When Are Gender Quotas Fulfilled? Party Strategy and Historical Memory in Ukrainian City Elections

Abstract

Many studies have investigated why countries adopt gender quotas for their elections. In this article, we answer a different question: why do political parties comply with gender quotas when the costs of noncompliance are absent or minimal? To answer this question, we analyze data from 1,600 party lists and 106 parties competing across 121 cities in the 2015 municipal elections in Ukraine. Our subnational approach tests whether contextual factors flagged by the broader gender literature explain variation in compliance across localities. The results of our models support our contention that Ukrainian political parties behaved strategically in terms of nominations and quota compliance. We find that urbanization and female incumbency fueled quota compliance. Parties, however, were less likely to comply with quotas in cities with more Ukrainian speakers. We suggest that the politics of memory explain this outcome, as Ukrainian speakers are more likely to remember of the costs of Soviet rule.

On October 25, 2015, Ukrainian voters went to the polls to elect city councils in 121 major cities. ¹ For the first time in any election in the postindependence period, Ukraine employed a gender quota for all party lists. The Law on Local Elections, which was passed by parliament on July 14, 2015, established a 30% gender quota for all party lists. The original legislation, however, did not contain any penalty for not adhering to the quota, and the Central Election Commission (CEC) waited until September 23, 2015, to rule that parties that failed to meet the quota would remain eligible to participate in the election. While the CEC decision was appealed through the courts, its decision was ultimately upheld. Consequently, parties knew prior to the election that adherence to the gender quota was voluntary. Yet, when you examine the lists of the 106 Ukrainian political parties who competed in the city elections, you find a striking result—quite a few parties complied with the quotas even without official sanctions. Of the total 1,630 lists, parties met the 30% quota 79.8% of the time. In fact, the average percentage of women in city legislators increased to 26.8% from 21% elected in 2010.

Research shows that parties are often unwilling to comply with quotas (Bjarnegård and Zetterberg 2016; Hinojosa 2012). Thus, quota laws often contain punishments for parties that fail to abide by them. Research finds that quotas with stronger enforcement mechanisms are more likely to be followed than others (e.g., Davidson-Schmich 2016; Schwindt-Bayer 2009; Baldez 2002; Htun and Jones 2002; Murray 2007). Yet, empirical studies of the effects of quotas on women’s representation find that quotas with no punishments or weak ones are still correlated with increases in women’s representation (Schwindt-Bayer 2009). The fact that quotas appear to increase women’s representation even if they are weakly enforced raises the question—why do parties abide by quotas that impose no formal costs for noncompliance? ²

Existing research suggests several explanations for this empirical pattern. Multiple studies have found that party characteristics matter, such as the commitment to gender equality (Kittilson 2006; Krook 2009; Murray, Krook, and Opello 2012; Opello 2006), the centralization of candidate selection mechanisms (Bjarnegård and Zetterberg 2016), the desire by party leaders to increase control over nominations (Weeks 2018), or the level of party institutionalization (Wiley 2018). Other research argues that parties abide by quotas to attract voters (e.g, Davidson-Schmich 2006; Kittilson 2006; Meier 2004; Murray, Krook, and Opello 2012). If parties perceive an electoral advantage to nominating female candidates, then it is likely that parties will have an incentive to abide by quota requirements to increase their share of female voters and other voters interested in gender equality.

The electoral logic behind quota compliance, we suggest, is complicated. We argue that variation in compliance for weakly enforced quotas is driven by two main factors: party ideology and characteristics of electoral districts. Parties with an ideological commitment are more likely to comply with weak quotas. In addition, we argue that the variation in preferences among voters implies that not all voters will respond to quota compliance similarly. The different possible voter reactions to female candidates suggest that parties may have an incentive to gauge their compliance based on the political and economic context of districts in which they compete for votes.

To test our arguments, we examine the party lists of all 106 parties that competed in the 2015 city elections in Ukraine. Ukraine provides us with several advantages. First, it is a good example of a case in which a weak quota existed; however, there is strong evidence that parties met the quota in multiple electoral districts. Second, there is significant regional variation in support for women’s representation tied to historical memories of the Soviet Union. We expect greater compliance with quotas in localities with more Russian-speaking voters, and less support in localities with more Ukrainian-speaking voters. Russian speakers are more likely to hold positive views of the Soviet past, while Ukrainian speakers are more likely to remember the costs of Soviet rule. As we explain in more detail later in the text, the divergent attitudes of Ukrainian and Russian speakers with respect to the Soviet past are a product of the different lived experiences of Ukrainians in western, central, southern, and eastern Ukraine. Several recent works on historical memory have used events in the Soviet period—including deaths during the1932–34 Ukrainian famine or Holodomor (Rozenas and Zhukov 2019) and the indiscriminate repression in localities in western Ukraine in the 1940s and 1950s (Rozenas, Schutte, and Zhukov 2017) to predict current electoral behavior in Ukraine.

Our statistical analysis finds that parties acted strategically. Parties placed more women on lists in localities with a higher percentage of female deputies elected in 2010, and fewer women on lists in localities with more Ukrainian-speaking voters. In terms of quota compliance, we find again that parties were more likely to meet the quota in cities where more women were elected in 2010, and less likely to meet it in localities with larger Ukrainian-speaking populations. We find no evidence that party ideology independently affected the number of women on lists or quota compliance. These findings support our argument that parties acted strategically when nominating candidates and making decisions about quota compliance. They also support our argument that the electoral chances of female candidates and support for gender quotas vary across localities in Ukraine. Political parties anticipated the differential support for female candidates across Ukraine when making decisions about adding women to lists and meeting the gender quota. We submit that parties looked at past elections, the ethnolinguistic composition of the electorate, and the different memories of quotas and the Soviet past across Ukraine’s regions when making these strategic decisions.

