Gender and Politics in Northeast Asia: Legislative Patterns and Substantive Representation in Korea and Taiwan

ABSTRACT The article examines the conditions under which female legislators are more likely to act on behalf of female electorates through two underexplored cases – South Korea and Taiwan. Specifically, it investigates the effect of three conditions – seat share, electoral rules, and legislator characteristics – on legislators’ sponsorship of women’s issue bills using an original bill submission dataset. The finding shows that, on the one hand, female legislators’ increasing seat proportion made legislators stress women’s issues more and, on the other hand, new legislators elected at the party tier with civil society experience became substantially more likely to advance women's issues. In light of the evidence, this article argues that women’s issues are more actively advanced when the political space allows women’s issue-promoting legislators to pursue both electoral and policy interests.


Introduction
In countries where key left-leaning political parties are characterized by issues more than socialdemocratic ideals, for example national security or national identity, how can the women's issue promotion agenda move forward? Examining the cases of two East Asian democracies -South Korea (henceforth, Korea) and the Republic of China (henceforth, Taiwan) -this article aims to shed light on the important role of individual legislators in promoting legislative measures concerning women's interests from the gender-equality perspective (henceforth, women's issues). Both Korea and Taiwan have seen significant increases in their proportions of female legislators between 1992 and 2016, from 1.5% to 17% in Korea and from 10% to 38% in Taiwan (Inter-Parliamentary Union, 2017). 1 Various pieces of evidence drawn from different regions have repeatedly demonstrated (Reingold 1992;Schwindt-Bayer 2010;Waylen 2007) that there is positive linkage between descriptive representation (the number of female legislators) and substantive representation (the degree to which female legislators act in the interests of female voters). Echoing this expectation, in both countries, the proportion of bills promoting women's interests rose from roughly 0.3% in 1992 to more than 2.3% in 2016. 2 However, we lack understanding about specific conditions under which female legislators' women's issue promotion is more likely.
The primary purpose of this article is to fill this gap by asking a series of related questions. Does a higher proportion of bills concerning women's issues simply reflect an increased seat share of female legislators who tend to have women-friendly preferences? Does it reflect a women's issue-promoting trend under which both male and female legislators increasingly prioritize women's issues along with the growing proportion of female legislators? In addition to legislators' gender, do other individual traits facilitate legislators' attention to women's issues? And under which political institutional arrangements are they more likely to promote issues benefitting female voters? Do party-tier-elected legislators tend to push women's issues forward more often compared to district-tier-elected legislators?
To address these questions, this article examines three conditions affecting women's substantive representation -female legislators' seat share, electoral rules, and legislators' attributes (elected terms and previous careers). It employs an original dataset that consists of the whole universe of bills submitted by individual legislators between 1992 and 2016 in both Korea and Taiwan (roughly 50,000 in total). The two countries are ideal cases to test specified conditions because of high variation in key conditions during the observed period, for example dramatic changes in women's seat shares over time or the existence of a mixed electoral system composed of district and party tiers, despite shared cultural, sociodemographic, and developmental characteristics.
The empirical results point that, in line with the extant literature, female legislators are substantially more likely to focus on women's issues compared to male legislators in Korea and Taiwan. However, the three conditions analyzed in the article deepen our understanding of which specific legislator traits and particular conditions facilitate the promotion of female voters' interests. First, as for the seat share, the article demonstrates two mechanisms connecting the numerical presence of female legislators and increased substantive representation of women's issues in Korea and Taiwan. On the one hand, straightforwardly, it happens through increased legislative activities of female legislators who tend to have substantially more pro-women inclinations. On the other hand, the increasing seat share of women over time makes both male and female legislators more women-friendly. Second, the results show that female legislators elected at the party-tier tend to prioritize women's issues substantially more than their colleagues elected at the district tier. Among party-tier-elected ones, female legislators elected through the gender quota after the introduction of the new electoral rule in each respective country (2004 in Korea and2008 in Taiwan) were especially keen on advancing women's interests. Third, female legislators who are newly elected or having civil society experience were found to be particularly more likely to be engaged in women's issues.
Judging by the effect of female legislators' increasing seat share on the one hand, I argue that the promotion of women's issues in Korea and Taiwan have become beneficial from the vote-seeking perspective for both male and female legislators. However, the fact that female legislators who are more likely to have strong pro-women policy motivations -party-tier elected legislators, newly elected legislators, or legislators with civil society experience -are particularly keener on advancing women's issues indicates that policy interests of legislators still matter. A key lesson we can draw from the findings of this article is that substantive representation is facilitated by a political space in which women's issue promotion becomes beneficial for both vote-seeking and policy-seeking interests of individual legislators.
The contribution of this article is twofold. First, the article offers new insights for the ongoing debates within the gender politics scholarship about the specific factors that facilitate or hamper substantive representation of female voters through elected female legislators (for example, Barnes 2014;Crowley 2004;Dahlerup 1988;Krook 2015). Findings here will be particularly relevant to young presidential democracies whose left-leaning parties are identified with political cleavages beyond class conflicts or minority protection (Shim 2020) because the article shows under which specific conditions individual legislators can play an equally important role in setting women's issue promotion agenda.
