Gender Differences in Policy Preferences of Legislators: Evidence from China’s National Legislature

Abstract

Do female legislators have different policy preferences than male legislators? Despite a large body of literature from liberal democracies and recent studies of electoral authoritarian regimes, this topic has received little attention in the context of single-party regimes. Based on quasi-experimental methods and regression models, we analyze original data from 38,383 proposals introduced during China’s 12th National People’s Congress and test the effect of gender on policies concerning conventionally selected feminine issues and “political stance,” issues that are unique to single-party regimes. The analysis confirms the effect of gender on policy preference across several feminine issues. However, the effect of gender is null on issues concerning political stance. Our findings suggest that while single-party regimes allow gender differences to emerge among legislators on issues that are not politically important, they tend to discourage such differences on politically prominent issues. This study advances the literature on both gender politics and authoritarian politics.

Literature from liberal democracies has provided compelling evidence to support an independent effect of gender on legislators’ performance, even when controlling for factors such as constituency needs, party membership, and ideology (Ramstetter and Habersack 2020). The underlying mechanism is that women overall have different experiences than men, and thus they bring different perspectives to the legislature (Lowande, Ritchie, and Lauterbach 2019; Phillips 1995; Piscopo 2011). Consequently, female legislators are found to hold more liberal policy preferences, they are particularly prone to spending on social welfare (Barnes 2012; Funk and Philips 2019; Mechkova and Carlitz 2021), and they are usually more concerned with feminine issues than male legislators are (Gottlieb, Grossman, and Robinson 2018; Schwindt-Bayer 2006; Shim 2021).

Does the same relationship between gender and legislative performance apply to legislatures outside of liberal democracies? Existing research suggests that female legislators in authoritarian regimes do push hard for women’s rights and interests: they have successfully campaigned for women’s greater participation during Iran’s Sixth Majles (parliament) (Moghadam and Haghighatjoo 2016), established a legislative agenda to tackle gender-based violence in autocratic Rwanda (Bauer and Burnet 2013), and lobbied for China’s first law against domestic violence (Jiang and Zhou 2022). Despite their importance, these case-based works focus mainly on the role of female legislators in advancing women’s interests. However, studies on whether women perform differently than men in authoritarian legislatures remain limited, and they are mostly conducted in electoral authoritarian regimes such as Singapore and Hong Kong (Tam 2017, 2020) or in developing democracies and regimes in transition (Clayton and Zetterberg 2021; Tønnessen and al-Nagar 2013). Because of data limitations, research on single-party regimes or monarchies is rare and preoccupied with women’s descriptive representation—that is, the number of women in parliament or their committee assignments (Joshi and Thimothy 2019; Shalaby and Elimam 2020).

To address this oversight, this study examines the relationship between gender and policy preference in the legislature of a single-party regime, the National People’s Congress (NPC) of China, testing whether gender affects proposal making within the newly opened space for genuine politics in this single-party legislature. The space is “newly opened” in the sense that while the NPC used to be a “rubber stamp,” it has become more assertive over the past few decades (Cho 2002; O’Brien 1994b), allowing for genuine politics, including legislators substantively representing their constituencies (Chen 2022; Manion 2015), building coalitions with ruling elites with competing policy preferences (Jiang and Zhou 2022; Lü, Liu, and Li 2020), and even holding up the legislative process to create gridlock (Truex 2020).

To do so, we refer to our compiled data set of 38,383 proposals submitted to the 12th NPC of China (2013–17) and utilize the quasi-experimental methods of entropy balancing and difference-in-difference (DID) and regression models. Additionally, the study relies on fieldwork conducted in China between 2016 and 2017 involving local congressional members in the provinces of Hunan and Hubei. Specifically, this study tests the effect of gender against issues traditionally associated with femininity as well as those unique to single-party regimes. We show that in a single-party legislature, gender does affect policy preferences concerning feminine issues: women are more likely to raise proposals related to women, children, family, education, and aging.¹ However, we find no such effect on legislators’ “political stance,” a concept rooted in China’s single-party rule and referring to a legislator’s closeness to national interests, Chinese Communist Party (CCP)’s ideology, and the party’s policy preferences.² In this study, we measure political stance using proposals on constituency interests and poverty alleviation.

