequently predefining the ranks of the political elite. 9
In this article, we aim to dissect the effects of certain institutional arrangements vis-à-vis popular upheavals on the politician-party attachment, therefore adding to the interpretation of the recent political history of Ukraine as well as knowledge of contextual effects in the formation of a party system. The study thus contributes to both—literature on intra-party politics and on Ukraine’s politics. Ukraine’s party system is often branded as fluid or volatile and as such represents a specific setting for political research. 10 It can generally be presumed that it was turbulent events and electoral-law engineering that magnified this shakiness. On the other hand, the literature concerning the consequences of the Ukrainian revolutions suggests that they brought more fundamental change to civil society than to political elites or institutions. This was much debated in the context of the Orange Revolution. 11 Konstantyn Federenko, Olena Rybiy, and Andreas Umland argue that after Euromaidan the essential characteristics of the country’s party system did not change as well. 12 As Ukraine’s politics has been often labeled oligarchic, the issue of a self-reproducing political class seems to pose an intrinsic research problem, while the overall impact of popular upheavals can show the degree to which domestic turbulence tangibly influenced political elites during the post-communist period. 13
In our analysis, we have used the East European Parliamentarian and Candidate Data (EAST PaC) on Ukraine. This encompasses the whole population of candidates to the Ukrainian parliament, the Verkhovna Rada (VR), from 1990 to 2014. It enables us to compare personnel fluidity in the ranks of the Ukrainian political class parallel to changes in electoral rules and to most significant political events. In other studies on this issue in the Ukrainian context, the scope of analysis has been confined to the parliamentary roster or higher positions in the state administration. 14 Our approach is more comprehensive as it encompasses relations between parties and all would-be MPs and is based on the assumption that there are important processes underlying the mutual assortment of individual politicians and parties, and that those processes may have implications for the stability of political representation and accountability. In the data covering only MPs, this assortment is overlaid by another selection process—the voters’ verdict. 15 What is more, switching between Verkhovna Rada factions was banned altogether in 2004–2010, which disables analysis of intra-parliamentary fluidity throughout the independence period. 16
Ukraine is a specific case for analysis of the “electoral game,” yet this conveys ample merit as it provides substantial variability of institutions and events to allow checking their consequences within the same polity. Conclusions drawn from analysis of individual data in a nationally bounded context are relatively safe as those based on aggregated data (on countries, parties) can be subjected to additional inference fallacies. For example, international comparative studies involve a relatively high risk of omitted-variable bias—the results might appear as an outcome of failing to take unique, underlying vernacular factors into account. A well-defined national or regional context is a safer setting in this respect. The advantage of a detailed national case is that it takes into account and enables researchers to control the plethora of social, cultural, and institutional factors. 17 Many scholars see the necessity to distinguish the effects of working rules—that is, actual political mechanisms—within certain developmental or national contexts. 18 Arendt Lijphart advised different institutional frameworks for new democracies, 19 whereas Andrew Wilson saw altogether different sets of rules governing democracy in Eastern Europe. 20 Sarah Birch argues that the impact of electoral institutions depends on the context in which they are embedded, while in post-communist countries political competition often takes place—at least partially—outside the traditional, narrow field of party politics. 21 Siavelis and Morgenstern affirm that expecting uniform responses and ignoring contextual factors is a serious flaw of institutional theories in the field, because the latter interact with institutions. 22 All these arguments suggest that the validity of observations concerning the impact of institutions on politics depends on developing a framework suitable to the specific national context. Our study is designed specifically for the Ukrainian one, yet precisely by taking its idiosyncrasies into account it can provide a general contribution to knowledge on the subject—by further nuancing it.
Research Framework
Institutional and Contextual Factors in Party Switching
There have been eight consecutive parliamentary elections that produced the independent Ukrainian legislature since 1990, none of which is fully comparable to the previous or the following one—either because of an alteration in the voting system or because of some exceptional circumstances (e.g., preterm elections were held in 1994, 2007, and 2014). This makes the data on VR candidacy an attractive case for the study of restructuring of the political class. It “supplies” analyses with additional, distinct features—popular upheavals, flux of electoral rules, and specific institutional setting. At the same time, many other contextual factors remain relatively stable—throughout the period, it is mostly the same population that produces both the ranks of candidates and the votes—which makes the ceteris paribus assumption relatively more viable. It allows us to look at events and particular rules as stimuli for political actors. With such a setting, we aim to empirically address two questions: one concerning the influence of the electoral system on a candidate’s attachment to a party or political bloc and another concerning the impact of revolutionary events on the fluidity of the Ukrainian political class.
As stated in the previous section, our purpose is to look beneath the “surface” of a party system, at the underlying processes of selection and recruitment. The basic analytical framework we apply assumes that the act of candidacy is a mutual choice by a politician and a party (or other political entity, like bloc or committee). It implies that the actual scope of party switching is shaped by matrixes of incentives on both sides of the “bargain”: parties (their leaders) and individual candidates. The selection processes play a crucial role in the shaping of a party system, though at the same time they are deemed to interact with other factors, such as the electoral system. 23 The incentives themselves, therefore, have to be seen as being shaped by institutional arrangements, including mainly electoral rules, but also procedures for coalition formation, and other formal and informal rules governing recruitment and selection. 24 In the case of Ukraine, it is possible to compare the effects of institutions with the political outcomes of revolutionary events that can be expected to magnify reshuffling in the political class. It is particularly advantageous to have such reference points as it allows a relatively straightforward interpretation of results.