Second, considering that other empirical works conducted with East Asian cases have focused largely on conditions affecting descriptive representation, this article fills a clear empirical gap on the issue of substantive representation by examining legislative behavior with a comprehensive dataset. Understanding the conditions of substantive representation also has important policy implications since numerous women's organizations and academics in both Korea and Taiwan have insisted on increasing the presence of female legislators even further. Their recommendations have taken various legislative or constitutional forms, such as the introduction of the 30% mandatory nomination of female candidates at the district-tier in Korea or the increase of the party-tier portion to 50% in Taiwan. The paper will proceed as follows: First, it will introduce relevant theoretical debates and findings about three conditions affecting female legislators' substantive representation of women's interests. Drawing from this, specific hypotheses regarding Korea and Taiwan will be derived after a brief introduction of related political contexts. Second, the paper will explain how the term "women's issues" is defined and why bill sponsorship data is selected to examine the relationship between female legislators and women's issue bills. Third, using descriptive statistics and logistic regressions, three key conditions of substantive representation -seat share, electoral rules, and legislator's attributes -will be tested. Fourth, the concluding section will summarize the key findings and implications of the paper and suggest promising avenues for future research on the topic.

Theory of descriptive and substantive representation
The primary insights of this article stem from Pitkin's (1967) distinction between descriptive and substantive representation. Based on her definition, descriptive representation ("standing for" representation) reflects certain sociodemographic characteristics of constituents such as race, class, or gender, while substantive representation ("acting for" representation) is about actually acting in the interests of the represented. In other words, when women are many, they are more likely to voice concerns about women's issues and produce relevant political results. That is, as initially noted by Phillips (1995) with her theory of the "politics of presence," the presence of women in public office has the potential to transform representation by reflecting women, their lives, and their perspectives in the policymaking process. This in turn increases governmental responsiveness to women's needs, interests, and perspectives (Burrell 2006;Swers 2002). There are several reasons behind this linkage. First, women are more likely to trust other women, who share a descriptive characteristic (gender) and gender-related experiences; to understand their needs; and present their substantive interests. Second, even male legislators hold the expectation that female legislators have more expertise in the area of women's issues. Third, several works suggest that female voters represented by female (rather than male) legislators are more likely to discuss politics with others (Atkeson 2003), express interests (Koch 1997), be aware of substantive policy positions of female legislators, and make female legislators more accountable to promises they made (Jones 2014).
A wide range of established empirical works from various regions show that having more female legislators actually makes a substantial difference in producing more women-friendly policies. Studies of members of the US House of Representatives, for example, have found that women are more likely than men to support policies that favor gender equity, day care programs, flextime in the work place, legal and accessible abortion, and reducing domestic violence and breast cancer (Thomas 1994). Similarly, gender politics scholars dealing with Rwanda or South Africa attribute the increase of new laws on gender-based violence or family law to women's increasing presence (Devlin and Elgie 2008;Waylen 2007). Likewise, works focused on Columbia, Costa Rica, or Hondura also provide evidence that female legislators tend to prioritize bills concerning women's rights (Schwindt-Bayer 2010;Taylor-Robinson and Heath 2003;Volden, Wiseman, and Wittmer 2018). Beyond legislative measures, recent empirical works have also confirmed the positive relationship between female legislators and their prioritization of women's issues through elite surveys (Coffé and Reiser 2018), as well as plenary and committee debates (Bäck, Debus, and Müller 2014;Clayton, Josefsson, and Wang 2017).

Conditions of substantive representation
Despite a large number of findings showing the positive connection between descriptive and substantive representation of women, it is nonetheless a probabilistic relation -that is, the number of female representatives per se does not guarantee that women's issues will be promoted. Therefore, works in gender politics have also examined other factors that determine the representation of women's interests such as legislators' personal traits, party identifications, or broader institutional contexts (Bratton 2005;Childs 2008;Childs and Lovenduski 2013). Taking a cue from this trend, this article focuses on three key conditions: female legislators' seat shares, electoral rules, and legislator attributes.
First, as for the seat share and substantive representation, the gender politics scholarship has shown a mixed picture. On the one hand, empirical findings demonstrate that a growing number of female legislators are associated with the increasing prioritization of women's issues by legislators (Bratton 2005). On the other, different scholars have observed that the percentage increase in elected women actually decreases the likelihood of legislators acting on behalf of women (Carroll 2001). Even the mechanisms behind each noted pattern have contrasting expectations for both male and female legislators. As for the legislative behavior of female legislators, some argue that the increasing presence of women in legislative chambers will enable them to form strategic coalitions with one another to promote women's issues (Thomas 1994). Relatedly, critical mass theory suggests that if female legislators reach a specific percentage threshold, they would thereafter be able to improve the representation of women's interests by altering the institutional culture (Dahlerup 1988). On the contrary, others posit that having a higher proportion of female legislators can lead to the deprioritizing of women's issues -since growing heterogeneity between female legislators can divert their priorities to other issues, or they might simply expect other female colleagues to focus on promoting women's interests (Carroll 2001;Schwindt-Bayer 2006).
Moreover, critics of the simple numerical presence approach argue that how male legislators will behave upon facing an increasing percentage of women is an empirical question (Childs and Krook 2009). On the one hand, some scholars argue that increased female representation makes women's issues popular, mainstream, and reelection friendly, which influences men's legislative behavior by making it focus more on women's issues. This, in turn, would minimize gender difference (Bratton 2005;Crowley 2004;Thomas 1994). On the other hand, others argue that the increased numerical presence of women can cause a backlash -male legislators will not hesitate to use a range of strategies to halt policy initiatives concerning women (Hawkesworth 2003;Krook 2015;Michelle Heath, Schwindt-Bayer, and Taylor-Robinson 2005). From this perspective, a lower proportion of women can serve as a more favorable political climate to promote women's concerns -since it would not appear as challenging male dominance (Crowley 2004).
