Legal Context

Gender quotas are considered to be an effective means for improving the gender balance in legislatures, but their impact varies with their target and level of enforcement. The Gender Quotas Database (International IDEA 2021) provides comprehensive data on the diversity of gender quotas around the world. At the national level, 28 countries have adopted a reserved seat system, which allocates a certain proportion of seats to women, in the lower house. 65 countries have legislated candidate quotas that require political parties to field a certain percentage of women (or both women and men) candidates at the national and/or the local level. Political parties in 56 countries voluntarily use gender quotas when they endorse candidates. It is not unusual for two or three types of gender quotas to be used in a single country. Taken together, 129 countries used gender quotas of some kind as of January 2021. More than 30 countries have gone so far as to enumerate gender quotas or require equal electoral opportunities in their national constitutions.

Japan, on the other hand, has long resisted the adoption of gender quotas (Gaunder Reference Gaunder2015). Women’s organizations have called for quotas since at least the 1990s, but the quota movement gained enough strength to pressure political elites only when a women’s organization called Q no Kai (Association to Promote Quotas) was created in 2012 (Miura Reference Miura, Hardcare, George, Komamura and Seraphim2021). An all-partisan parliamentary group was formed in 2014 and drafted the Gender Parity Law, which was enacted with unanimous support in 2018. It urges political parties to field an equal number of men and women candidates in all elections. The law also encourages parties to take special measures, including numerical targets, to increase women candidates. Because the law is not binding, only some progressive parties have established such targets. In 2021, an amendment was passed in the Diet, again unanimously, that introduced a new clause to require legislatures and governments to prevent sexual harassment against women candidates and legislators.Footnote ² While the all-partisan parliamentary group that drafted the original Gender Parity Law and its amendment had sought to make numerical targets compulsory, it gave up on that idea because of resistance from the long-ruling conservative LDP. Instead, it included several options that political parties are encouraged to adopt, such as improvements to the candidate endorsement process, providing training for women candidates, and adopting preventive measures against the sexual harassment of women.

A major reason for the LDP’s reluctance to set candidacy targets, despite calls to do so from some of its own women legislators, is intraparty constraints. For progressive opposition parties, quotas offer an opportunity to advocate their vision for gender equality. They are also in the minority and thus have fewer men incumbents. In order to win elections, they need to field many new candidates, which gives them room to implement numerical targets for women. The Japan Communist Party (JCP) has set targets of 30% for local elections and 50% for national elections. The Constitutional Democratic Party (CDP) and the Democratic Party for the People (DPP) have set a 30% target for national contests. Indeed, even with the Gender Parity Law’s nonbinding mandate, the proportion of women opposition candidates increased slightly, from 22% in the 2017 lower house election to 26% in 2021. When looking solely at non-incumbents, the numbers are 24% and 30%, respectively.

By contrast, the LDP is predominantly male, including 92% of incumbents in the lower house and 85% in the upper house. Any numerical target would force a sizable proportion of LDP legislators to forfeit their seats, creating the risk of intraparty fissure. Moreover, the LDP’s candidate nomination process is highly decentralized, making it all the more difficult for national party leaders to impose numerical targets on its local branches. As incumbents are customarily granted automatic party endorsement, the LDP faces practical difficulties in setting a numerical target unless it drastically changes its endorsement process, especially in single-member districts. Its share of women candidates in lower house elections only increased from 8% to 10% between 2017 and 2021; among non-incumbents, the shares were just 7% and 15%. The exception is the open-list proportional representation (PR) tier of the upper house, in which voters can cast a vote either for an individual candidate or party. The LDP set a 30% target for proportional representation in the 2022 upper house election and achieved that goal, although open-list PR seats in Japan are considered to be difficult to win for inexperienced candidates.

Public Attitudes toward Quotas

Binding or compulsory gender quotas are still rare in Japanese government and society, although various actors have proposed voluntary targets. For example, the Cabinet Office of the executive branch has set various numerical goals to close gender gaps in a wide range of fields, including a 35% target for electoral candidacy by 2030. These targets, many of which were set in 2003, were ambitious from the start, and as of 2020, almost all decision-making bodies failed to pass this bar.Footnote ³ For example, the government was compelled to postpone the 30% target for women to take leadership positions across various sectors by 2020 to “as early as possible within the next decade.”