In the case of the effects of electoral systems, Duverger’s Law becomes a departure point in at least two respects. 25 The “classical” argument on accountability, largely based on the Law, underlines the difference between voting in single mandate district (SMD) and on party lists. 26 While the argument is mostly focused on the clarity of alteration in power with majoritarian systems, it also implies that an individual with a strong position in a particular constituency can be a valued asset for political parties in majoritarian elections. 27 From the proposed “incentives” perspective, this might in fact allow candidates to choose among parties’ offers and to select affiliations that reward them most. In PR elections, where ideological representation and popularity of party leaders have more weight, candidates have much stronger incentives to be loyal, as their election is more contingent on the success of the whole party. In other words, with different electoral rules either side can be more of a “chooser.” Therefore, in certain contexts, a plurality vote might in fact undermine accountability (especially given that party switching often illustrates a divergence between the accountability of a politician and of a party).
The intra-party perspective clearly diverged from the Duvergerian tradition. In particular, Giovanni Sartori pointed out that while the electoral system cannot precisely explain the party system itself, it is an important factor determining the amount of internal fission and fusion. 28 We hypothesize that while the PR system does not necessarily encourage party cohesion, politician–party agency balance is of more importance in cases with less institutionalised party systems, for example, Ukraine. 29 The argument is that more weight at the individual scale in certain institutional and social arrangements (which are discussed below for the Ukrainian case) in fact makes the party scene more prone to individual affiliation switching. Therefore, the outcome of the agency balance is a crucial ingredient in the relation between the voting system and the party system described in the Duvergerian theoretical tradition. In other words, formal and informal institutions, including electoral laws but not limited to them, impact not only the party-level logic of politics but also intra-party logics, in particular, individual politicians’ strategies.
Studies in intra-party politics bring a nuanced view to the issue, as more contextual factors are taken into account. Lise Rakner, Lars Svåsand, and Nixon Khembo demonstrated that first-past-the-post electoral systems encourage politicians to run independently rather than join forces. 30 In such settings, parties are often seen as “vehicles for political advancement rather than as communities of ideologically like-minded individuals.” 31 Elad Klein has shown that parties that experience diminishing support are more likely to lose members in candidate-centered voting systems than in party-centered ones. 32 This would imply that the “crew of a drowning ship” can avoid sinking by finding shelter in other, more stable “vessels.” Nevertheless, the assertion that electoral systems that encourage personal voting provoke disunity and fission is not uniformly accepted. 33 Diana O’Brian and Yael Shomer found some support for the claim that systems which encourage personalized behavior are more likely to witness switching, but the direct effect of institutions was minimal in their study. 34 Many cases serve as straightforward exceptions. Carol Mershon and Olga Shvetsova analyzed a cross-national data set and found a negative relation between single-mandate districts (SMDs) and party switching in parliaments of nine developed countries. 35 William Heller and Carol Mershon argued that in Italy party switching was conditioned mostly individually and that there was no difference between the effects of party-lists and SMDs in this respect. 36 Scott Desposato compared chambers elected with different rules in Brazil and came to a similar conclusion. 37 Those studies strongly suggest that the relation between electoral system and party switching is contingent on other “ingredients” in the political context. It can be argued that this is especially the case in Eastern Europe, where the specificity of interactions between formal and informal institutions have been repeatedly accentuated. 38
In the Ukrainian context there are at least three factors that can hypothetically induce the effects of electoral rules on political attachment, as there may exist a specific structure of institutional incentives. First of all, while in many contexts (re)election is defined as the primary reward of a candidate, in some situations it can be treated as a secondary objective or merely a means (among viable others) to achieving a goal. To a large degree, this is inscribed in the course of coalition formation and distribution of important governmental positions. Party unity is expected to be lower under presidentialism. 39 As it was in the case of Ukraine, especially with Leonid Kuchma’s presidency after the adoption of the new Constitution in 1996, when the importance of configuration of parties in the parliament was diminished by the fact that the head of state had leverage over government formation and the ability to effectively “sweep” individual MPs and their affiliates into the VR majority. This was especially the case with independent candidates, who were not bound by party loyalty in their choices, but it applied to others as well. The phenomenon was branded “party of power” and was argued to be of central importance to parliamentary politics in Ukraine. 40 As Sarah Birch noted, “the large number of independent parliamentarians, most of whom have very weak ideological commitment, has facilitated the presidents’ ability to generate supportive coalitions through the distribution of extra-parliamentary ‘incentives.’” 41 The insight into that issue can be gained by treating independent candidacies not merely as a reference category but as a substantial affiliation category.