Going beyond the effect of numbers, another strand of research has examined how specific political institutions facilitate substantive representation. And one of the most frequently recognized political institutions known to be affecting women in politics is electoral rules. Empirical works have repeatedly revealed that the proportional representative electoral rule is more conducive to the entry of female politicians, in comparison to the district-based rule with low district magnitude (Matland 1998;Salmond 2006). Extending the effect of electoral rules to the domain of the substantive representation of women is a newly developing research field. Between party-and district-tierelected legislators, it has been noted that party-tier ones (either nationwide or statewide) are more likely to promote women's issues compared to district-tier peers -since the former are free from obligations related to their local constituencies and, at the same time, incentivized to develop a more global representation style that can benefit their party as a whole (Tremblay 2006).
The latest empirical works based on countries with a mixed-member electoral system have demonstrated that party-tier legislators tend to prioritize voters beyond particular localities and focus on a wide range of social groups such as the advancement of female voters' interests. For instance, in New Zealand and Germany, legislators elected at the party-tier tend to take their key representational focus from particular social groups such as women or ethnic groups cross-cutting regional boundaries (Coffé 2018). Moreover, Höhmann's (2019) work goes further by explicitly theorizing the aforementioned positive relationship between being party-tier legislators and their women-friendly policy orientation from the vote-seeking perspective, as well as empirically testing it with legislators' oral and written parliamentary questions in Germany.
Building on this line of inquiry, I argue that party-tier legislators are more likely to promote women-friendly policies even from the individual legislators' policy-seeking perspective. That is, we cannot be certain if party-tier legislators can or will continuously seek to be reelected at the same tier without moving to the district one; or, they might even terminate their political career afterward. Even under these circumstances, I expect party-tier legislators to promote women's issues more than their district colleagues do because they themselves are more likely to be nominated by their parties in light of their policy-seeking potential going beyond particular local constituencies (especially under the closed-list system). From the perspective of parties, what is crucial for the party-tier is to maximize the state or nationwide vote share of the party. Hence, it is a rational strategy to nominate candidates who have demonstrated their preference and expertise on policy issues reaching broader interests not bounded by any particular locality. Considering that politicians are also known to be policy-seeking in addition to being office-or vote-seeking (Müller et al. 1999), once elected, party-tier legislators are more likely to promote women-friendly policies out of their policy-seeking motivation as compared to district-tier peers.
Third, individual characteristics of legislators merit more attention. Carroll (2001) points out that the chance of women acting as a cohesive legislative group declines as the group's size grows and it consequently becomes increasingly heterogeneous. Several empirical works focus on the neutralizing role increasing diversity has within the female legislator cohort. They find that we can expect female legislators' legislative priorities to be influenced by differences in the life experiences, perspectives, needs, and interests between white women and women of color, poor women and wealthy women, and rural women and urban women (Dodson 2006). Although women can be directly compared with other minority groups, gender -unlike ethnicity or race -tends to crosscut key political cleavages and is, therefore, more susceptible to political neutralization (Htun 2004). For instance, Dalton's (2015) work shows how Japanese female legislators promote a conservative understanding of femininity based on stereotypes of motherhood and housewifery, while Lovenduski (1986) points out that European socialist men once opposed female representation in the legislature based on the reason that female legislators mostly come from white, middle-class, and elite backgrounds. Moreover, empirical evidence demonstrates that those with feminist backgrounds or women's organization experience make the most difference when they get elected (Dodson 2006). In this regard, what ought to be considered for understanding elected politicians' women's issue-promoting efforts is not simply their gender but also other attributes such as their ideology, previous career, income level, and number of elected terms. In this sense, the paper aligns with the "critical actors" approach by recognizing the fact that influential individual characteristics can be found beyond one's gender among those key individuals who resolve to promote women's interests (Childs 2008;Childs and Krook 2009).

Gender politics in Korea and Taiwan: political contexts and hypotheses
Aforementioned expectations on conditions of women's substantive representation are largely drawn from empirical works based on Anglo-European democracies or Latin American countries. To derive more specific testable hypotheses for the two understudied East Asian cases examined here, political contexts and institutional arrangements related to gender politics in each country should be understood. During the period of observation, 1992-2016, there were several noteworthy political developments regarding women's issues in Korea and Taiwan. First, democratization from the late 1980s opened up possibilities for female activists to advance women's issues in both countries (Clark and Lee 2000). In Korea, many women's groups fought together for democratic transition and after achieving their goal in 1987 they focused on women's issues (Kim and Kim 2010). In Taiwan, women's activism was reflected by the diversity and number of feminist groups after the dissolution of martial law in 1987 (Clark and Clark 2008). This pattern is in line with established works that show democracies create citizens more supportive of gender equality (Norris, Inglehart, and Welzel 2002) and that democracy and women's status is strongly correlated (Tripp 2013).
However, democratization was far from sufficient in advancing women's interests. The crucial watershed moment leading to "gender mainstreaming" in Korea and Taiwan was the United Nations' Fourth World Conference on Women, which was held in Beijing in 1995 and attended by 40,000 NGO and 181 state representatives. The conference set 12 action plans to further women's rights and spurred transnational information sharing on quotas via academic networks and personal contacts between femocrats, feminist politicians, and women's movements (Krook 2006).