In Japan, the government has used the term positive action (“affirmative action” in American parlance), which mixes concrete quotas with aspirational goals and timetables. Thus, the term “quota” remains unfamiliar to many Japanese citizens. Neither corporations nor universities have used concrete quotas for their selection of managers or students, although some universities have recruited women on an ad hoc basis in order to redress gender imbalances in science and engineering faculties.

Public opinion data suggest that Japanese citizens do not give high credence to quotas, particularly mandatory ones. According to surveys by the Cabinet Office in 1995, 2000, and 2004 that asked respondents to choose appropriate “positive action” measures, only 20% to 23% of respondents selected voluntary candidate quotas by political parties. Quotas for companies’ recruitment and managers received slightly higher approval, between 19% and 30%.Footnote ⁴ A more recent survey of voters conducted by the University of Tokyo and Asahi Shimbun (UTAS; Taniguchi and Asahi Shimbun Reference Taniguchi and Shimbun2016) confirms this pattern. Its 2016 survey found that just 25% of men and 29% of women respondents agreed with the introduction of quotas to allocate a certain proportion of seats or candidacies to women.

That said, gender quotas are not an electorally polarizing issue, as seen by the ambiguous attitude of political elites. UTAS runs a parallel survey of election candidates, which includes identical questions to those asked of voters. Looking at the responses of victors in the 2016 upper house election,Footnote ⁵ 42% of men and 65% of women winners agreed with the introduction of gender quotas, while 20% and 13% disagreed, respectively. Among LDP members, however, agreement was only 10% for men and 17% for women winners. Gender has a clear correlation with quota support: the agreement ratio of men and women politicians have statistically significant differences at the 5% level. We cannot realistically estimate gender differences by legislators’ party—something that Bohigues and Piscopo (Reference Bohigues and Piscopo2021) have shown in Latin American countries—as the number of elected women is very low. However, it is worth noting that 59% of men and 50% of women LDP respondents answered that they neither agreed nor disagreed with quotas. While we cannot discount social desirability bias on the part of legislators,Footnote ⁶ the high level of neutral responses suggests that elites are not eager to openly oppose quotas. This implies that elite partisan framing may be less salient on this topic, at least in Japan.

Benevolent, Hostile, and Modern Sexism

Previous scholarship offers insights into the role of gender beliefs. The question of whether to implement legislative quotas is inexorably tied to one’s views about the fitness of men versus women to serve in office, as well as women’s “appropriate roles” and “virtues.” Not surprisingly, those who hold gender-egalitarian attitudes tend to support gender quotas (Keenan and McElroy Reference Keenan and McElroy2017). However, the opposite attitude—sexism—needs to be analyzed with some nuance. For example, Glick and Fiske’s (Reference Glick and Fiske1996) seminal work posits the importance of differentiating “hostile” sexism from “benevolent” sexism. Hostile sexism represents antagonistic attitudes toward women, while benevolent sexism reflects paternalistic ideas about women’s role in society that relies on a “kinder and gentler justification of male dominance and prescribed gender roles” (Glick and Fiske Reference Glick and Fiske1996, 121).

Past studies show that hostile sexism is correlated with opposition to gender quotas (Keenan and McElroy Reference Keenan and McElroy2017) and women’s candidacy (Winter Reference Winter2022). By contrast, benevolent sexism can lead to support for gender quotas, because these can be a means to foster gender complementarities and stereotypes about women’s purity. Batista Pereira and Porto (Reference Batista Pereira and Nathália2020) show that those who embrace benevolent sexism tend to reject gender equality but support gender quotas. Similarly, Beauregard and Sheppard (Reference Beauregard and Sheppard2021) find that benevolent sexists are more likely to support gender quotas for “condescending and patronizing reasons,” such as that women need men’s assistance and protection. Women’s contribution in the private sphere justifies their participation in politics, as they are expected to bring “women’s perspectives” into policy making.