This article makes four contributions to the literature. First, we analyze quota fulfillment at the subnational level. As stated by Davidson-Schmich (2016, 212), “One reason that the implementation of party quotas is understudied is that much of the research on women’s representation is focused on the national level.” Second, this research adds to our understanding of women’s representation in postcommunist Ukraine by looking beyond the national level to the possible causes of regional differences in women’s representation at the local level. Third, it extends our understanding of the strategic behavior of political parties in terms of quota compliance. While there is a significant amount of research on quota adoption, quota compliance is less studied. Fourth, we apply the politics of memory to study quota compliance. While multiple scholars have studied the effect of Soviet legacies on gender politics in Central and Eastern Europe (Dahlerup and Gaber 2017; Horn 2008; Kapoor 2016), we analyze how the Soviet legacy plays out differently across Ukraine’s regions and ethnic groups.

The article proceeds as follows. First, we review the literature on quota compliance to develop our theory about how context affects the politics of quota compliance. Second, we discuss gender and electoral politics in Ukraine. In this section, we explain the specifics of the Ukrainian case and develop hypotheses that are in line with our theoretical expectations. Third, we present a quantitative analysis of the 2015 Ukrainian local elections. Our empirical results demonstrate that Ukrainian parties responded strategically to the context of the cities that they competed in when deciding on nomination strategies and quota compliance.

Understanding Quota Compliance

Broadly, there are three major types of gender quotas—reserved seat quotas, legislative quotas, and party quotas (Krook 2007). Reserved seat quotas specify that a certain percentage of legislative seats must be held by female legislators. Much more common are compulsory party quotas that require political parties to nominate a certain percentage of female candidates (Tripp and Kang 2008). Such quotas are adopted by states and apply to all political parties. Finally, parties may voluntarily adopt quotas that require them to nominate slates that contain a certain percentage of women.

There is ample evidence that quotas increase women’s representation. Yet, parties often fail to comply with compulsory party quotas, even when there are costs for noncompliance, and with voluntary party quotas, which the parties adopt themselves (Bjarnegård and Zetterberg 2016; Hinojosa 2012). For parties, quota compliance presents a challenge. Parties are often wary of complying with quotas since meeting them challenges the existing power structure that favors men (Bjarnegård and Zetterberg 2016). After all, quotas require nominating more women, which entails replacing male candidates with female ones (Murray 2007). Since quotas can destabilize the status quo, it is not a surprise that parties will, at times, resist them.

The reluctance of parties to comply with quotas does raise the question, why do parties decide to comply with quotas? Murray (2007) suggests three arguments to explain why parties implement quotas. First, the ideological approach argues that some parties have a greater ideological affinity for quotas than do others. Three factors push parties to ignore other factors and embrace quotas—whether parties display ideological support for the goal of gender equality, whether they agree that state intervention is necessary to solve the problem, and whether the party’s ideological support is strong enough to overcome obstacles to quotas (Murray 2007). Current research supports this argument, showing that parties with a commitment to gender equity are more like to adopt quotas than are other parties (Kittilson 2006; Krook 2009; Murray, Krook, and Opello 2012; Opello 2006). Davidson-Schmich (2016) argues, however, that the adoption of quotas by parties across the ideological spectrum in the German case suggests that ideology alone is not sufficient to explain party quota adoption in all contexts.

Second, the institutional approach suggests that variation in quota compliance is explained by variation in institutions, such as electoral systems, party systems, or the quota legislation. Bjarnegård and Zetterberg (2016) show that more bureaucratized candidate mechanisms increase compliance versus less formal ones. ³ Htun and Jones (2002) find that party quotas are more effective in closed-list proportional representation (PR) systems than they are in open-list PR systems; however, other research demonstrates that quotas in open-list PR systems are effective (Matland 2006; Schmidt and Saunders 2004). Wylie (2018) argues that weakly institutionalized parties lack the resources to support female candidates and that this explains, in part, why Brazilian parties failed to comply with gender quotas.

Another factor is the compliance mechanism. There is a significant amount of research demonstrating that quotas with effective compliance requirements produce better results (Baldez 2002; Davidson-Schmich 2006, 2016; Htun and Jones 2002; Schwindt-Bayer 2009). Murray (2007) argues that the funding penalties for quota non-fulfillment in France push smaller parties, who cannot afford losses in party funds, to abide by the quota, but not larger parties, which can afford the losses. Davidson-Schmich (2016) notes that while the voluntary party quotas enacted by German parties are written to be legally binding, the quota rules of many parties contained loopholes that allowed them to avoid fulfilling them under certain circumstances.