The effect of this conference clearly manifested itself in each country, albeit in a different form. Taiwan was unable to participate in the 1995 Beijing conference as an independent country. However, Jo Freeman, an American feminist who was participating in a feminist training camp in October 1995, emphasized the importance of having a critical mass in order to make a substantial impact in any organization. In response, feminist activists attempted to increase the number of reserved seats for elections at all levels. The Awakening Foundation -a strong women's group -was closely involved in pressuring the government to establish a formal committee. And their efforts resulted in the prime minister convening the Commission for the Promotion of Women's Rights in 1995. With respect to key legislative changes, activists helped bring about changes in the late 1990s that made divorce law and parenting law more woman-friendly and also helped secure the provision of parental leave and childcare facilities for employees in 2002 and 2007, respectively. In Korea, influenced by the Beijing Conference, women's issues gained salience in the political scene in the late 1990s. Several NGOs advocating women's rights mushroomed and formed a broad political coalition called the Korean Female Organization Coalition during that period, which was closely followed by the Women's Network for Transparent Politics around the turn of the century. These pan organizations actively engaged in a wide range of political activities, such as filing legal petitions, holding press conferences, running education programs for female leaders, evaluating women's issue manifestos, or creating blacklists or white lists for upcoming elections (e.g. among 101 recommended female candidates, 46 were accepted by major parties in 2003). As a result, the 1995 basic Framework Act for Women's Development was enacted in 1995 and followed by the 1999 Gender Discrimination Prevention law. However, like their Taiwanese counterparts, their chief achievements were the introduction of the 50% legal quota for party-tier seats from the 2004 general election and abolition of ho-ju law (which treated men as the heads of their families) in 2008.
Against this backdrop, numerous studies have demonstrated the role of women's movements on substantive representation. For instance, Cooke (2010) illustrates, after the democratization, how the strategic efforts of a wide range of women's organizations resulted in the creation of more political opportunities and advanced gender equality in Korea. In a similar vein, Huang (2015) finds Taiwan's adoption of reserved seats in the early 1950s stemmed from the Chinese women's movements while the gender quota in the 1990s derived from the international norm-inspired feminist movements.
Another avenue of research that students of East Asian gender politics have investigated is the effect of introducing seat quotas for female legislators. So far, empirical works demonstrated various aspects of quota implementation. On the one hand, Shin's study (2014) shows the spill over effect of gender quotas from one elected tier to another in Korea -female legislators elected via party-tier later being nominated and reelected to the district-tier. Yang's work with the Taiwanese case (2018) underscores that the gender quota in the party tier changed the landscape of women's representation since it brought a more diverse pool of candidates into the political scene. In a similar vein, Huang (2016) shows that the majority of female legislators elected through the reserved seats are equally or more competent than the men they unseated. On the other hand, Lee and Shin (2016) attribute the lack of implementation of the gender quota at the district-level in Korea to the non-routinized and clientelistic candidate selection process, which favors established and well-resourced male candidates. Similarly, Yoon and Shin (2015) point out how established parties take advantage of institutional loopholes and resist compliance -for example, there is no sanction for noncompliance or predominantly placing male candidates in winnable seats. Moreover, based on her quantitative and qualitative comparison of Korea, Taiwan, and Singapore, Tan (2016) concludes that the effect of quotas is not mechanical but contingent on factors such as the degree of party system institutionalization, electoral competitiveness, or societal attitudes toward women.

Conditions of substantive representation in Korea and Taiwan
Despite the importance of the gender quota effect and the role of women's movements, the aforementioned studies have failed to empirically scrutinize whether and how elected female legislators actually engage in women's issues. Here, I will specify relevant country-specific details in light of the three conditions affecting substantive representation -seat share, electoral rules, and legislator attributes -and, from that, derive testable hypotheses. Other than the first hypothesis, expectations are not gendered and apply to both male and female legislators.
First, as pointed out in the introduction, both countries have seen a substantial change in the seat share of female legislators during the period of observation. Drawing directly from the specified general debates, we can derive the following set of hypotheses: Hypothesis 1-a: Female legislators are more likely to prioritize women's issues as female legislators increase their seat proportion.
Hypothesis 1-b: Female legislators are less likely to prioritize women's issues as female legislators increase their seat proportion.
Hypothesis 1-c: Male legislators are more likely to prioritize women's issues as female legislators increase their seat proportion.
Hypothesis 1-d: Male legislators are less likely to prioritize women's issues as female legislators increase their seat proportion.
In relation to the electoral rules, both Korea and Taiwan have had mixed electoral systems during the period of observation. Prior to the 2002 election, Korea had both a district-tier (all single-member districts) and a party-tier for general elections, and the seats for the latter were filled in proportion to seat or vote shares at the district-tier. The party-tier was nationwide, so based on the aforementioned theoretical expectations we can expect party-tier legislators to focus more on women-friendly issues vis-à-vis their district-tier counterparts. Similar to the Korean case, before its 2008 election, Taiwan had a party-tier whose seats were distributed according to the vote share at the district-tier. However, the latter consisted not of single-member districts but multi-member ones with a medium level of district magnitude (MMD) in combination with single, non-transferable voting rules (SNTV). According to established studies on electoral rules, for example Carey and Shugart (1995), MMD-SNTV lend themselves to highly personal appeals/clientelistic practices through intra-party competition between co-partisans. In this sense, we could guess that party-tier legislators focus much less on personal appeals and more on public goods such as promoting women's issues -from which female electorates, in general, can benefit.