While these two well-used measures of sexism—hostile and benevolent—focus on women’s appropriate roles in society generally, modern sexism may be more relevant to understanding attitudes toward gender quotas. Modern sexism captures how respondents perceive and/or dismiss the very existence of gender discrimination and how this shapes their resentment toward those who are unhappy with the status quo. The original Modern Sexism Scale, developed by Swim et al. (Reference Swim, Aikin, Hall and Hunter1995), consists of components that measure denial of continuing discrimination, antagonism toward women’s demands, and resentment about special dispensations toward women. The Modern Sexism Scale has been used to understand resistance to equality measures in various organizations, including universities (cf. Skewes, Skewes, and Ryan Reference Skewes, Skewes and Ryan2019) and workplaces (Archer and Kam Reference Archer and Kam2021), as well as concrete political behavior including voter turnout (Kam and Archer Reference Kam and Allison2021).

In the context of the principle-policy puzzle, modern sexism is particularly pertinent to understanding why some citizens support gender equality but nevertheless oppose quotas. They may believe that structural barriers to women’s entry have been resolved or that remaining inequalities are due to differences in the career preferences and efforts of women. They may also not agree with positive action measures as a remedy because of concerns about reverse discrimination.

For the purposes of our research, we measure modern sexism using survey items that have direct relevance to political processes and gender quotas in electoral democracy, rather than broader attitudes to gender inequality as captured by the Modern Sexism Scale. Two relate to beliefs about the reason for women’s underrepresentation. Prior research reveals a gender gap in people’s understanding of the causes of women’s relative absence. Women are more likely to attribute underrepresentation to structural-level factors, whereas men tend to focus on the level of individual women (Barnes and Córdova Reference Barnes and Córdova2016; Meier Reference Meier2008). Those who find fault with structural impediments, such as the gatekeeping role of political parties or the limited availability of resources such as money and time, are more likely to support gender quotas (Keenan and McElroy Reference Keenan and McElroy2017).

In order to examine the association between the perceived causes of gender imbalances and support for quotas, we directly asked respondents whether they believe that not enough women are recruited by political parties and/or are not interested in politics. From the foregoing discussion, we construct the following hypotheses.

Beliefs about the effects of gender quotas on the quality of representative democracy should also affect citizens’ approval of such means. One naive fear of quotas is that they will lead to more unqualified people in office. A number of studies on gender quotas reveal that such changes do not take place after their adoption. Instead, women who are equally—if not more—qualified than men counterparts are more likely to step into a political career (Allen, Cutts, and Campbell Reference Allen, Cutts and Campbell2016; Besleyet al. Reference Besley, Folke, Persson and Rickne2017; Childs and Webb Reference Childs and Webb2012; Huang Reference Huang2016; Murray Reference Murray2010; Nugent and Krook Reference Nugent and Krook2016; O’Brien Reference O’Brien, Franceschet, Krook and Piscopo2012). In the Japanese context, Kage, Rosenbluth, and Tanaka (Reference Kage, Rosenbluth and Tanaka2019) show that voters do not find women candidates to be less competent than their men counterparts.Footnote ⁷

However, some voters may fear that the introduction of gender quotas would give unfair opportunities for unqualified women to win seats, as they may be unaware of findings in political science that deny such myths. Therefore, we construct the following hypothesis.

Government Interventionism

Even if one believes that women’s inequality is undesirable and its root causes are structural, there may be misapprehension about mandating gender parity through legislation. These relate to concerns about government overreach that are often captured as differences among progressives and conservatives (or libertarians) for “big” versus “small” government. Earlier research suggests that this dimension has important implications for gender quotas as well. Kane and Whipkey (Reference Kane and Whipkey2009) find that beliefs about the role of government in limiting economic stratification predict both men’s and women’s support for gender-related affirmative action in the United States. Similarly, Barnes and Córdova (Reference Barnes and Córdova2016) show that citizens’ normative beliefs about the role of government are important predictors of support for state-mandated gender quotas, independent of sex, gender-egalitarian attitudes, and ideology.

We take this view one step further and measure beliefs about the appropriate roles of government in social policy. Citizens have distinct preferences about the degree to which government should intervene in the private sector to realize egalitarian outcomes. As far as gender equality is concerned, government can affect not only women’s participation in politics, but also women’s social status on matters such as marriage, parenting, or sexual harassment.