Second, it is also important what political resources can be effective in securing reelection. There is an ongoing discussion on the role of “administrative resources” in Ukrainian politics. 42 Wilson argues that with SMDs “cruder types of administrative resources can be more directly applied,” 43 like enforcing support from local businesses by administrative harassment or judicial interventions aimed at discrediting political competition. Semenova notes that in this respect presidentialism interacts with the plurality vote, as elections in SMDs are more easily affected using access to government—particularly in Ukraine. 44 In a political system where reelection can be relatively easily secured with access to administrative power, there are incentives to gain the latter by joining government. An analogical argument can be made for oligarchic political setting—if weakly restrained or unrestrained campaign finances matter more than ideologies and agendas then political parties play a relatively minor role. 45 As Nathalie Giger and Heike Kluver showed for Switzerland, even in legislative decisions MPs tend to be loyal not only to constituencies but also to interest groups. 46 Introducing some proxy measures for administrative power and access to financial capital into the analysis should answer the question about the effectiveness of those resources.
The third crucial factor concerns the relative position of parties in the political system. Causes of inter-party (and inter-bloc) flows can be based on the predicted popularity of certain political brands and on calculations concerning potential reconfigurations of the party scene. If ideological and organizational persistence is not considered to be rewarded by voters, there are strong enticements for a candidate to look for presumed “bandwagons.” This is one of the reasons why in young democracies with weakly institutionalized party systems, parties are often considered to have minor importance for understanding political competition. 47 As Mair noted, the political elites of post-communist societies are less likely to be motivated by party loyalties and commitments and more prone to strategic affiliation change. 48
What is more, heavy circulation in political representation might by itself reduce the politician–party attachment. In some political systems, the mutual assortment of politicians and parties is done mostly once in a lifetime (as in the United States), while in others it is made more regularly (as in Ukraine). In the Verkhovna Rada, the continuity of personnel is particularly low, as during 1990–2014, the average percentage of reelection was 37 (the number is partly inflated by the snap 2007 elections, when most of the MPs from 2006 won their seats again). In the United States, the probability of reelection is usually higher than 90 percent (with two-year terms, but nonetheless producing almost twice the length of service as in Ukraine 49 ). When there are high rates of replacement in parliamentary ranks, that is, when circulation is particularly strong, voters are more able to reward and punish parties than individual candidates. John Carey calls it “the party label’s information value.” 50 Though there are signs of personal accountability in Ukrainian politics, most individual parliamentarians can easily avoid being evaluated on the basis of policies—particularly when they switch affiliation. 51
Finally, with the party system still in the process of reshaping, candidates gain even more bargaining power as parties struggle to form a nationwide presence. This task is exceptionally hard with a deep, geographically pronounced political cleavage, as in Ukraine. 52 Low nationalization of the party scene serves as an additional factor in the candidate–party bargain where the plurality vote works especially strongly toward undermining party loyalty. 53
All the above-mentioned arguments suggest that in the setting of the Ukrainian political system the balance of agency between party and individual candidates might be tipped toward the latter. In such a context, PR vote might be a measure to reinforce party position in the process of selection and recruitment, while with personal vote seeking in SMDs the choice of affiliation by candidates would be much less constrained. The observable effect would be more intensive individual party switching with a majoritarian vote.
Our second research question concerns the effects of Ukrainian revolutions on the political class (here defined as the population of VR candidates). Ukraine serves as an important case with respect to the level of bottom-up change accompanying popular upheavals as the latter occurred more than once since Independence: in 2004 as the Orange Revolution and at the turn of 2014 as the Euromaidan (Revolution of Dignity). The threat of coordinated mass opposition is considered to be a basic requirement of self-enforcing democracy, thereby giving society a chance to punish “cheating” ruling elites and change their composition. 54 On the other hand, individual politicians can deploy strategies to counter such electoral punishment. 55 In line with the rationale given in the context of hypotheses concerning electoral rules, a variable party scene (i.e., the one with a low threshold of entry for a new political label and a high threshold for maintaining one) creates conditions in which party-switching allows politicians to remain in the electoral game despite the defeat of their hitherto party or bloc. 56 In sum, in the aftermath of revolutions, we expect heightened rates of individual dropping out of politics (for a part of the political class discouraged or discredited by the events), but also of switching of party colors (for those politicians who tried to stay in politics). The alternative hypothesis would state that change on the political scene had mostly a character of “rebranding” while dropping out of political competition and personnel reshuffling between parties remained close to the country average. In fact there seems to be stronger support for this alternative in the literature on Ukraine’s revolutions. Studies of the Orange Revolution mostly argue that there was more revolutionary semblance than substance. 57 There are already similar concerns for the Revolution of Dignity. 58
Data on Ukrainian elections
EastPAC-Ukraine data, which serves as an empirical base in our analysis, contains 35,798 cases of individual candidates (46,586 instances of running in elections) organized into eight elections that differed in applied rules and political contexts. In the parliamentary elections held in March 1990, Ukrainians were actually choosing deputies for the twelfth chamber of the Ukrainian Soviet Socialist Republic—not yet an independent state. 59 Though the members of this Supreme Soviet were to become the first parliament of independent Ukraine, the 1990 elections cannot be seen as completely free or democratic, as the Communist Party still held sway (especially with control over state apparatus and the media) and political pluralism was almost absent. Yet there were some rudimentary conditions for political representation and independent candidacy, brought about by the 1989 amendment of electoral laws made in the spirit of Gorbachov’s perestroika and glasnost, reforms. The elected deputies came to be the first Verkhovna Rada (VR) of independent Ukraine in the aftermath of the 1991 referendum. In the preterm 1994 elections a turnout threshold was added, which necessitated that the winner achieve an absolute majority of more than 50 percent of eligible voters (so at least 25 percent of all possible votes). This led to a series of runoffs in 1994 and 1995 for as many as 112 seats. 60 The 1994 elections were the first de facto democratic ones in Ukraine’s history, as attested by international observers. 61
In 1998 the mixed system was put in place. Half of the MPs were elected from nationwide lists (threshold of 4 percent), while the other half competed in the SMDs, this time according to the plurality (first-past-the-post) rule. 62 These were also the only elections in which candidates could simultaneously run in SMDs and on the party lists. The latter provision was abolished before the 2002 elections, but the mixed system remained in place.