Since the 2004 election in Korea and the 2008 one in Taiwan, both countries have had general elections based on newly introduced systems which do not allow compensation between tiers, and the electorate cast votes for each separately: single-member districts for the district-tier and national closed proportional representative rules for the party one. However, what characterizes the party-tier in both countries is that major parties employed non-renomination practices. That is, legislators who were nominated and elected at the party-tier are pressured to run at the district-tier in the next election. Although this convention clearly undermines why party-tier legislators are more likely to promote women's issues from the individual-level, vote-seeking logic described earlier, from the parties' perspective they had incentives to nominate candidates who have the potential to serve nationwide interests -such as child experts, nurses, or NGO representatives. And this clearly has been the case in both countries (Shim 2019;Yang 2018). Thus, from the policy-seeking perspective, we could expect these candidates, with their expertise and interests, to promote policies with nationwide resonance -like women's issues -after getting elected.
Another noteworthy aspect of electoral reform in both countries is that the gender quota was applied at the party tier. That is, echoing previous works demonstrating that gender quota laws are easier to implement with proportional representation (Matland 2006), parties in Korea and Taiwan have been legally obliged to fill 50% of the party-tier list with female candidates. In light of the empirical findings showing that quota-elected women tend to be instrumental for the successful passage of bills related to women's rights (Kerevel and Atkeson 2013), we can expect female legislators who were elected through the gender quota to be particularly keen on advancing women's interests. These institutional contexts and practices of the two countries lead to the following two hypotheses: Hypothesis 2-a: Male and female legislators elected at the party-tier are more likely to prioritize women's issues compared to their colleagues elected at the district-tier Hypothesis 2-b: After the electoral reform, female legislators elected at the party-tier are more likely to prioritize women's issues compared to other legislators Third, as for the effect of legislators' individual attributes on women's issues, previous career and elected terms are particularly relevant in the Korean and Taiwanese context 3 . First, a legislator's previous career can be defined based on the period between the end of their education and the beginning of their first terms as elected legislators. For both countries, the four most frequent career categories are (see the supplementary materials for the percentage breakdown by legislator's gender in each country) i) journalist/broadcaster, ii) civil society organizations, iii) professional (e.g., doctor, lawyer, professor, pharmacist, etc.), and iv) politician/civil servant. As noted earlier, considering the significance of women's organizations on the progress of women's issues in the two countries, among other career categories, we can expect that legislators with civil society experience tend to promote substantive representation of women much more than others, as "critical actors." In this regard, the comparison can be made between this category and the rest.
Second, elected terms of legislators have ranged from first to seventh term in both countries. Here, we can expect newcomers to focus more on substantive representation of women. First, newly elected legislators tend to be younger in both countries and, based on previous works, age is an important positive factor supporting post-material values like feminism (Inglehart 1981). Therefore, newly elected legislators are more likely to be progressive and keener to support women's issues. Second, women's issue promotion is a median-voter friendly issue resonating with half of the electorate. Therefore, it can help individual legislators to build a positive reputation with the public which often get rewarded either by performance-evaluating organizations or their affiliated parties. In contrast, it is not hard to imagine that more experienced legislators will be less inclined to promote women's issues given their stable position within their party and alternative means to secure votes. Experienced legislators not only tend to have closer access to state resources but also have a wide range of entrenched networks and channels to distribute those resources to specific groups and geographical constituents. For instance, Barnes (2014) demonstrates that a legislator's seniority is positively associated with appointments to powerful committees, while negatively associated with committees focusing on social issues.

Hypothesis 3-a: Legislators with civil society backgrounds are more likely to prioritize women's issues compared to legislators with other backgrounds
Hypothesis 3-b: The fewer terms a legislator served, the more they are likely to prioritize women's issues

Data: bill sponsorship
Against the backdrop of increased gender diversity in the national legislatures in Korea and Taiwan, this paper taps into the whole universe of individual-legislator sponsored bills since 1992 to examine the impact of female legislators on promoting women's issues. The bill sponsorship data are used because legislation has brought about the most substantial changes for female electors over the past two decades in both countries -for instance, increased parental leave and financial penalties for workplace discrimination. Moreover, sponsorship data provide a rich source of information on legislators' preferences, power, and networks with other legislators. For instance, sponsoring a bill is an opportunity to express support for a certain issue through which a legislator can send signals to their constituents (Campbell 1982) or median voters (Kessler and Krehbiel 1996). Korea and Taiwan are no exception to this. Although the amount of effort a legislator puts into sponsoring or cosponsoring a bill can vary, legislators in both countries take this task seriously since bill sponsorship record can be a source of praise or criticism (e.g. by NGO Monitor Group in Korea or by Citizen Congress Watch in Taiwan). Furthermore, in addition to the fact that co-sponsorship activities in general tend not to be structured by party discipline (Talbert and Potoski 2002), the evidence demonstrates that the two presidential democracies' "separation of survival" logic -both the president and legislators are guaranteed to hold office for a fixed term (Samuels and Shugart 2010) -has facilitated legislators' entrepreneur potential and, as a result, enabled them to serve as an "independent law-making institution" (Shim 2019). All in all, bill sponsorship in both countries can be utilized as an ideal estimate of legislators' policy preferences -and is not merely a reflection of their affiliated party's policy directions.