Finally, the electoral competition approach argues that quota compliance is the result of strategic behavior by parties seeking to maximize votes. Assuming that parties are rational actors whose primary preference is winning seats, parties will nominate candidates who are best able to secure votes, ignoring ideological or other factors. If parties assume that voters are less likely to vote for female than male candidates, then parties will discriminate against female candidates. The literature on electoral systems, for example, finds that women are less likely to be elected in systems that create personal vote incentives in comparison with more party-centric systems (Murray 2007; Rule 1987; Thames and Williams 2010). This finding is explained, in part, by the concerns of party leaders that voters prefer male candidates. Thus, nominating female candidates risks losing vote share. ⁴

However, electoral competition may create incentives for nominating female candidates. One of the main explanations for the adoption of gender quotas assumes that parties use quotas to capture female votes (Davidson-Schmich 2006; Kittilson 2006; Meier 2004; Murray, Krook, and Opello 2012). The need to remain competitive among female voters pushes parties to support quotas when their opponents do (Davidson-Schmich 2006; Giger et al. 2014; Matland and Studlar 1996; Wängnerud 2000). ⁵ Electoral competition is seen by some as a crucial incentive for male elites to adopt quotas (Baldez 2002; Fréchette, Maniquet, and Morelli 2008; Weeks 2018). Based on this logic, parties will fully implement quotas to the extent that it increases their vote share. Conversely, parties will avoid implementing quotas in situations in which it would lose them votes.

How voters respond to the gender of candidates depends on their ideology. There is evidence that voters who share more egalitarian attitudes as well as those who embrace left-wing ideologies are more likely to vote for female candidates (Caul 1999; Dolan 2008; Kittilson 2006; Salmond 2006). Religious and nationalist voters and parties, on the other hand, might be less open to female candidates. Regional and religious proxies for attitudes toward women appear to correlate with women’s representation in cross-national research (Tripp and Kang 2008). Inglehart and Norris (2003) find that less religious societies are more likely to hold attitudes that support gender equality and, by extension, female representation. National-level indicators, however, might mask variation within countries. Setzler (2015) argues that support for women’s representation in Latin America varies based on religious faith. Nationalism has also been shown to depress female representation by stressing male leadership traits (Banerjee 2007). ⁶ There is also research that finds less support for female candidates among voters on the far right (Erzeel and Caluwaerts 2015; Golder et al. 2017).

Political parties do matter. Badas and Stauffer (2019) find that gender affinity voting is more common in nonpartisan contexts than in partisan ones. This suggests that the likelihood that female, or even male, voters will support female candidates depends on partisanship. Dolan (2014) argues that voter perceptions of candidates are filtered by partisanship. While there is strong evidence that voters evaluate candidates partly based on abstract gender stereotypes that often lead them to devalue female candidates (e.g., Fox and Smith 1998; Huddy and Terkildsen 1993; Lawless 2004), the effect of these stereotypes is diminished because of the importance of partisan concerns.

For parties seeking to maximize votes through quota compliance, the trick is identifying when and where compliance increases votes. Valdini (2012) argues that party leaders are aware that voters may use gender as a cue, which may have negative connotations for some voters. What types of populations are more likely to support female candidates? Research finds correlations between district socioeconomic factors and support for female candidates. Districts and states that are wealthier, more urbanized, and feature less traditional political cultures are more likely to nominate and elect women (Diamond 1977; Fox 2000; Hill 1981; Nechemias 1987; Ondercin and Welch 2009). Thus, voters’ support for female candidates may vary based on local demographic factors.

Based on the foregoing discussion, we have two expectations concerning quota compliance. First, we expect that parties will behave strategically in terms of quota compliance. If the support for female legislators is not universal even among female voters, then parties have a strong incentive to behave strategically in terms of quota compliance. In systems in which the costs of compliance are high, it may not make sense for parties to comply with quotas, as the costs of doing so may outweigh the electoral benefits. In systems with weaker costs of noncompliance, parties have a strong incentive to uphold quotas in those areas where voters are more predisposed to support female candidates and not comply in areas where voters are, on average, less supportive of female candidates.

Second, we expect that the patterns of compliance will be heavily dependent upon context within countries. Support for female candidates should vary based on differences in key socioeconomic and demographic variables, not only between but, most importantly, within countries. In addition, compliance should vary based on important institutional differences such as electoral rules and compliance costs.

The Ukrainian Case

There is strong evidence of a patriarchal, traditional political culture in postcommunist Ukraine (e.g., Galligan and Clavero 2008; Hankivsky and Salnykova 2010; Hrycak 2005, 2006, 2010, 2011; LaFont 2001; Martseniuk 2016; Pyshulina 2016; Rubchak 2012; Thames 2018). For many, the role of women is limited to child rearing and home life. As a result, one can find only weak support for women as political leaders. A 2012 survey of Ukrainian adults found that 40% of respondents preferred male over female candidates and 77% believed that men were better suited for politics (Lake, Gotoff, and Pondel 2012). With the possible exception of figures such as Yuliia Tymoshenko—who was twice prime minister—women remain outside of traditional parties.

Yet, if you unpack the data, you find that the effect of gender on electoral outcomes in Ukraine is more complicated than it may seem. For example, research on elections to the national legislature—the Rada—finds that women make up an increasing number of legislators. While women constituted a mere 12.1% of members of parliament in 2014 (Martseniuk 2016, 236), that share increased to 20.8% following the 2019 elections. ⁷ The research on women competing in single-member district (SMD) elections to the Rada shows that voters do not discriminate against female candidates; however, parties are more likely to nominate women in PR elections than in SMD elections (Thames 2018). Thus, there is evidence that while discriminatory attitudes do exist, Ukrainian parties and voters will support female candidates.