The 2006 elections, held in the aftermath of the Orange Revolution, saw the introduction of the closed-list PR system and 3 percent threshold (length of term was set to five years). Independent candidacies and intra-parliamentary faction switches were disallowed. 63 It was argued that the plurality vote gave an electoral advantage to local business and administrative elites, hence enabling them to retain political influence, at the same time undermining parties and producing phenomena like the “party of power.” 64 These changes have assisted the emergence of a more pronounced party system but led to cohabitation, resulting in a political crisis. As a result, snap elections had to be held in 2007, with the same electoral rules in place. After Yanukovych successfully sought presidential election, he began to dismantle the institutional legacy of the rule of the “Orange coalition,” including bringing back the mixed vote for 2012 elections (this time with 5 percent threshold and a ban on electoral blocs). 65 Although those decisions brought much criticism from the political opposition, the electoral system itself stayed in place after the Euromaidan—the 2014 vote was decided according to the same rules as in 2012.
Those developments, together with other elements of Ukrainian transition, had many consequences for party politics. The elections of 1994 brought the emergence of 33 political parties, a number that grew later. Parties often united into blocs (in 2014 the latter were still present, though “dressed up” as parties), with the names of parties and blocs often coinciding, making the system relatively complex. Candidates running in SMDs as independents often enjoyed the support of political forces (sometimes in the form of not putting forward rival candidates) while candidates on the party list have not always been formal members of that party. This complexity influences how we treat candidates’ political affiliation. In our study, it is defined on the basis of a presence on a particular electoral list (with the exception of the 1990 elections). The notion of affiliation is hence treated in a relatively flexible manner, in step with the electoral rules in Ukraine. In particular, when a party entered any of the major political blocs, this was not treated as a change of affiliation (party switching) for its members. 66 Likewise with party mergers, splits, and simple rebranding—the operational definition of switching is narrow, and hence safe. Additionally, party coding for the 1990 elections takes into account only formal membership in the Communist Party of Ukraine (CPU), while the political factions of the first Ukrainian parliament included the rightist “Rukh,” moderate democratic “Demokratychna platform,” and several others—representing non-party nomenklatura, non-party professionals, and cultural elites. 67 Membership in those groups was highly volatile and informal, and in fact the demarcation line between communist and independent or pro-democracy politicians was not always clear. 68 Therefore, for the 1990–1994 period, switches were defined only as changing affiliation from the CPU or to the CPU. All other changes of affiliation were treated as retained party attachment.
Analysis
Historical Account of Affiliation Instability in Independent Ukraine
We introduce the analysis of party switching in Ukraine with a brief, historical description of personnel flows between major political forces. Figure 1 shows movement of the “actors” on the political scene during each of the seven inter-election spells. After the first term of the VR, most of those former CPU members who decided to stay in politics became independent candidates, while the rest split mainly between left and center-left parties: the new, reborn Communist Party of Ukraine (CPU), Peasant Party of Ukraine (SelPU), Socialist Party of Ukraine (SPU), and Social-Democratic Party of Ukraine (SDPU). Those who were not members of the CPU in 1990 split between the People’s Movement of Ukraine (NRU), other minor parties, or ran again as independents. In a way, this has been a “big bang” for the Ukrainian political scene—the rising number of parties was accompanied by the highest to-date flows of politicians. Elections in 1998 witnessed the majority of candidates staying relatively loyal to the aforementioned political forces, with only minor transfers between (what were considered to be) rightist and leftist parties. The important development was the rise of two new political entities in the center of the political scene—one was liberal-leaning Hromada (of the ex-PM Pavlo Lazarenko), the second was the pro-government People’s Democratic Party (NDP, led by Anatoliy Matviyenko and Valeriy Pustovoitenko, the PM at the time). Both parties formed out of large numbers of hitherto independent political actors, in many cases former or current officials. There was a large influx of independent candidates to other political parties as well – including the surprisingly successful Greens. In 2002, Yushchenko’s Our Ukraine bloc was formed, which prominently included NRU. Hromada was “replaced” by BYuT, headed by Tymoshenko, former minister in Yuschenko’s cabinet. Just as four years earlier, this showed the relative importance of state administration in successful mustering of new political entities. The NDP was replaced by a broader political “centrist nomenklatura-oligarchic” platform 69 —the “For United Ukraine” bloc (FUU), with the Party of Regions (PoR) rising to primary position within its ranks. General inter-party movement became intensified, as the new players came to dominant positions in the political scene. After this point, affiliation switching became less intensive in Ukrainian politics.