We thus hypothesize that those who support government intervention to ensure gender equality in general should be more likely to also support quotas to achieve gender-balanced representation in parliament. Mandatory quotas are a particularly strong intervention, as political parties are forced to nominate a certain number of women candidates. Those who agree with state intervention, particularly in sociocultural matters, should be more comfortable with such laws, whereas those who are skeptical of government intervention should be more hesitant. Our survey incorporates items to assess respondents’ attitudes toward such intervention in the context of gender.

Benevolent versus Hostile Sexism

Views on equitable gender representation are shaped by fundamental social attitudes toward the role of women in society. Our theoretical focus is the role of “modern sexism,” as discussed earlier, but we begin by confirming whether patterns found between ambivalent sexism and quota support in other contexts also apply to Japan. We analyze the relationship between benevolent and hostile sexism and Diet quota, drawing on a battery of questions taken from the Japanese translation of the Ambivalent Sexism Inventory. Ui and Yamamoto’s (Reference Ui and Yamamoto2001) instrument includes 22 items, from which we select 6 with the highest factor loadings. Responses were given on a 6-point Likert scale (cf. Appendix Table B2). The first three items relate to benevolent sexism, while the next three relate to hostile sexism.

In order to estimate latent dimensions of these two sexism types, we calculate the mean responses to the benevolent (Q1–3) and hostile (Q4–6) sexism questions. Scales are reversed from the original, such that higher values denote greater sexism. On benevolent sexism, the mean response for men is 3.46 and for women is 3.33; their difference (–0.13) is statistically significant at conventional levels (t = 4.07). On hostile sexism, the mean response for men is 3.11 and for women is 2.65; their difference (–0.46) is also statistically significant (t = 12.55). As the level of mean responses suggests, benevolent sexism is more prevalent than hostile sexism. The correlation between these two variables is a moderate 0.38.

We run separate models for men and women respondents, regressing Diet quota against the benevolent and hostile sexism scores. In lieu of predicted levels of quota support, Figure 2 shows the marginal change in Diet quota from a 1-point increase in each sexism score. For both men and women respondents, hostile sexism reduces support for gender quotas at statistically significant levels. On benevolent sexism, however, men with higher scores are more likely to support quotas, but the coefficient for women is statistically insignificant. In a separate model that collapses men and women respondents together, we find that benevolent sexism overall has a positive and statistically significant effect on Diet quota. Raw regression results, including for models that include partisan affinity and age as control variables, can be found in Appendix Table D1.

Figure 2. Benevolent and hostile sexism and quota support. Figure shows the marginal change in support for legislative gender quotas by benevolent and hostile sexism, estimated separately for men and women respondents. Estimates for men respondents are in blue; those for women respondents are in red. Markers denote point estimates, and bars represent 95% confidence intervals.

Our result confirms the pattern observed in other countries such as Ireland (Keenan and McElroy Reference Keenan and McElroy2017), Brazil (Batista Pereira and Porto Reference Batista Pereira and Nathália2020), and Australia (Beauregard and Sheppard Reference Beauregard and Sheppard2021). The addition of Japan in the study of sexism and quota support indicates the robustness of the impact of these two variants of sexism. We also find that the impact of benevolent sexism on women is insignificant. This suggests that Japanese women do not consider gender differences in virtues or attributes to be a reason to ensure more women in politics, whereas men do, perhaps for paternalistic reasons. This may reflect the asymmetrical position of men and women in Japanese society.

Modern Sexism

We now move to our test of three hypotheses about modern sexism that directly address beliefs that underlie support or opposition to gender quotas. Citizens’ preferences are likely to vary with their perception of why women’s representation has been impeded. If they believe that the cause is attributable to structural factors, such as the failure of parties to actively recruit women candidates ( H₁ ), then they may see quotas as a reasonable means to ameliorate gender imbalances. If they believe that the fault is with women themselves, such as their lack of interest in politics ( H₂ ), then they may be less favorable.

To test these relationships, we asked respondents whether they agreed (dichotomous yes/no) with two statements relating to the perceived causes of women’s underrepresentation.