According to the Korean and Taiwanese constitutions, both legislative and executive branches can propose bills. In the case of legislative-branch proposed bills, it can be divided into individual legislator submitted ones and standing committee-head submitted ones in Korea. In Taiwan, the distinction is made between bills submitted by individual legislators and party caucus. In the case of legislatorinitiated bills (also known as private members' bills) each sponsoring legislator must have the support (co-sponsorship) of 10 or more members of the National Assembly in Korea and 15 or more members of the Legislative Yuan in Taiwan. 4 There is no upper limit to the number of cosponsors in either country.
Insofar as legislator submitted bills are concerned, as can be clear from Figure 1, both countries have seen an increase in the proportion of submitted bills by female legislators in tandem with the increasing presence of female legislators.
Before proceeding further, we need to clarify what "women's issues" are, so as to select pertinent bills. In relation to this, there have been theory-free approaches to defining women's issues such as identifying women's issues as those on which female legislators introduce significantly more bills than male legislators (Volden, Wiseman, and Wittmer 2018) or using external sources (e.g. Congressional Caucus for Women's Issues or Institute of Women's Policy Research) to delineate the boundaries of women's issues. These approaches are not ideal for this paper since both countries lack reliable external sources from which women's issues can be identified regularly for the whole period of observation; moreover, if one defines women's issues as something female legislators focus more on, by definition it becomes impossible to compare to what extent male or female legislators prioritize women's issues. Gender studies combined with a theory-orientation has approached the definition in two major ways. On the one hand, women's issues have been approached rather broadly by equating them with policy areas where the topics of concern relate to care, compassion, and civil rights -such as health, education, welfare, inequalities for minorities, or reproduction health (Atkinson and Windett

2019
). On the other, they have been defined as issues that disproportionately affect women as a group (Jones 2016). This could be interpreted narrowly as issues promoting gender equality and, at the same time, directly targeting women, such as abortion rights, maternity leave, subsidies for mothers, antifamily violence, anti-workplace/employment discrimination, female career advancement, and maternal healthcare.
Defining women's issues broadly can make them indistinguishable from general social policies, or "soft issues". At the same time, following a narrow definition could exclude policy issues that sound gender neutral but where women are disproportionately more likely to be affected in practice. This article opts for an eclectic approach by adopting the narrow definition -but, at the same time, adding in some broad issues that particularly concern women in Korea and Taiwan. As noted in the gender politics scholarship (Reingold and Swers 2011), which issues are more likely to affect women varies over time and across contexts. Per a broad definition, issues related to the long-term care of the elderly, childcare, and work-family balance should be treated as women's ones because women are disproportionately more affected by them in Korea and Taiwan. There, women have often been expected to look after the elderly and children at the expense of their careers.
Based on the broad definition of women's issue bills, a total of 1,526 bills were submitted between 1992 and 2016 in Korea and Taiwan, of which 1,340 were sponsored by individual legislators.
This article is only based on these 1,340 legislator-initiated women's issue bills from which the initiator's gender can be traced. 5 This is roughly 2.49% of total bills submitted by individual legislators in Korea and 2.35% in Taiwan (which is 38,864 and 14,567, respectively). And legislators affiliated with both ruling parties and opposition parties have been significant contributors -40-60% each side -to women's issue bills during the period of observation. With regards to the contents of women's issue bills, each can be classified as an attempt concerning: (i) equal employment-promotion or general gender-equality promotion, such as education or training in schools or the workplace; (ii) parental leave and childcare support; or, (iii) abortion, contraception, sexual harassment, or domestic violence (see the supplementary materials for the percentage breakdown by country and specific coding rules).

Gender and sponsorship of women's issue bills
Before testing the three conditions affecting substantive representation of women, the article examines whether one finding repeatedly noted in the existing literature -female legislators are more likely to prioritize women's issues than male legislators (Bäck et all, 2014;Clayton, Josefsson, and Wang 2017;Coffé and Reiser 2018;Devlin and Elgie 2008;Schwindt-Bayer 2010;Thomas 1994;Waylen 2007)applies to Korea and Taiwan. For this goal, the proportions of women's issue bills and non-women's issue bills sponsored by male legislators, on the one hand, and female legislators, on the other, can be compared. Intuitively, we can associate a higher proportion of women's issue bills with having more issue attention to women's issues. Figure 2 (left-side) shows the proportions of women's issue bills sponsored by male legislators and female legislators from the total submitted bills. In line with expectations, female legislators are two to four times more likely to sponsor women's issue bills than are male legislators.
Despite female legislators' attention to women's issues, this does not mean that overall women's issue bills are substantially influenced by female legislators. After all, male legislators on average had a far larger share of seats during our period of observation (roughly 90% in Korea and 80% in Taiwan). Thus, I examined the proportional difference between male and female legislators who sponsored women's issue bills. As clearly illustrated in Figure 2 (right-side), female legislators are on par with male legislators in sponsoring women's issue bills. In both countries, roughly half of the women's issue bills are submitted by female legislators. By contrast, female legislators only sponsored 17% of nonwomen's issue bills in Korea and 31% in Taiwan.
The clear gender difference in the women's issue bill sponsorship patterns notwithstanding, there is a chance that women's issue bills submitted by female legislators are merely trivial changes or reverse the existing benefits. Bearing this concern in mind, I coded the significance and direction of women's issue bills. The "significance" aspect of each bill is primarily based on the coding scheme in the established welfare state research 6 while the direction aspect (retrenchment or expansion) mainly follows the welfare bill coding by Klitgaard and Elmelund-Praestekaer (2013, 60).