Candidates’ mobility between parties and electoral blocs in the Ukrainian elections of 1990–2014
The 2006 VR elections were the somewhat delayed aftermath of the Orange Revolution. On the institutional dimension, it brought in a new voting system, but at the same time it froze the existing political cleavage. In the PR voting, independent candidates disappeared—all of them being dispersed among existing parties. This could have inflated party switching at that time—which would ostensibly back our hypothesis of higher interparty mobility during the post-revolution period—but no unusual rise of switching odds can be observed in the 2002–2006 period. In 2007, as was mentioned, there were relatively few transfers with the exception of candidates from minor parties jumping on presumed “bandwagons.” It is worth noting that there was more movement from the NU and BYuT towards the PoR than the other way around. The 2012 election witnessed the demise of the NU and the dispersal of its candidates, mainly to Volodymyr Klychko’s UDAR, but also to many less successful parties. Independent candidacies were allowed again, and many candidates resigned from their previous affiliations, which was especially significant in case of the LB (in total, more candidates from 2007 left its ranks than stayed; in most cases, they became independents). The “spinoff” Radical Party emerged from the ranks of the BYuT, while for the first time Svoboda entered the Parliament. Apart from that, the 2007–2012 period was a time of relative stability for the political scene, despite the fact that the imperative mandate was abolished and flows toward “independent” affiliation were allowed again. As was already stated, all the evidence points in the direction of the binding effect of a stronger socio-political cleavage at the time.
After Euromaidan, the party scene was reshaped once more and inter-party movement became more intense. As marked by the volume of nodes in Figure 1, Petro Poroshenko’s Bloc became the most attractive platform to join, mostly for members of parties previously out of the VR (the latter became more numerous than UDAR’s candidates, even if we omit independents in the calculations). There was a similar pattern for the People’s Front, though it attracted relatively more independents. On the other hand, Svoboda and the CPU proved to be relatively hermetic, and both failed to reach the 5 percent threshold. Once more this showed that in Ukrainian politics, a party’s roster (or “political HR”) can be a crucial factor in the electoral success.
Factors Influencing Party Switching in Ukrainian Elections
The analysis presented here considers switches between major political forces (parties and blocs) as well as between parties and the status of independent candidate. The latter, in line with the framework laid out in previous sections, is treated as a special kind of affiliation. As Table 1 shows, changes to the population of candidates happened mostly through dropping out—a decision that was likely the result of an unsuccessful attempt to enter parliamentary politics. Additional analyses showed that with statistical control of individual characteristics there were no clear, statistically significant effects of electoral rules or any of the two revolutions on dropping out (effects have been positive, though very weak). If the party switches and dropouts are taken together, then relatively many of them happened in the pre-term elections of 1994, while the highest share of candidates kept the affiliation in the pre-term elections of 2007 and 2014. The fall of communism and the “restart” of politics in the conditions of independence brought the biggest personnel reshuffle. There is no clear pattern in such nationally aggregated data with respect to electoral rules or—more surprisingly—revolutionary events.
Retaining Political Affiliations, Switching Them, and Dropping Out from Subsequent Races by Candidates in Ukrainian Parliamentary Elections of 1990–2014
To gain further insight into the issue of candidate–party attachment, a more detailed analysis, including individual candidates’ traits, needs to be applied. With the prevalent mixed system, we gain the opportunity to dissect differences between those running on party lists and those running in SMDs in the same elections. Ukrainian data also encompass cases of SMD candidacies in both plurality and mixed elections, and conversely, PR candidacies in both proportional and mixed elections. In this sense, we have a crossed comparison of voting systems (between and within elections), of elections (with different rules and partly repeated candidacies), and of individual races (in different systems and elections). We look at how the ties between individuals and political entities changed—whether they were preserved or changed across consecutive elections. Analysis conducted on combined data in the long form necessitates distinguishing between the five temporal types of independent variables: separate inter-election periods (chronological), encompassing all periods (constants, like sex), describing the beginning of a period (start variables) and the end of that period (lead variables), and finally those that apply to previous periods (lagged variables).