1. (1) Few women are interested in politics. (69.6% agreement)

2. (2) Political parties are not serious in their recruitment of women. (78.9% agreement)

H₃ posits that respondents would be less likely to support quotas if they perceive the cure to be worse than the disease. We asked respondents whether they agreed with the following statement on a 6-point Likert scale, which we converted into a binary indicator so that coefficients are comparable to those for (1) and (2):

1. (3) If quotas become compulsory, the number of unqualified women candidates will increase. (67.2% agreement)

We include all three variables simultaneously as covariates in ordinary least squares regressions, with Diet quota as the dependent variable. Figure 3 shows the marginal change in support for quotas when each variable takes the value of 1 (i.e., agreement).Footnote ⁹ All have clear, predicted effects on quota support, with similar magnitudes across gender. Regarding the causes of women’s underrepresentation, those who attribute it to women’s lack of political interests are –0.34 (men) and –0.32 (women) points less likely to support quotas. The effect estimates are reversed for those who believe that parties are not recruiting women seriously, at +0.92 and +0.87 points greater support, respectively. In substantive terms, the starkest relationship is for unqualified women. Among men who agree that quotas will increase the number of unqualified women legislators, support for gender quotas falls by an astounding –1.02 points; for women, the coefficient is –0.79.Footnote ¹⁰

Figure 3. Modern sexism and support for quotas. Figure shows the marginal change in support for legislative gender quotas (6-point Likert scale) by perceived reasons for women’s underrepresentation and the potential effects of gender quotas, conditional on respondent gender. Responses by men respondents are in blue; those by women respondents are in red. Markers denote point estimates, and bars represent 95% confidence intervals.

Collectively, the results presented in Figure 3 indicate that modern sexism is a strong predictor of support for gender quotas. Those who attribute women’s underrepresentation to women themselves do not think that quotas are necessary, while those who associate it with political causes such as stunted elite recruitment are in favor of quotas. Consequences also matter: those who believe that quotas will decrease the quality of political representation oppose quotas. While this linkage is perhaps not surprising, it speaks to the need for more communication, particularly by academics, that this supposed relationship is not borne out by comparative evidence. We return to this point in the conclusion.

Government Interventionism

A separate reason for opposition to gender quotas may be apprehension about inviting state intervention in political affairs, particularly on the decision-making freedom of political parties. Even if voters seek more diverse legislatures, they may not see direct government mandates as an appropriate price to pay—that is, the ends may not justify the means. By contrast, those who generally support direct government intervention may be more likely to also back gender quotas, as posited by H₄ .

To isolate the effects of beliefs about the appropriateness of government intervention, as opposed to general left-right political ideology, our survey instrument included four questions relating to statutory requirements to improve gender diversity in nonpolitical spheres. These include support for (1) mandatory paternity leave, (2) same-sex marriage, (3) not requiring surname changes upon marriage, and (4) penalties for sexual harassment. The breakdown of responses to these questions is available in Appendix Table B3. While attitudes toward government activism in the private sphere can span multiple policy dimensions, including religion, race/ethnicity, and education, we purposefully chose items that relate to gender, so as to better assess its linkage to gender quotas in particular.

We use these responses to estimate a single latent dimension of support for government intervention, based on a graded response model that draws on item response theory. Interventionism has a mean of 0.00 and a standard deviation of 0.82; higher values are correlated with greater agreement with each of the four questions above. For ease of analysis, we collapse this latent dimension into quartiles, which we label “strong passive,” “weak passive,” “weak active,” and “strong active.”Footnote ¹¹

Figure 4 depicts the predicted level of Diet quota at each quaternary level of interventionism, conditional on gender.Footnote ¹² As support for government intervention increases, so does support for gender quotas. Among men, the predicted levels between “strong passive” and “strong active” interventionists are 3.01 and 4.22, respectively; for women, this is 3.29 and 4.59, respectively. In other words, Diet quota increases by more than 1 point (on a 6-point scale) between those who strongly oppose versus support government intervention. This corroborates our contention in H₄ that a key explanatory factor in support for quotas is beliefs about the appropriateness of government intervention to achieve greater gender egalitarianism in the legislature.

Figure 4. Government interventionism and quota support. Figure shows the predicted level of support for legislative gender quotas by support for government interventionism (4-point scale), conditional on respondent gender. Responses by men respondents are in blue; those by women respondents are in red. Solid lines denote predicted levels, and dashed lines represent 95% confidence intervals.