While this research is based on China’s NPC, it generates broad theoretical and methodological implications. Theoretically, we address a long-existing gap in the literature of gender and politics attributable to data limitations by showing that the effect of gender on feminine issues transcends regime types and that, despite the ruling party’s control of the legislature, women in single-party regimes bring different perspectives rooted in their experiences (Forman-Rabinovici and Sommer 2019). Moreover, this article reveals that gender differences exist in single-party legislatures concerning feminine and politically unimportant issues but not politically prominent issues, thereby advancing knowledge on single-party and authoritarian legislatures, which is in a nascent stage of development because of limited data/fieldwork access and the assumption that legislatures do not matter in such regimes (Gandhi, Noble, and Svolik 2020; Williamson and Magaloni 2020). Methodologically, researchers of gender differences in policy preferences typically select issue areas related to femininity (e.g., children, education, health care) while dismissing political and cultural contexts. This study follows a bottom-up approach in its selection of policy areas and contributes new outcome variables rooted in China’s single-party legislature that measure one’s political stance for future studies of gender politics, authoritarian politics, and Chinese politics.

Proposal Making in China’s National Legislature

Known as the NPC, China’s national legislature consists of about 3,000 legislators elected mostly by the Provincial People’s Congresses.³ Among all legislative activities, proposal introduction works as the most important way for NPC legislators to express their policy preferences. Compared with bills, proposals have a much lower threshold: they can be introduced by as few as one legislator and need not go to the floor, demanding only a response from the relevant CCP or state agencies, whereas, once passed, bills become laws or resolutions. Because of their importance, bills that have reached the floor are those that have won enough support from the ruling elites, and votes are therefore largely unanimous (Tanner 1995). Consequently, legislators tend to put forth many more proposals than bills, and proposal introduction becomes a safer way of signaling preferences than casting a nay vote on a bill during the last stage (Lü, Liu, and Li 2020).

Proposal making in the NPC both differs from and resembles doing so in liberal democracies. First, much like in other single-party regimes, the NPC is subjugated organizationally and politically to the rule of the CCP, with the nomination and election of legislators very much controlled by the party (Jiang 2022; Tanner and Chen 1998). Thus, unlike its counterparts in liberal democracies (Dingler and Ramstetter 2023; Mummolo, Peterson, and Westwood 2021), the NPC and its members seldom challenge the party. Critiques are raised without threating regime stability (O’Brien 1994a; Truex 2016). Often, the NPC works closely with the CCP, offering support for the party and its policies. Doing so has earned China’s legislators a reputation as “regime agents” and “remonstrators” who explain government policies to their constituents and help rectify administration (O’Brien 1994a). As a result, even though the congresses, as an organization, have certainly consolidated their place in Chinese politics, the idea of working with rather than initiating conflict with the party remains relevant (Ma and Lin 2015; Xia 2000).

Second, despite the CCP’s influence, the NPC has undergone important changes since its establishment in 1954, making it much more like its counterparts in liberal democracies in that the legislative process is now susceptible to lobbying efforts by both state and nonstate actors (Chen and Huang 2023; Jiang and Zhou 2022; Kennedy 2009; Mertha 2009). Two mechanisms are particularly important in shaping Chinese legislators’ proposals: parochial politics and elite politics. First, legislators perform substantively for parochial or constituency interests (Manion 2015; Truex 2016; Zuo 2022), and some of them have successfully brought more resources, projects, and preferential policies to their constituencies (Chen 2019, 2022). Second, the policy-making process in China is a battlefield for elite bargaining, and coalitions often arise between legislators and major stakeholders, including ministries and local bureaus. Meeting only once a year, China’s unpaid and part-time legislators devote the majority of their time to their full-time jobs outside the legislature.⁴ Because of that, legislators are easily influenced by deeply divided bureaucratic interests (Mertha 2009), and legislators’ employment ties often work as a basis for coalition building (Lü, Liu, and Li 2020). Ministries and other workplaces purposely seek proxies who share employment ties with them to advance their bureaucratic preferences in the legislature. For instance, the All-China Women’s Federation (ACWF), a CCP-sponsored women’s organization with more than 2,000 chapters, has famously pushed for the passage of the 1992 Law on the Protection of Women’s Rights and Interests and the 2016 Anti-Domestic Violence Law by endorsing legislators who work for the ACWF or its local chapters (Jiang and Zhou 2022).