In addition, Ukraine provides us an opportunity to analyze quota compliance at the local level, which remains understudied not just with respect to Ukraine but worldwide. Ukraine adopted a compulsory 30% party gender quota for the 2015 local elections. The quota was added to the Law on Local Elections, which passed in parliament on July 14, 2015, by an amendment proposed by Yurii Lutsenko, the leader of the pro-presidential Block of Petro Poroshenko (Blok Petra Poroshenka, BPP) (Antoniuk 2015, 145–47). While the law was signed by President Poroshenko, quota enforcement remained unsettled, as no costs were outlined in the legislation. On September 23, 2015, the CEC concluded that the failure to meet the quota did not prevent parties from registering party lists. On September 28, the decision was challenged in court with an appeal by deputies from the Self-Help (Samopomich) Party, which actively ran on the gender quota. The Kyiv Appellate Court sided with the Samopomich deputies and revoked the CEC decision. This decision, however, was further appealed to the High Administrative Court of Ukraine, which eventually upheld the CEC decision. ⁸ This meant that parties were not punished for failing to meet the quota. Consequently, while Ukraine adopted a compulsory party quota, the lack of enforcement mechanism meant that quota compliance was essentially voluntary. ⁹

Similar disputes also occurred at the regional level. On October 2, 2015, the Volyn’ Regional Electoral Commission denied the registration of lists from four parties that did not meet the gender quota. The candidates who were denied registration took their cases to the L’viv Appellate Court, which sided on behalf of the candidates whose lists were not registered. Oksana Yarosh, a political scientist from Volyn’, explained the lack of adherence to the gender quota in the following way: “as parties explained, the reason for not abiding by the gender quota (was) the absence of sanctions [tse vidsutnist’ sanktsii].” ¹⁰

Political actors were betting that courts at the regional or national level were unlikely to affirm the mandatory gender quota and block the registration of lists that failed to meet it. These court decisions should not be seen as a battle between political institutions. Ukrainian politicians understood which side was likely to win in the end. Ukraine’s judiciary is not independent, but rather “politicized and submissive” (Popova and Beers 2020, 137). In this light, it is not odd that that the party that proposed the quota in the first place—the pro-presidential BPP—only adhered to it for 65% of its lists and made certain that the courts allowed the CEC to register lists that did not meet the gender quota. President Poroshenko had the ability to change the makeup of the CEC, whose members had been installed by his predecessor, Viktor Yanukovych. In the end, Poroshenko decided to keep them, likely because they showed their lack of independence. Ukrainian courts have been especially subservient to the government when it comes to electoral law (Popova 2010, 1220–21).

Quota compliance can vary based on institutional factors such as the electoral system. In this case, all Ukrainian local elections utilized the same electoral system, allowing us to hold the electoral system constant across all cases. The electoral system employed to elect local governments in 2015 was new, and complicated, fundamentally changing the electoral rules that are used in provinces (oblasty), cities (mista), districts (raiony), and urban districts (raiony u mistakh). In Ukraine, the system is often referred to as open-list proportional representation. However, it differs fundamentally from other open-list PR systems. While the system awards seats proportionally to candidates who pass the electoral threshold, there is no intraparty competition for votes. ¹¹

Intraparty dynamics might explain why women are not placed on lists. In most cases, party leaders maintained significant control over nominations. Rahat and Hazan (2001) place the potential selectorates of parties along a continuum from exclusive to inclusive. Using this framework, Ukrainian political parties would likely be considered exclusive. When asked how parties formed their lists in the 2012 parliamentary elections, Viktor Chumak, of the Ukrainian Democratic Alliance for Reform, explained that the decisions were generally made by the “members of the Politboro of the parties and rather influential groups [vliiatel’nykh grupp] in the party.” ¹² If this is the case, men are likely overrepresented in the selectorates of most Ukrainian parties. In a forum on women’s representation organized by the Committee of Voters of Ukraine in the western Ukrainian city of Rivne in 2016, participants pointed out that increased female representation depended, beyond simply quotas, on parties themselves becoming more friendly (druzhnimy) toward women. ¹³ As these were local elections, local party members are likely to play an active role in candidate selection. We expect that these local party members broadly share the attitudes of a given locality with respect to the electoral viability of female candidates and the benefits of enforcing the gender quota. We argue that these attitudes are also shaped by the broader politics of memory.

Our theory claims that parties need to compete for votes; therefore, they will comply with quotas in cities where voters are more likely to support female candidates. On average, 79.8% of party lists in cities met the 30% gender quota. Why was compliance so high in this case? As indicated previously, there is evidence that quotas with weak or no enforcement often increase women’s representation (Schwindt-Bayer 2009). In addition, recent research on national elections in Ukraine finds little evidence that voters discriminate against female candidates (Thames 2018). Thus, it is likely that parties have less fear that nominating women will undermine their votes. It is also likely that uncertainty over the quota enforcement specifics created an incentive for parties to meet the threshold to avoid possible sanctions.

Yet, there was significant variation in the number of lists that met the quota by party. Figure 1 plots the percentage of city lists by major party where parties met the 30% gender quota. ¹⁴ The data presented in Figure 2 reveal significant variation in quota compliance among major Ukrainian parties in the election. The Unity Party, which competed in only four cities, met the quota on all its lists. Yet, Trust Deeds met the quota in only 50% of cities.

Figure 1. Percentage of lists meeting quota by major party.

Figure 2. Average percentage of party lists in city elections meeting the quota by region.