Principle-Policy Puzzle

Not all citizens who seek more women in the legislature support gender quotas. The literature has long noted this “principle-policy puzzle”: those who want more political diversity—whether based on race, religion, or, in our case, women—do not necessarily believe that this should be achieved through quotas. To estimate the extent to which this phenomenon is pertinent to Japan, we examine how the effect of our main explanatory factors—modern sexism and interventionism—are moderated by a desire for a more egalitarian legislature.

Our survey asked respondents for their ideal percentage of women in the Diet (range 0–100). The average percentages were 37.8 for men (median = 35, SD = 13.8) and 40.3 for women (median = 40, SD = 12.9). In general, those who desire more women politicians also support quotas, with a correlation coefficient of 0.324. That said, while 58.3% of respondents believe that women should comprise more than 40% of the Diet, of those, 23.9% nevertheless oppose gender quotas.

Here, we estimate Diet quota as a function of the interaction between respondents’ stated ideal percentage of women and our three variables for modern sexism (women uninterested, parties unserious, and unqualified women) and one variable for government intervention, as well as their respective additive terms. For ease of comparison, we convert intervention into a binary term that equals 1 if the original latent score is greater than 0, and 0 otherwise. Separate regressions are run for each of the four key variables. The raw regression coefficients can be found in Appendix Table D2.

Figure 5 presents the results as four separate panels, one each for our four explanatory factors. Each line depicts the predicted level of Diet quota as a function of the ideal percentage of women in the Diet, conditional on agreement (black) or disagreement (gray) with each sentiment. Dashed lines are 95% confidence intervals. There are three points to note about these results. First, all slopes are positive, indicating that support for quotas increases among those who seek more women in the Diet. Not surprisingly, those who desire a more egalitarian legislature are generally more supportive of quotas as a means to achieve that aim.

Figure 5. Quota support and the principle-policy puzzle. Figure shows the predicted level of support for legislative gender quotas by the desired percentage of women in the Diet, conditional on agreement with the following four statements. Top-left panel: women’s underrepresentation is due to women’s disinterest in politics. Top-right panel: women’s underrepresentation is due to the lack of pro-egalitarian recruitment by political parties. Bottom-left panel: quotas will increase the number of unqualified women politicians. Bottom-right panel: support for active government intervention in gender affairs. Black lines denote those who agree with the statement; gray lines denote disagreement. Solid lines are predicted levels, and dashed lines are 95% confidence intervals.

Second, the slopes conditional on statement agreement are roughly parallel, suggesting that the main distinction is that of levels or intercepts. There are baseline differences in Diet quota, such that those who espouse modern sexism and oppose government interventionism are less supportive of gender quotas. This difference remains even as the desired percentage of women legislators increases, although the lines begin to converge for the variable women uninterested.

Third, the level of ideal percentage of women at which each slope is significantly above 3.5 (the middle level on the 1–6 scale) varies by the moderator. Among those who agree that unequal representation is due to women’s disinterest (top-left panel, black line), support for quotas goes into positive territory (above 3.5) when the ideal percentage of women is above 37%. For those who disagree that parties are unserious about recruiting women candidates (top-right panel, gray line), the predicted level is higher than 60%. For those who agree that quotas will increase the number of unqualified women legislators (bottom-left panel, black line), it is 45%. Finally, for those who are opposed to government interventionism on gender matters (bottom-right panel, gray line), the ideal percentage of women in the Diet needs to surpass 43% before the predicted level of Diet quota is in positive territory.

The main takeaway from this analysis is that to increase support for gender quotas, convincing citizens to support more women in the legislature is insufficient. For those who reject modern sexism and are open to government interventionism in gender matters, support for quotas is high at even moderate levels of desired parliamentary egalitarianism. Instead, the key hurdle is to convince people that parties can still do more to nominate women, that structural barriers for women legislators are real, and that gender quotas will not produce less qualified representatives. The first factor—elite recruitment—may deserve particular attention, given that those who think parties are doing enough already are least likely to support quotas, even when they want more women legislators (Figure 5, top-right panel).

As we discuss next, these attitudes may change in response to educational campaigns that inform citizens about structural societal constraints and rectify unwarranted concerns. The same may be said about convincing people to accept greater government intervention, but this attitude may be closer to a more fundamental political value that is less mutable. At a minimum, more research is necessary to determine how preferences about positive action are shaped, both in general and more specifically in Japan.