What is the implication of gender for Chinese legislators’ proposal making? In arguing that legislators are motivated by parochial or bureaucratic interests, previous studies have essentially suggested that legislators have little discretionary room in which to maneuver when conveying constituency/bureaucratic preferences. Yet persuasive voices are to be found on the other side of this debate, suggesting that personal traits such as CCP membership, gender, and ethnicity do have an impact on legislators’ performance in terms of their degree of “activism” and “parochialism” (He and Liu 2013; Huang and Chen 2015; Zuo, Zhang, and Pan 2020; Zuo 2022). If legislators are able to exert agency in these activities, how might gender influence legislators’ policy preferences when introducing proposals? In the following two sections, we propose hypotheses to test the effects of gender on selected issue areas based on the existing literature and our fieldwork.

Gender Differences on Feminine Issues

The argument for bringing more women into the legislature hinges on the idea that female legislators have different preferences than their male colleagues, as reflected in voting, proposal making, and debates, among other things. Therefore, women’s underrepresentation in the legislature undoubtedly hurts women’s interests, as men might overlook issues such as childcare and maternal employment or hold different opinions on issues that disproportionately affect women. This argument has received increasing support in recent years, with empirical evidence drawn from different parts of the world and across time: studies show that women are more likely to introduce, sponsor, cosponsor, and vote for issues with “feminine” characteristics and straightforward “feminist policies” (Gottlieb, Grossman, and Robinson 2018; Ramstetter and Habersack 2020; Shim 2021; Tam 2017, 2020). For instance, female legislators in Hong Kong ask more questions during plenary meetings on education and women’s rights (Tam 2017). Their counterparts in Argentina and the United States, on the other hand, tend to cosponsor more bills on social welfare (Barnes 2012; Swers 2005). The reported gender difference also emerges in policy outcomes, particularly in spending. Many note that women are more likely to support spending on “feminine” causes (Funk and Philips 2019), including health (Mechkova and Carlitz 2021), drinking water (Gottlieb, Grossman, and Robinson 2018), and education (Barnes, Beall, and Holman 2020).

We propose that the same difference concerning feminine issues arises in China’s NPC. Prior research has typically used two coding schemes when testing gender differences in policy preferences. The most common is to separate “women’s issues” as a category distinct from the rest, which capture “feminist” issues or “strategic gender interests” (Molyneux 1985; Schwindt-Bayer 2006; Tam 2017). These issues derive from women’s subordination. Policies typically intend to overcome the gendered hierarchy through measures such as the abolition of the “sexual division of labor” and the adoption of measures against gender-based violence (Molyneux 1985, 228). Alternatively, scholars have coded issues by gender based on the traditional public/private domains (Barnes, Beall, and Holman 2020; Krook and O’Brien 2012), which entails a horizontal gendered division of labor and distinguishes gender by the “policy areas in which men and women work” (Raaum 1995, 29). Following this method, “feminine” policy areas usually include education, health, and childcare, while “masculine” policy areas include the economy, the military, and foreign affairs (Krook and O’Brien 2012).

Whether they are based on “women’s issues” or on the public/private division, the coding schemes used in previous research have almost become a priori, with little attention paid to differences across cultures, political regimes, and time (Ewig 2018). To avoid coding bias, we combine the two schemes and adapt them to reflect the institutional and cultural characteristics of China. First, instead of coding women’s issues based on a feminist agenda, we do so by determining whether a proposal relates directly to women. Such proposals might not necessarily challenge gender-based hierarchies but help ease women’s burdens within a gendered division of labor (Molyneux 1985). As a result of the CCP’s censorship of feminist agendas, the use of a “feminist” coding criterion would result in zero cases, as state actors, including legislators, have carefully distanced themselves from the label “feminist” in recent years (Jiang and Zhou 2022). Instead, most cases in our data set are drawn from women’s practical interests, such as maternity leave or dowry (Sohu News 2021), as well as less controversial strategic interests, including land rights and hiring discrimination. Therefore, we propose the following:

H_1a: Compared with male legislators, female legislators are more likely to make proposals associated with women’s issues.