What could explain the results in Figure 1? As discussed previously, quota adoption and compliance can vary by party ideology. ¹⁵ It is not easy to place Ukrainian political parties on a traditional left-right economic continuum. We distinguish here between left and ethnonationalist parties. While most parties in Ukraine offer voters populist economic appeals, especially before elections, parties that we code as left call for increasing the role of the state in the economy and maintaining social guarantees inherited from the Soviet era. Ukrainian parties also differ in their attitudes toward the Soviet past, relations with Russia, and the role of ethnic Ukrainians and the Ukrainian language in the polity. This ethnopolitical continuum has been a constant feature of elections since independence, but it has arguably hardened in the aftermath of Russia’s annexation of Crimea and the war with Russian-backed separatists in eastern Ukraine. Parties that we code as ethnonationalist aim their messages at Ukrainian speakers and support ethnic Ukrainian culture, traditions, and family and religious values. Ethnonationalist parties played key roles in anti-regime mobilization that initiated and sustained both the Orange Revolution in 2004–05 and the Euro-Maidan protests in 2013–14. A small number of parties that voiced the group-based demands of Hungarian and Russian speakers were also coded as ethnonationalist, as the broader scholarly literature suggests that ethnonationalist parties are less supportive of female representation. Ethnonationalist parties that seek the votes of Russian speakers generally look more favorably on the Soviet past and welcome closer ties with Russia. The Opposition Block, which emerged out of the ashes of the Party of Regions—the governing party of former president Viktor Yanukovych—currently aims its ethnonationalist messages at Russian-speaking voters.

How would the ideological positioning of Ukrainian parties along this ethnopolitical continuum affect their nomination strategies vis-à-vis female candidates? Ethnonationalist parties are more socially conservative and support traditional family values and gender roles. Many of these parties have their roots in Rukh, the Ukrainian national independence movement. While women participated in organizations that fought against Soviet rule, these organizations were typically founded on nationalist and traditional family values, and they were generally opposed to women’s political equality (Hrycak 2006, 2010). Ethnonationalist parties’ support of traditional family values is reinforced by Ukraine’s various Orthodox and Uniate churches. Even when women are employed, they are expected to maintain the home and rear the children. Bolzendahl and Gracheva (2018) find that opposition to feminism and gay rights in Eastern Europe is much higher among individuals who strongly identify with their nation. These individuals view the imposition of these “Western” norms as a violation of their national sovereignty.

Some ethnonationalists go so far as to claim that the 2015 gender quota was alien to Ukraine and imposed from abroad. Mar’ian Krеkoten’, a deputy from the extreme-right Svoboda (Freedom) party on the city council in Ternopil’ in western Ukraine, claimed that one had to be a “European idiot” to support the quota and suggested that “positive discrimination” was nonetheless sex discrimination and prohibited by the Ukrainian constitution. ¹⁶ Ukrainian ethnonationalist parties traditionally reject feminism and have advocated for legislation that would punish those who spread “gay propaganda.” Some groups have gone further. Gays and lesbians in Ukraine have been routinely attacked by members of nationalist and extreme-right groups. The hostility against those participating in gay pride marches is so intense that the police have had to form human shields to separate gay rights supporters from extreme-right attackers (Hrybanov and Kravchuk 2018, 138–41). We hypothesize, therefore, that ethnonationalist parties are less likely to nominate female candidates and comply with quotas than are other parties.

Left-wing parties in Ukraine tend to be more statist and seek to maintain increase the role of the state in the economy, despite the already huge role played by the state in Ukraine’s semireformed economy. In the past, these voters often gravitated to socialist and communist parties. Russian speakers make up a major constituency of left parties. Not surprisingly, left parties also tend to look more favorably on the Soviet past, and call for the government to live up to Soviet-era “social guarantees.” We hypothesize, therefore, that left parties are more likely to nominate female candidates and comply with quotas than are other parties.

We believe, however, that differences in party ideology explain only part of the variation in quota compliance. Our theory emphasizes that contextual factors will also matter. In Ukraine, variation in party support, urbanization, and ethnicity exhibits a strong regional cleavage that defines Ukrainian politics (Bloom and Shulman 2011; Holovakha and Panina 1998; Katchanovski 2006; Khmelko and Wilson 1998; Wilson 1997, 2005). Differences in views from western to eastern Ukraine—including on controversial issues such as language use (Ukrainian or Russian) or NATO membership—are, at least in part, rooted in different historical experiences (Riabchuk 2003, 28–37). Western Ukraine was not incorporated into the Soviet Union until 1944, while the rest of Ukraine has deeper roots in the Russian Empire and the Soviet Union (Subtelny 1994, 487–95). Eastern and southern Ukraine were also more affected by Russification policies. In the industrial cities of the east and south, immigrants from Russia and Ukrainians from the countryside worked together in factories and sent their children to Russian-language schools, fulfilling in some sense the Communist Party’s goals—introduced in the 1970s—of the merging of nationalities and the creation of “Soviet man” (Lieven 1999, 50–52). Russification and Sovietization were less pronounced in central Ukraine and absent in western Ukraine. As a result, views of the Soviet past vary significantly across Ukraine’s regions (Wanner 1998, 171–99).

Inglehart and Norris (2003, 135–46) suggest that the relative stability of support across time for gender quotas in Eastern European countries can be explained by their prior experience of gender quotas in the communist era. We are intrigued by the notion that the legacy of gender quotas in the Soviet era could explain quota compliance in the 2015 election. We do not, however, believe that the effect of historical memory will be uniform across Ukraine’s cities. Given the stark differences in attitudes about the Soviet period across Ukraine’s regions, we expect more support for the gender quotas in localities with more Russian-speaking voters—who have more sympathetic views of the Soviet Union, and less support in localities with more Ukrainian-speaking voters—who hold more negative views of the Soviet past.