The result shows that 90% percent of women's issue bills were not trivial changes and 95% targeted expanded benefits for women in various forms in both Korea and Taiwan, regardless of the sponsor's gender. Therefore, it can be said that women's issue bills neither represent trivial proposals for signaling purposes nor look to roll back the existing benefits. Looking at the details, many of the successful attempts made by female legislators include changes whose impact can be felt by a large number of female electorates. For instance, in Korea, these are i) legalizing maternity leave before and after the childbirth, ii) government subsidies on premature infants, or iii) shortening working hours during mother's pregnancy. Likewise, in Taiwan, legislative changes include measures such as i) financial loans to domestic violence victims, ii) protecting mother's right to breast-feed in public areas by setting up relevant public facilities, or iii) extending the number of days for parental leave.

Logistic regressions
Descriptive legislation patterns suggest that female legislators prioritize women's issue bills over nonwomen's issue bills more than male legislators and, at the same time, their legislative contribution on women's issue bills is quite significant compared to non-women's issue bills. However, the key purpose of this paper is to test three conditions affecting substantive representation of women and, for this goal, I carried out a pooled logistic regression for all legislator-sponsored bills in both countries for the period of 1992-2016.
From Model 1 to Model 5, the outcome variable is whether a bill is related to women's issues (nonwomen's issue bills 0; women's issue bills 1) and the predictor variables differ from one model to another. Model 1 is the most basic form and tests the effect of gender (male 0; female 1), while Model 2 adds women's seat share proportion at a particular time (continuous) to test H1(a-d). Model 3 and Model 4 are designed to test H2 (a-b) and H3 (a-b), which examine the effect of elected tiers and legislator characteristics, respectively. As for Model 3, legislator's elected tier variable -(district tier 0; party tier 1) -is added while Model 4 includes elected terms (continuous variable) and previous career (civil society organization 1; or others 0). And Model 5 includes all variables simultaneously.
In the case of control variables, the following four are added. Initiation status of a bill 7 (whether a bill was a new enactment 0; or amended an existing bill 1), ideological spectrum of legislator's affiliated party 8 (right-leaning 0, left-leaning 1, and others such as independent candidates 2 9 ), and country and year dummies. In addition to the individual-level variables included here, other factors such as whether a legislator holds a leadership position (within the party, government, or parliament) or is affiliated to standing or temporary committees related to women's issues can affect the proclivity to represent such issues. However, due to the frequent changes in legislators' leadership responsibilities and committee membership in both countries, they were not included during the data-collection stage. To offset this limitation, for all models the results are based on standard errors clustered around individual legislators. And, the unit of analysis in this paper is sponsored bills. Since the primary purpose of the paper is to examine the conditions facilitating substantive representation of women by female legislators, the effect of each condition will be illustrated separately between male and female legislators in related figures.
First, the result in Model 1 confirms what we found from descriptive statistics -being a female legislator is a strong predictor of a submitted bill being a women's issue bill, while holding other variables to their means. Specifically, as described in Figure 3 (left), the predicted probability increases from 1.57% to 4.89% and is statistically significant at the 1% level. Model 2 adds the seat share of female legislators and shows that women's substantive representation is not merely a result of increasing the number of female legislators who tend to have more pro-women orientation than male legislators. As can be clear from Figure 3 (right), the increased seat share of women goes hand-inhand with both male and female legislators becoming more women's issue friendly -by changing the level of seat share from the minimum level to the maximum level, the predicted probability of a submitted bill concerning women moves dramatically from 0.86% to 2.99% for male legislators and from 2.72% to 9.04% for female legislators. Although the absolute degree of change is much less compared to their female counterparts, male legislators tend to prioritize women's issue bills along with the increasing female presence in the legislature. Based on this result, we can support H1-a and H1-c and, at the same time, reject H1-b and H1-d.
As for H2-a and H2-b, the effect of the elected-tier is in line with previous expectations based on the results in Model 3. Specifically, the results in Model 3-a clearly confirm that legislators elected at the party-tier tended to be more geared toward women's issue bills than their district-tier colleagues during the period of observation -by shifting from district-to party-tier, Figure 4 (left-hand side) shows that the predicted probability of a bill concerning women's issues increases from 1.46% to 2.11% for male legislators and from 4% to 5.71% for female ones. Although not included as a separate model in Table 1 so as to conserve space, the effect of a gender quota was tested by comparing the prowomen's issues bill tendency across male and female legislators following the electoral reform in each country. The predicted probability illustrated in Figure 4 (right-hand side) demonstrates that partytier female legislators (all of whom were elected via the gender quota) are clearly the most women's issues-oriented category of the four after the electoral reform. Hence, these results support both H2-a and H2-b.  Notes: (1) ***p <0.01, **p <0.05, *p < 0.1 (2) Difference in the number of observations between the models can be attributed to the lack of information, e.g. previous career, regarding legislators elected through by-elections or reelections.
Moreover, Model 4 confirms that both legislator characteristic variables included for H3-a and H3b -career background and elected terms -follow the specified expectations. Namely, legislators with lower numbers of elected terms or with civil society experience (compared to those without) prioritize women's issue bills more.