Second, in addition to women’s issues, we test issues that are traditionally associated with “feminine” characteristics. Here, we follow the “double definition” coding scheme used by Krook and O’Brien (2012), which relies on both the public/private division and “traditional views on men’s and women’s roles.” The second criterion is especially useful here because it allows for cultural and social variations in views on femininity. In the case of China, on one hand, under the CCP’s commitment to liberate women, the role of women in public spaces has been strengthened, with the educational and economic status of women largely improved (Evans 2021). Chinese women’s labor participation is far above the world average despite a drop in recent years, from 73.24% in 1990 to 60.57% in 2019, compared with the global drop from 51.19% to 47.29% (International Labour Organization 2023). On the other hand, women are subject to both traditional and modern gender norms brought by China’s embrace of “modernity” and “Western culture,” both of which consider women the primary—in most cases the sole—caretaker in private spaces. While traditional gender norms prescribe the role of a “virtuous wife and good mother” (xian qi liang mu) (Robinson 1985), modern norms promote femininity in terms of differences, sexual appeal to men, motherhood, and consumerism (Hershatter 2004).

The definition of femininity in China therefore needs to consider Chinese women’s dual roles in public and private spaces. Thus, we test policy areas that either relate to occupations in which women make up the majority (e.g., education, health) or reflect women’s traditional roles in private spaces. These selection criteria give rise to six policy areas: children, family, education, aging, social security, and health care. We propose the following:

H_1b: Compared with male legislators, female legislators are more likely to make proposals that relate to women’s roles in both public and private spaces.

Gender Differences in Political Stance

Existing studies reveal that authoritarian legislatures are usually established to consolidate authoritarian rule (Boix and Svolik 2013; Gandhi and Przeworski 2007; Magaloni 2008; Manion 2015; Truex 2016). As a result, such legislatures and their members are controlled by authoritarian rulers and seldom challenge them; instead, they tend to be loyal to and cooperative with them, often catering to their policy preferences (Johnson 2016; O’Brien 1994b; Schuler 2020). In this section, we incorporate this feature of single-party legislatures and examine the gender gap in policy issues that are core to China’s single-party regime—that is, issues that measure legislators’ “political stance” (zhengzhi zhanwei). While the CCP has never clearly defined this term, it promotes the term widely in its official discourse.

This dependent variable is selected inductively based on our fieldwork. Between 2016 and 2017, we conducted semistructured interviews with 32 female and 23 male local legislators at county-level congresses from four counties in Hunan and Hubei Provinces. Their occupations varied, including local CCP cadres, primary school teachers, entrepreneurs, and artists. In the interviews, we asked an open-ended question: “Do you think there is a difference between female and male legislators’ proposal making? If yes, what?” Based on the interview data, we summarize the three most frequent answers: “political stance” (n = 18), “women are more active” (n = 10), and “no difference” (n = 6). According to our respondents, male legislators hold higher political stance than female legislators in the sense that men (1) typically focus more on national interests as opposed to women’s preoccupation with parochial interests and (2) are more knowledgeable of recent party policies and priorities and more likely to align their proposals with them.

To further validate the definition of political stance, we cross-checked interviewees’ understandings of the term against official interpretations. While the term “political stance” has existed for decades, Xi Jinping’s rise to power heightened its use, as the CCP constantly encourages its cadres, party members, and state organs to hold a high political stance. Xi commented, “Party committees and leaders at all levels must consciously emphasize politics (jiang zhengzhi), … pay close attention to what the Party center is concerned with, and have a deep understanding of the most important interest of the Party and the country, and the stance that needs to be firmly upheld” (Xinhua News 2020). In the CCP’s official publication, Seeking Truth (qiu shi), a variety of interpretations of political stance have been proposed—for example, to “be knowledgeable of the Party and state’s development, principle, and strategy” and to always “subsume individual/local interests to collective/national interests” (Dong 2018). In one case, local officials were accused of holding a low political stance after it was alleged that they had “pollute[d] the environment despite the central government’s push for improving environmental quality” (Ming 2019).