Some might object to using past historical experiences to predict current political behavior. Historical arguments, however, are experiencing a renaissance of sorts in comparative politics research (Mares and Mohammed 2019; Pop-Elecher and Tucker 2017). Two articles by Arturas Rozenas and his colleagues use events in Soviet Ukraine—including the 1932–34 Ukrainian famine or Holodomor (Rozenas and Zhukov 2019) and repression in localities in western Ukraine in the 1940s and 1950s (Rozenas, Schutte, and Zhukov 2017)—to predict political mobilization and electoral behavior during the late Soviet and postindependence periods. They find that localities that experienced more deaths during the Holodomor were less likely to vote for pro-Russian parties in elections from 2002 to 2014 (Rozenas and Zhukov 2019, 576) and that the electoral margin for pro-Russian parties was lower in districts that experienced more Soviet-era deportations (Rozenas, Schutte, and Zhukov 2019, 1153–54).

For any politics of memory argument, it is important to explain how historical legacies are transmitted across generations. Rozenas and Zhuvkov (2019, 570–71) discuss three possible modes of transmission. First, parents share their views of events and ideology with their children. Second, children learn about events in schools and churches. Third, individuals construct shared norms though their interactions with neighbors and peers. Many Ukrainians know about relatives that died during the Holodomor or were deported to Siberia. While this topic was taboo during the Soviet period, it is now openly discussed in families. The crimes of the Soviet Union are also part of the school curriculum in Ukrainian schools (Kosianov 2014), although teachers in Russian-language schools in eastern Ukraine have been found to deflect criticism of Russia (Rodgers 2007). Finally, Ukrainians undoubtedly learn from their daily interactions with peers. A key source of information, beyond social interactions, is the media. Ukrainian-language media is more likely to provide a negative view of today’s Russia and the Soviet past than Russian-language media. ¹⁷

Table 1 shows characteristics of Ukraine’s regions that might influence parties’ decisions about whether to fulfill the gender quota in specific localities, including several that have been flagged by the broader gender literature as factors affecting female representation. Western and central Ukraine are more ethnically and linguistically homogenous than southern and eastern Ukraine. The differences between the percentages of ethnic Ukrainian and Ukrainian speakers capture the effects of Russification in the east and south. Inhabitants of western Ukraine are the most religious, followed by those in the center. Measures of religiosity are lower in the south and east. Those living in the west and center are also more likely to support Christian democratic, national democratic, and nationalist parties. Support for these parties is lower in the south and totally absent in the east.

Table 1. Ukrainian regional differences (percentages)

Sources: Osoblyvosti relihiinoho (2018, 12, 19); Zlobina et al. (2017, 92); Ukrainian State Statistical Committee; Ukrainian Central Electoral Commission.

As previously noted, there is significant prior research that leads us to expect that support for female candidates will be lower in those cities with greater support for traditional political cultures. Unfortunately, we lack disaggregated survey data for cities to create a measure that directly captures differences in attitudes. Since polling data are not available at the level of cities, we rely on the percentage of the population that speaks Ukrainian across cities to measure traditional popular culture. ¹⁸ The Ukrainian-speaking population tracks fairly closely with levels of religiosity and support for Christian democratic, national democratic, and nationalist parties across cities, as well as attitudes toward the Soviet past. Residents of the south, and especially the east, are more likely to support communist parties. Consequently, we hypothesize that as the percentage of the Ukrainian-speaking population increases in a city, the percentage of women nominated on party lists and quota compliance will decrease.

Given the strong east-west/left-right cleavage in Ukraine, it is not surprising that we see variation in quota compliance regionally. Figure 2 plots a heat map of the average percentage of party lists in cities that met the 30% gender quota by state (oblast’). The data show a clear trend in quota compliance among oblasts in eastern and southern Ukraine, in comparison with those in central and especially western Ukraine. ¹⁹

Finally, for parties seeking to maximize votes through the nomination of female candidates, the best indicator of future support might be past success. Consequently, we expect parties to consider the results of the previous election, paying particular attention to the success of female candidates. In this case, parties may find that nominating women in cities that saw past success for female candidates will allow them to meet the quota and not lose vote share. Thus, we expect that the percentage of female candidates on city lists and quota compliance will increase as the percentage of elected female deputies in the 2010 election increases.

The remainder of this article will test our hypotheses.

Data and Methods

To test our hypotheses, we created a data set of all 106 parties that competed in the 121 cities of oblast significance in the 2015 Ukrainian elections. ²⁰ Our data set totals 1,641 observations, and the unit of analysis is the party list/city. ²¹

We test our hypotheses with two dependent variables calculated using our sample of the 2015 electoral data. First, we calculate for each party list in each city the percentage of women on the list. Second, we create a dummy variable, met quota, that is coded 1 for all parties that complied with the 30% quota on the party list in the city and 0 for all those parties that did not meet the 30% quota. We employ two dependent variables to more fully understand party nomination strategies. Given that the percentage of women on the list variable is continuous, it contains more information than the simple met quota dummy variable does. Moreover, if our intuition is correct, then we should see the expected relationships between both the percentage of women on the list variable as well as the met quota variable.