To analyze the influence of the aforementioned factors on affiliation fluidity we used multilevel logistic modeling. This is a multivariable statistical technique that allows us to check relationships between dichotomous outcome and multiple independent variables, and at the same time enables us to take into account the complex structure of the data (in this case, longitudinal). Table 2 contains seven differently specified models explaining party-switching for all consecutive races from 1990 to 2014. In the logistic models, chances of changing affiliation for each factor are expressed as odds, that is, a ratio of probability of “success” (in this case, switching affiliation) to probability of “failure” (retaining it). The constant term shows the odds of switching in the reference category (i.e., those running in 1998, women, those aged between 36 and 59, and so on) and the coefficients below are the multiplicative odds ratios, that is, ratios of the odds in the particular category relative to odds in the reference category. Values greater than 1 show the higher-than-average odds of switching, while those lower than 1 indicate below-average switching. 70 The first two models use standard, “flat” regression with a logit link function, while the rest involve multilevel specification.
Logistic Models of Switching Affiliation (As Compared with Retaining It)
p<0.001, **p<0.01, *p<0.5, †p<0.1.
Only periods leading to elections with mixed rules (1994–1998, 1998–2002, 2007–2012, 2012–2014).
Only the winners of (start) elections (MPs).
Model I involves only the period-specific fluidity in party affiliations. It shows that the gross effect of particular circumstances is relatively high, as it accounts for 9 percent of the whole deviance. We treat the 1998–2002 period as a reference category (preserved mixed voting, full term, some mass protests, but without decisive political consequences) and with the exception of the preceding VR term (1994–1998), all other periods differ significantly in the scope of political mobility. As was already shown, the probability of retaining party affiliation was particularly low in the 1994 elections, and very high in 2007.
Despite that the two shortest VR terms are characterized by less switching, there is no simple relation between length of term and attachment (snap elections were held in 1994, 2007, and 2014, while the longest term ended in 2012). In the aftermath of both revolutions, the odds of switching were, respectively, 0.70 and 0.68, therefore less than in the whole period of 1990–2002, but more than in the two VR terms between 2006 and 2012. The strengthened party attachment in this period can be explained in at least two ways. First, the pronounced cleavage that emerged after the Orange Revolution could have brought a period of stronger political attachment. It would be a matter of a wider (hence harder to cross) ideological gap between political forces. This gap would have been at least as wide in the 2012–2014 term, but relatively more switches were brought about by the changes in the aftermath of Euromaidan. Another explanation might involve the imperative mandate introduced in 2004, which was overturned in 2010 and then reintroduced in 2014. In this case, however, we would expect clear difference in party attachment between MPs running for reelection and all the other VR candidates, as it was only the former who were directly impacted by the imperative mandate. Additional analysis showed that this is not the case. MPs were, surprisingly, relatively more prone to changing banners after 2004 than before. This result suggests that it was rather the cleavage itself that stabilized party loyalty.
Period-specific explanations tell only a part of the story: Both in politically turbulent times and in times of relative peace, some people preserved their affiliations while others reconsidered. Much of the explanation lies elsewhere. Adding individual characteristics allows us to look closer at the personal preconditions of affiliation switching (models II-IV), and at the same time slightly changes the period-specific odds. While the spells of 1990–1994 and 2002–2006 seem less outstanding with the control of individual-level variables, the period of 1994–1998 presents itself as less similar to the subsequent period (the difference was partly conflated by changes in composition of candidates). It is worth noting that the intercept falls considerably, indicating that the introduced predictors explain volatility, more so than loyalty (most candidate-specific odds are higher than one). 71
Dummy variables indicating categories of managers and officials are based on declarations made by candidates themselves and they involve a wide variety of roles and power positions. 72 This is therefore a proxy measure of availability of administrative and financial resources. With this in mind, we can state that some kind of broadly understood power or influence made candidate-party commitments shakier. As was hypothesized, any relevant resource gives an individual a better bargaining position, thus intensifying mutual assortment. Volatility came also with experience (which can be also treated as a kind of asset), as those with more previous electoral races (lagged cumulative count) changed party colors more often (controlling for the adverse effect of age). It seems evident that practical strategies in the “game of seats” changed significantly according to individual characteristics of the players.
Differences between models II, III, and IV lie in the underlying assumption about the independence of observations. The standard “flat” regression (II) is actually biased as it treats all instances of switching or retaining affiliation as independent. In the long data format candidacies are actually “nested” within individual candidates. Model III takes this into account by treating the within-candidate part of the variance separately (as a random effect), hence checking whether the personal tendency of an individual candidate to switch affiliations interacts with the results. Model IV additionally accounts for the variance in affiliation switching by candidates running in SMDs (random slopes), but there seems not to be any similar “personal tendency” concerning district races. 73
Those corrections in fact make some of the odds ratios more pronounced, especially for candidates who competed in SMDs (in both majoritarian and mixed systems). The mode of running in the elections has been introduced as a series of interaction terms. Two consecutive election races in an SMD are characterized by the odds ratio of 2.41, which means that the odds of switching for those who entered a race on a party-list at both times were exceptionally low. The ratio is similar (2.46) for candidates who ran in an SMD at first, to later compete in the PR system. It is even higher for candidates who moved from party-list to district race (2.97). Changing banners in many cases coincided with a change of electoral strategy. Nonetheless, running in an SMD is an important factor in party switching—more so than demographics, electoral experience, or occupational position.