Even though the CCP has never clearly defined “political stance,” we are able to summarize the most important aspects of this term based on its various interpretations from official news outlets: it signals one’s closeness to national interests and to the CCP’s ideology and policy preferences. In other words, to hold a high political stance means to both think and act like the party. Notably, political stance differs from party discipline in that while party discipline shows one’s party loyalty and “members’ acquiescence to the party’s leadership” (Clayton and Zetterberg 2021, 870), high political stance has a much higher threshold. While political stance also signals loyalty, loyalty is the bottom line of CCP membership, as the CCP requires “absolute loyalty” (juedui zhongcheng) (Wu 2018). Loyalty is therefore only the starting point of political stance. Compared with party discipline, high political stance requires one to go above and beyond and to proactively align oneself with the party’s ideology, interests, and policy priorities.

The cross-check of interviewees’ understandings of political stance and its official interpretations shows that interviewees’ two-dimensional understanding of the term is a subset of its official interpretation. High political stance in proposal making therefore means that one (1) is concerned with national interests as opposed to personal or provincial interests and (2) is knowledgeable about the party’s recent policy priorities and preferences and proactively shapes proposals to align with them (see Table 1). We did not include the ideology aspect for further analysis because interviewees did not report it and because measuring ideology is impractical in a single-party regime, where no one is meant to question the CCP’s ideology (e.g., Marxism and Maoism).

Table 1. Conceptualization of political stance and its operationalization

We select two policy areas as proxy measurements for political stance based on the two dimensions of national interests and party policy preferences. First, we select the policy area of constituency interests—that is, whether a proposal’s concern is beyond or within the immediate jurisdiction of the legislator who submitted it. This measurement is a proxy for political stance in terms of legislators’ concern for provincial/parochial versus national interests. It indicates the “micro/macro” division of one’s political stance, as constituency needs suggest a low political stance because of their limited jurisdictional coverage. Though NPC legislators are allowed to articulate local interests, they are encouraged by the party center and the NPC to raise suggestions and appeals from the perspective of the whole country. As our data show, among the 38,383 proposals submitted by 12th NPC legislators, only about 30% articulated legislators’ constituency interests. Furthermore, as some legislators reported in our fieldwork, female legislators tend to care more about “micro” or “petty” issues that are limited to their neighborhoods or electoral districts, while men tend to be more ambitious in raising issues related to the whole country. Therefore, we propose the following:

H_2a: Compared with male legislators, female legislators are more likely to make proposals concerning their immediate jurisdictions.

Second, we select the policy area of poverty alleviation to measure political stance in terms of whether legislators catered to party policy preferences in recent years. Poverty alleviation was one of the most politically prominent policies during the 12th NPC. Considered one of the three major “fights” initiated by Xi during his presidency,⁵ the “fight against poverty” aimed to lift over 70 million residents above the poverty line before the end of 2020. Since the initiation of this policy in November 2015, the government has rolled out all sorts of financial and personnel resources for its implementation. Cadres (i.e., bureaucrats) have to leave office to help with poverty-alleviation tasks such as conducting surveys on household income to identify households under the poverty line and helping households apply for funds or loans (Zeng 2020). Thus, NPC legislators have been motivated to put forth more proposals about poverty alleviation since the “fight against poverty” was initiated. Figure 1 indicates that both the number and proportion of NPC proposals concerning poverty alleviation increased sharply after 2015. Therefore, this policy can be used to measure legislators’ political stance. Notably, this measurement also resonates with our fieldwork findings, as some legislators consider male legislators to be more informed about the party’s policy priorities, while women are “apolitical,” aiming to solve immediate problems. We hypothesize the following:

H_2b: Compared with male legislators, female legislators are less likely to make proposals on poverty alleviation to cater to the “fight against poverty.”

Figure 1. Number and proportion of NPC proposals about poverty alleviation, 2013–17.

These new outcome variables concerning political stance have several advantages. First, a consideration of political stance in the study of gender gaps in legislators’ proposal making captures one of the characteristics of single-party legislatures: not only do legislators rarely challenge the party, they are encouraged to cater to the party’s needs and priorities. Doing so furthers the understanding of differences between legislatures in different types of regimes. Second, while most gender coding of policies reflects a gendered “horizontal division of labor,” our coding also sheds light on the gendered “vertical division of labor,” which captures hierarchies in politics: high political prestige implies “high visibility” and better “access to resources,” and vice versa (Krook and O’Brien 2012, 842). Political stance touches on the political hierarchy within the CCP, as a lower stance derives from jurisdictional and departmental concerns, while a higher stance aims for a politically and jurisdictionally broader impact. Third, in selecting the policy area with the biggest gender gap based on our interviewees’ experiences working in the legislature, we incorporate cultural interpretations of gender differences into our research methodology. This is a feminist methodological tradition, as it problematizes social experiences that are characteristic of Western societies by and large (Harding 1987) and instead “uses women’s experiences to design research” (Tickner 2005, 7).