We employ four independent variables to test our hypotheses. First, we include two dummy variables to test the impact of party ideology, based on the Ukrainian context—left party and ethnonationalist party. Parties are coded as left if their platforms or public statements focused on policies that would increase the role of the state in the economy and maintain social guarantees from the Soviet period. Parties are coded as ethnonationalist if their platforms or public statements were aimed primarily at co-ethnics and supported the culture and traditions of their group. The overwhelming majority of these parties targeted Ukrainian speakers, although some parties representing the interests of Russian and Hungarian speakers were also coded as ethnonationalist based on their platforms and public statements. A list of the 106 parties and their coding can be found in the supplementary materials online. ²² Second, we include a variable that codes each city by its percentage of Ukrainian-speaking population. ²³ Finally, we calculate the percentage of female deputies in 2010 for each city and code all observations with its value. ²⁴

All our models include several control variables. We control for differing economic conditions by including the percentage unemployment for all cities in our data set. As indicated in Table 1, unemployment was higher in eastern than western Ukraine. ²⁵ We include the percentage of party deputies in 2010 for each party. This number is simply the percentage of each city legislature elected by party in the 2010 elections. We include this measure to control for the possibility that differences in overall party support will explain variation in the number of female candidates nominated by parties. We want to control for potential differences in the behavior of larger parties versus smaller parties. To do this, we include a variable measuring the percentage of cities in which parties ran lists. We also include the percentage of the vote that each party received in the 2014 PR election for the Ukrainian parliament, the Rada (percentage Rada vote 2014). It is possible that parties with strong national support may behave differently than those with only limited national support. Parties that did not compete in the 2014 Rada elections were coded as 0% of the vote. We include the log of party age to control for differences in party institutionalization. Women’s representation is often greater in urban areas; thus, we include the urbanization percentage, which is the percentage of the urban population for each city as measured in the 2001 Ukrainian census. ²⁶ Some research finds a positive correlation between women’s representation and the number of legislative seats (e.g., Eder, Fortin-Rittberger, and Kroeber 2016; Kjaer and Elklit 2014). Other scholars, however, find mixed results (Salmond 2006). We code each city with its legislature’s number of seats to control for the potential effect of legislative size.

We use non-nested Bayesian hierarchical models estimated with PyStan (Stan Development Team 2018). We include random intercepts for both individual political parties and individual cities in all equations. The random intercepts allow us to control for potential error variances based on different cities and parties. Thus, we can control for heterogeneity based on different cities and parties. In addition, several of our independent variables are city or party level indicators, consequently it is appropriate to include random intercepts at the group level. For our models using the percentage of women on the list dependent variable, we use a linear regression model. We use a logit specification for our models using the met quota variable. ²⁷

Results

Figure 3 plots the posterior medians and 90% credibility intervals of the independent variable first differences of Model 1 using the percentage women on the list dependent variable. ²⁸ For dummy variables, the first difference represents the difference in the expected percentage of women on the list of changing the value of the variable from 0 to 1 while holding all other variables at their means and modes. For continuous variables, the first differences represent an increase in the value of the variable from one standard deviation below the mean to one standard deviation above it while holding the value of all other variables at their means and modes. The results presented in Figure 3 provide mixed support for our arguments. In our first two hypotheses, we predicted that the party ideology dummy variables, ethnonationalist party and left party, would impact the percentage of women on party lists. The results, however, show no statistically significant impact of party ideology on the number of women on party lists in the 2015 elections. Party ideology, therefore, is not a good predictor of the percentage of women on party lists.

Figure 3. First differences, Model 1 (DV = percentage women on the list).

We do, however, find support for our other two hypotheses. Increasing the percentage Ukrainian-speaking population variable from one standard deviation below to one standard deviation of the mean reduces the percentage of women on the list by 1.2%. Thus, our expectations concerning the effect Ukraine’s historical memory are supported. In addition, we find evidence that a city’s previous level of support for female legislators matters. Increasing the percentage of female deputies elected in 2010 from one standard deviation below to one standard deviation above its mean increases women’s representation on lists by 2.5%. This result also supports our hypothesis.

Our control variables provided a few interesting results. Increasing the percentage urban population from one standard deviation below to one standard deviation above their means increased the percentage of female candidate on party lists in cities by a modest 1%. Increasing the percentage unemployment from one standard deviation below its mean to one above increased the percentage of women on the list on average by 1.2%. Given that unemployment was higher in eastern and southern Ukraine compared to western Ukraine, this is, perhaps, not surprising. Changes in the log of party age, percentage of cities, percentage of party deputies elected in 2010 do not have an independent effect on the percentage of women nominated on party lists. A similar increase in number of seats variables decreased the percentage of women on party lists in cities by 1.7%. Thus, cities with a greater number of seats had fewer women on party lists, on average.

What about quota compliance? Figure 4 presents the posterior median average marginal effects and 90% credibility intervals for each of independent variables in Model 2 using the met quota dependent variable. We calculate the effects using changes in the values of the variables as we did for the results in Figure 3. Once again, we find no evidence that party ideology alone explains the likelihood of a party meeting the quota. Both the ethnonationalist party and left party variables are not statistically significant in Model 2. Thus, we find more evidence that party ideology does not have an independent effect on meeting the 2015 quota.

Figure 4. First differences, Model 2 (DV = met quota).

Yet, we find strong support for our two contextual variables—the percentage of Ukrainian-speaking population and the percentage of female deputies in 2010. We find a robust effect of increasing the percentage Ukrainian-speaking population variable, which results in a 5.1% decrease in the probability of meeting the quota. This is an important finding because it demonstrates parties responding to differences in context, regardless of party ideology. In addition, increasing the percentage of female deputies in 2010 variable improved the probability of complying with the gender quota by 5.4%. As we found in Model 1, Model 2 provides strong support for our final two hypotheses.

The impacts of the control variables are less consistent in Model 2. None of the control variables are statistically significant; therefore, they have little independent effect on whether parties met the 2015 quota law or not.