MPs were more prone to switching affiliation, but only slightly—in most of the models, the effect is insignificant. Past electoral success undoubtedly serves as an additional advantage in the bargain between candidate and party, but there is also a counterbalance—becoming a high-ranking party member often induces stronger attachment (the insider bargain has different rules). We can also turn this induction around and ask whether switching leads to better chances of success. The answer is negative—the odds of switching for those who actually won the mandate at the end of some period were lower than 1 (models II-VI). Party-switching might be in some cases a strategy to stay afloat—especially with a fast-changing political scene, as in Ukraine—but in general the relation between changing parties and becoming an MP is slightly negative. In the period of 1990–2014, there were relatively more instances of party loyalty awarded by voters than punished by them.
The purpose of models V and VI is to further validate these findings. In model V, we have taken into account only those periods that ended with mixed elections. This is to show that purely majoritarian and purely proportional systems did not influence our findings. As the odds ratios deviate little from those in model III, there is more certainty that the results are not the outcome of mere changes in electoral laws (which at times might have induced or restrained affiliation changes). In the case of SMD races, the effects are even stronger. Model VI involves only MPs running for reelection. Its aim is to ascertain that the results are not based mainly on unsuccessful candidates, who constitute most of the population under study. This model also corroborates other results. It only deviates from the previous ones by showing much more fluidity among MPs in the 2014 elections (as a countermeasure to the wide reconfiguration of the party system).
Model VII shows the overall intensity of affiliation changing and retaining by candidates of the main affiliation categories. Independent candidates were relatively more prone to switching, but the difference between them and partisan candidates is unexpectedly low. The corresponding coefficient is an interaction term (as all independents ran in SMDs), but it is still lower than for some blocs, like the Bloc of Yulia Tymoshenko (BYuT) or Our Ukraine. As expected, there was much less movement towards becoming an independent candidate (lead variable). In the case of the CPU, we see a great out-migration of members while there were very few candidates who changed to CPU affiliation. Generally, all those political entities that were already past their heyday experienced a higher outflow than inflow. In contrast, coefficients for UDAR (2012) and Poroshenko Bloc (2014) show smaller outflow and an above-average inflow (vitality symptoms for a party). With the control of the main affiliations, the effect of electoral rules holds.
There is one more check that can be made for a similar purpose. In the 1998 elections, candidates were allowed to run in districts and on party-lists simultaneously, which offers an interesting reference measure. As shown in Table 3, in this election, candidates that ran solely in SMD more often changed affiliation than retained it. In the case of party-lists, it was the other way around. For candidates using both modes, sticking with their affiliation was also prevalent, but the ratio is smaller, putting this group between the two aforementioned. There is also less dropping out, which can be explained by the fact that in general parties allowed a “double run” especially for their most prominent members (“electoral locomotives”). There seems to be no reason to doubt the strong effect of local district races on weaker party attachment in Ukraine during the past twenty-five years—and this is regardless of whether a majoritarian or a mixed system was in place.
Party attachment in 1998 elections by the mode of running
Findings
There is no clear indication of a specific influence of the Ukrainian revolutions on party attachment or heightened dropping out of politics. There is even little evidence for more intensive circulation in the VR ranks. Reelection chances in 2006 (36 percent) and in 2014 (33 percent) were higher than in 1994 (13 percent), and similar to those in 1998 (33 percent). To put it in a broader perspective: The average fraction of reelected MPs in the chamber was 28 percent during 1994–2002, and rose to 44 percent in 2006–2014 (36 percent when the 2007 elections are excluded). This means lower replacement rates in more politically turbulent times. Mass protests might have had some influence on the lower chances of retaining MP status in the PR vote, but the reverse was true for SMDs (it compensated for lower party system stability).
In turn, there was an evident effect connected with electoral rules. To check its robustness, we have presented as many as four ways of approaching the empirical question. First, we have shown that in the whole population of candidates during all the elections in the 1990–2014 period, those running in a majoritarian system had twice to three times higher odds of switching affiliation than those running in a PR system. In some cases, it might mean that running in SMDs makes a candidate more prone to reconsider party choices while in others a district race might be a convenient mode of changing banners. Substantially, both situations are equivalent—that is, SMDs enhanced politicians’ mobility. Second, we show that a very similar pattern is true for MPs, which shows that the rules of the electoral game had relevance irrespective of the “play” being carried out inside or outside the parliamentary chamber. Third, a more rigid comparison can be made using solely candidacies in elections with the mixed system in place—this way, we ascertain that the results are not simply contingent on the concurrence of voting rules and certain historical events and political developments. In fact, when we took only those cases into account, the pattern was even more pronounced. Finally, to briefly put the variable historical context aside, we have looked at the 1998 data in isolation and recorded that even in the instance of parallel SMD and party-list running (as was allowed at that time), the probability of switching was apparently higher when a candidate competed in the local district. The results of all four analyses point in the same direction. There can be hardly any doubt that in post-Soviet Ukraine, politicians running in SMDs were more prone to switching affiliations.