To examine whether there exists any gender difference in two conditions positively affecting substantive representation of women -civil society backgrounds and elected terms-I examined the predicted probability differences by gender as illustrated in Figure 5. The result shows that having a civil society experience prior to legislators' first term and the number of terms both have a positive impact on both male and female legislators. However, the positive effects are particularly pronounced for female legislators. Finally, Model 5 includes all variables simultaneously, and the results demonstrate that the degree and direction of statistical significance observed in Models 1-4 stay the same except for the women's seat share proportion variable. This indicates that the positive effect of women's seat share might have been realized through other explanatory variables; for example along with a women's seat share increase, female legislators as a group ask for a type of electoral rules change through which more women-friendly measures can be adopted or women-friendly candidates voted in.
Although not the core focus of this paper, the statistical results in Models 4 and 5 demonstrate that a left-leaning party affiliation does not make a legislator more likely to support women's issues bills than being right-leaning. A lack of partisan bias might be considered a counterintuitive finding from the perspective of the existing literature, which points out that -with their emphasis on inclusiveness, diversity, and equality -left-leaning parties are more likely to select female candidates, adopt gender quotas, support women's movements, and produce legislation pertinent to females (Beckwith and Cowell-Meyers 2007;Kittilson 2006). However, the latest gender politics scholarship also notes that factors such as pressure from women's movements or electoral threats from a left-wing party that spurs the gender-equality contagion can push right-leaning parties to promote women's interests (Celis and Erzeel 2015).
Some highly salient qualitative evidence in Korea and Taiwan aligns with this. The center-right Grand National Party, for example, nominated the highest proportion of female candidates for the 2010 local elections in Korea. Furthermore, the center-right Saenuri party was the first party to nominate a female presidential candidate (Park Geun-Hye) and also nominated the first non-ethnic Korean female candidate for the 2012 general election (Jasmine Lee). In Taiwan, the adoption of partylevel gender quota is a clear example of "contagion from left". Initially adopted by the left-leaning Democratic Progressive Party in 1996 as 25% quota for women candidates, the right-leaning KMT followed its footsteps in 2000 and both parties finally agreed to put the 50% nomination rule (in the party tier) during its 2005 constitutional reform. Also, in 2004 several legislators from the center-right Pan-Blue coalition formed the Female Legislators Alliance to promote women's welfare in the Legislative Yuan.

Concluding remarks
The paper contributes to the gender politics scholarship by testing three conditions affecting female legislators' substantive representation of female voters -seat share, electoral rules, legislator characteristics (career background and elected terms). On the one hand, findings based on seat share suggests that women's issues have become beneficial from the vote-seeking perspective in Korea and Taiwan. On the other hand, findings from the effect of legislators' elected tier and individual attributes indicate that female electorates' substantive representation is advanced more actively when electoral and policy interests of legislators coincide. From this, it can be said that the ideal environment for substantive representation of female voters is a political space where both policy and election-oriented legislators can engage in women's issue promotion. The article also makes a key contribution by investigating the legislative behavior of female legislators in two under-studied countries, where academic attention was largely paid to the impact of gender quotas or the role of women's movements in promoting women's issues. With respect to the policy implications, the results in this article clearly direct to the further increase of female seats in both countries. Although substantive representation of female voters is not the only goal for more descriptive representation of female legislators 10 , the findings here demonstrate that higher descriptive representation does translate into women's issue promotion. That is, contrary to the potential concerns raised by previous works which argue that the increasing presence of female legislators in the legislature can result in female legislators shifting away from focusing on women's issues or the possibility of facing a backlash from male legislators, both males and females have become increasingly women's issues-friendly in Korea and Taiwan.
Building on this article, what are future avenues of research on Korea and Taiwan? First, a more nuanced approach can be taken by dividing women's issues into different categories, e.g. childcare support, divorce law, domestic violence, equal employment, and comparisons of male and female legislators" or between different parties. Going further, women's issues themselves can be compared to more "male" issues, e.g. immigration, taxes, energy, military, crime, agriculture, construction, or business, and examine legislative behavior differences between male and female legislators. Second, insofar as Korea and Taiwan are concerned, cross-party or cross-gender cooperation on women's issue bills can be systematically compared based on bill co-sponsorship. For both countries, it has been mandatory for legislators to have 10-20 co-sponsors for any submitted bill. Therefore, going beyond the sponsorship records employed for this paper, legislative networks of co-sponsors on women's issue bills can be compared over time, which can provide valuable insights into another under-studied aspect of substantive representation.

Notes
1. Unlike Korea's constitution, Taiwan's 1946 constitution states that seats should be reserved for women in the legislative body (the exact seat varied from 10% to 25%). This partly explains female legislators' higher proportion of seats in Taiwan (for details, see Huang 2015). 2. 1992 is chosen because that is the year Taiwan had the first direct election (for Korea, the first direction election held after its recent re-democratization was in 1987). 3. Potentially relevant aspects such as income or family background could not be included due to the lack of data while other factors such as education, religion, or race are not relevant because they are either not politicized or do not vary much between legislators. 4. The threshold was 20 until 2003 in Korea and 33 until 2008 in Taiwan. Although the number varies over time, there were approximately 300 house members in Korea and 225 (later halved to 113) in Taiwan before 2008. 5. For other bills, gender of bill-initiator cannot be traced, e.g. party caucus or the cabinet, or, if it is possible, intention cannot be linked to the bill initiator since s/he is just signing as a result of collective decision-making, e.g. standing committee heads.