Discussion of Results

Taken together, the results of models using our two dependent variables, the percentage of women on the list and the met quota, provide support for our contention that Ukrainian political parties behaved strategically in terms of women’s overall nominations and quota compliance. On the one hand, we found little evidence that party ideology alone explains whether parties met the weakly enforced party quota. Model 1 found some evidence that parties that ran lists in a greater percentage of cities nominate fewer women; however, we did not find similar results in Model 2 using the met quota variable. The fact that we included measures of various party characteristics and included random intercepts for individual political parties strongly suggests that we controlled for party effects; however, the results imply that these characteristics did not independently explain variation in quota compliance. Thus, we found no support for our first two hypotheses regarding ideology.

On the other hand, we have strong evidence that variation in local context impacted party decisions. Our argument assumes that parties want to win votes and will behave strategically to get votes to maximize their seat share. Thus, when faced with a weakly enforced quota, parties will vary their compliance, placing a greater number of female candidates in those cities where support for female candidates is greater. Based on the politics of memory in Ukraine, we expected that parties would be less likely to meet the quota in those cities with greater proportions of Ukrainian speakers. The results of both our models support this contention. Parties clearly placed more female candidates and were more likely to reach the 30% quota in cities with fewer Ukrainian speakers. In addition, we found evidence that parties were more likely to nominate women and meet the quota in cities that elected more women in the 2010 election. This supports our contention that women’s nominations and quota compliance were greater in those cities where parties had evidence that voters rewarded female candidates. In those cities where fewer women won seats in 2010, parties were less likely to meet the quota, nominating fewer women.

Conclusion

In 2015, Ukraine enacted a 30% gender quota for local elections. The quota, however, was effectively costless since the legislation contained no enforcement mechanism or penalties for failing to meet the quota. Even with a weak quota, Ukraine not only increased the average percentage of women elected to city legislatures since the last election, to 26.8% from 21%, but 79.8% of the time, parties met the quota. Perhaps this finding is not unexpected, given that the literature finds that even weak quotas increase women’s representation (Schwindt-Bayer 2009); however, it does raise the question, why do parties comply with weakly enforced quotas?

Our review of the broader literature on quota compliance formed the basis for our expectations that parties would behave strategically, and that quota compliance would be affected by contextual factors as well as party ideology. Our statistical analysis provided mixed support for our expectations. We found little evidence that party ideology independently impacted the number of women parties nominated and quota compliance. Yet, we found strong evidence that contextual factors mattered. Quota compliance varied not only with the percentage of Ukrainian speakers in a city, but also with the percentage of women elected in the 2010 local elections.

For scholars of Ukraine and post-Soviet states more broadly, these results provide evidence that historical memory of the Soviet Union can influence more recent political outcomes. The Soviet Union created fundamentally different legacies across Ukraine. Soviet rule left inhabitants of eastern and southern Ukraine more Sovietized than Ukrainians living in central and western Ukraine, where more aspects of traditional culture—including religion and patriarchy—survived the Soviet experiment. As other studies have found, views of the Soviet past are passed down through generations. Political parties, we suggest, anticipated that voters in the east and south would be more open to the gender quota, and female candidates more generally, than voters in the center and west. This finding is line with recent research (Rozenas, Schutte, and Zhukov 2017; Rozenas and Zhukov 2019) that also demonstrates how the Soviet experience continues to affect contemporary Ukraine.

With respect to future research, Ukraine’s adoption of a new, national-level electoral law in 2019 with a 40% gender quota raises the question, will we find similar patterns of quota compliance at the national level that we have seen at the city level? Davidson-Schmich (2016) correctly points out that much of the quota research focuses only on the national level. This article has sought to help rectify this issue. Nonetheless, the adoption of a national level quota in Ukraine provides research opportunities to determine whether national and regional patterns of compliance are similar or different.

The findings also add to our understanding of the effects of gender quotas more broadly. Research on gender quotas has found that enforcement provisions explain, in part, the effectiveness of quotas (e.g., Baldez 2002; Davidson-Schmich 2016; Htun and Jones 2002; Murray 2007; Schwindt-Bayer 2009). The research presented here expands our understanding of quotas with weak enforcement mechanisms. The evidence from the 2015 Ukrainian local elections is clear—parties did not simply ignore the quota. Instead, parties varied their quota compliance based on the characteristics of electoral districts. Parties were willing to nominate more women and to meet the weakly enforced quota rules in those cities that were more likely to support female candidates. This is clear evidence that parties behave strategically when faced with weakly enforced gender quotas. More importantly, the findings explain conditions under which parties will abide by quota regulations even when they face little to no punishment for ignoring them. For scholars of gender quotas, this is important because it suggests that even weakly enforced quotas matter.

Our results pose one important question for future research—are our findings generalizable beyond the Ukrainian case? Ukraine is a good case to test our argument given its low level of gender equality, weakly enforced quota, clear evidence of compliance variation, and strong regional divisions. Yet, not all countries have similar regional cleavages or low levels of gender equality. In addition, the sources of strategic variation may differ in other contexts. Thus, more research on quota compliance in different contests must be done to further test our expectations.

The results also raise the possibility that the politics of memory may be an important determinant of gendered outcomes in other contexts. While the specifics of the Soviet experience may be unique to post-Soviet countries, there is little reason to expect that the effect of historical memory is limited to them. Further research on historical memory and gendered outcomes in other contexts will help us better understand the extent to which the past continues to influence the future.