We hypothesized that affiliation volatility stems from positions in a bargain between politicians and parties, and from a structure of incentives created by an institutional environment which includes—among other things—voting system, party system, and rules of formal and informal distribution of political resources and rewards. In particular we expected that those factors interact, creating situation in which individual candidates have a relatively high degree of agency in processes of mutual selection. We may look for further clues on whether such interpretation is valid. There are two variables available in the models that can serve as indicators of the favorable position of a politician in relation to political organizations (such as parties, blocs, or committees). One is an MP mandate, the second is some other official position (a proxy measure for “administrative resources”). In the light of the “incentive model” laid out at the beginning of the article, we should expect relatively heightened party-switching in both categories—and indeed this is what we have observed. In the case of mandate holders there was a positive net effect that occurred despite obvious confounding factors, like stronger party attachment among its core leaders. When only the mixed elections were taken into account or when major political affiliations were included in the model, the effect was even more pronounced. As for the administrative resources, we also have some clues on whether they amplify the majoritarian voting effect on undermining party loyalty. Adding the 3-way interaction between officials and consecutive SMD races (start and lead) to the model III in Table 3 provides an additional odds ratio of 3.46 (while leaving it for officials in subsequent PR races at 1.21). It shows that candidates who were officials and were running in a local district had the highest probability of switching parties. Other “resources” working in conjunction with majoritarian voting, especially the ability to muster financial support, seemingly had an analogous effect: There was a similarly significant interaction with managerial positions. Just as expected, personal availability of resources made candidates in a plurality vote more capable of being political “choosers.”
Discussion
In the article, we posed empirical questions concerning the impact of revolutions and institutional rules on party switching in Ukrainian parliamentary elections. The analysis showed that post-revolution elections have not been characterized by widened flows of candidates between parties. It was only the MPs who were more exposed to political transfers; therefore, it can be said that circulation of the political class was relatively more intensified only among the narrow parliamentary elite (who were more inclined to try to avoid voters’ “punishment” directed at certain political forces). It seems plausible that any direct effect on switching connected with revolutionary events was nullified by the fact that attachment of rank-and-file party members became relatively strong in times of intensified political conflict, yet the gross effects of revolutions on reshuffling in the political class seems marginal.
Unlike in some other countries studied in similar respects, 74 in Ukraine we found voting rules to be important factors influencing the political system, the party scene in particular. However, the fact that we observed strong effects caused by the electoral system does not readily position our findings in the debate between institutionalism and cultural modernization theory. 75 The study on party switching in Brazil led Desposato to observe: “We have overemphasized the exogenous power of electoral rules and underspecified the power of informal institutions.” 76 Though our findings for Ukraine differ in many respects, this observation has not been proven incorrect as the results indeed highlight an interplay of certain factors: electoral system, party scene, and structure of power and influence. The practical mechanisms of the electoral “game of seats” were shaped not only by formal rules but also according to the general balance in the complex bargain between politicians, parties, interest groups in constituencies, and other influential stakeholders. When important resources in that game— access to administrative power, personal support, donors, etc.—are available to independent politicians, the advantage provided by the majoritarian system becomes strongly pronounced. In Ukraine, where nationalization of the party system has been weak and the structure of power and wealth has exerted a relatively high political impact, single candidates have been able to gain a strong bargaining position in relation to both parties and constituencies. The patterns we found inscribed in the data on parliamentary elections seem to fit that interpretation well. This observation is also coherent with the findings on Ukrainian MPs, who changed parties relatively often and mostly to join the “party of power” or to secure reelection. 77
Despite the clear effect of majoritarian and proportional rules observed on the party system, this impact cannot be deemed exogenous to the political system as a whole. As Kenneth Benoit argues, electoral rules are in fact at least partly endogenous to the system, 78 which finds a relevant illustration in the unresolved Ukrainian debate on open list PR voting. There is also a clear pattern concerning electoral systems and an authoritarian trajectory in the transition of post-Soviet states. 79 All this has implications for the classical accountability argument. In the Ukrainian case, it was the PR vote that led to a more explicit alteration of power and there was no indication that Duverger’s Law was “in force.” 80 This is an important observation as the accountability thesis treats the Law as an assumption. Second, the weakly nationalized Ukrainian party scene illustrates the contradiction inscribed in the problem. While the argument about constituency accountability is better at conveying regional representation and therefore a more complex structure of representation, Duverger’s Law involves an argument about the simplicity of the latter in a two-party system. When there is a clear trade-off between certain policies in benefits to a particular constituency and the rest of the country (as often happens in deeply divided societies), then the substance of accountability becomes contradictory. The third implication conveys the issue of candidate-party bargain analyzed here. Popper argued that the relative empowerment of individual political actors enables them to bring changes to a party. 81 However, “changing party” is equivocal—instead of exerting influence from within, it might mean switching altogether. In the case of Ukraine, the balance has been tilted toward individual politicians’ choices, which led to relatively weak parties and to an unstable party system, seemingly creating a vicious circle.