Research Design

We created an original data set of 38,383 NPC proposals from 2013 to 2017 based on the NPC Yearbook. The data set contains a nearly complete list⁶ of all the proposals made during the 12th NPC, including the title of the proposal, the year when the proposal was submitted, and the name and delegation of the sponsor.

This research considers two groups of policy issues. The first includes the seven policy issues traditionally regarded as feminine, including “women’s issues” directly related to women and six issues relating to women’s traditional gender roles in both public and private spaces: children, family, education, aging, social security, and health care. The second group of policy issues concerns political stance, including issues of constituency interests and poverty alleviation.

We then code NPC proposals according to whether they belong to each policy issue, identified independently. A proposal may belong to more than one policy issue considered in this study or to no such policy issues. For example, a proposal can belong to both women’s issues and poverty alleviation issues if it aims to decrease the poverty rate among rural women. A proposal is coded 1 for a corresponding variable if it belongs to a certain issue; otherwise, it is coded 0. Each proposal was coded by two student research assistants from a Chinese university majoring in social sciences who had previously been trained by the authors. Students coded each proposal independently based on its title in Chinese; the Krippendorff’s alpha for each policy issue is larger than 0.8, meaning that the intercoder reliability is acceptable (Krippendorff 2004). Had the coding of the two research assistants been inconsistent, one of the authors would have coded the proposal again. Definitions and examples of different policy issues are provided in Table 2.

Table 2. Definitions and examples of different policy issues

* Social special care refers to special care for soldiers, veterans, and families of martyrs.

** In this article, the constituencies of NPC legislators refer to the provinces represented by the delegations to which those legislators belong.

We also created a data set containing the biographies of the 12th NPC legislators. The two data sets containing NPC proposals and legislators’ biographic information were combined; the unit of analysis is the legislator. NPC legislators belonging to the delegations of Hong Kong, Macao, Taiwan, the People’s Liberation Army and the Armed Police, as well as legislators who did not complete their five-year term, have been excluded from the analysis.

Entropy balancing, a quasi-experimental method, is utilized to reveal gender differences in NPC legislators’ policy preferences. This method, proposed by Hainmueller (2012), aims to balance a series of covariates between the treatment group (female legislators) and the control group (male legislators). Entropy balancing is based on a maximum entropy reweighting scheme, which is used to search for weights of observations in the control group that satisfy a potentially large set of balance constraints. After reweighting the observations, entropy balancing can achieve a nearly perfect covariate balance between the treatment and control groups and retain all possible information in the preprocessed data (Hainmueller 2012; Hainmueller and Xu 2013).

The covariates balanced in this study include ethnicity,⁷ age,⁸ education level,⁹ party affiliation,¹⁰ occupation,¹¹ level of employment affiliation,¹² Standing Committee membership,¹³ membership on the nine special committees,¹⁴ legislator of the last NPC, number of proposals,¹⁵ and province dummy variables. The first moments (i.e., means) of all covariates are adjusted. Summary statistics of the treatment variables and covariates are shown in Table A1 in the online appendix.

Based on the data set reweighted by entropy balancing, we first compare the per capita number of proposals concerning each issue between female and male legislators during their five-year term. Then, negative binomial regressions are applied to examine relationships between legislators’ genders and the number of proposals those legislators submit on different policy issues during their five-year term, as the dependent variables are count variables with a large number of 0 values. Furthermore, linear regressions are used to examine the effects of legislators’ genders on the proportion of proposals concerning certain policy issues in all proposals submitted by those legislators during their five-year term. In all of these regression models, covariates need not be added because they have been balanced by entropy balancing.

Furthermore, to examine whether female legislators are less likely than male legislators to cater to politically prominent policies, we utilize the “fight against poverty” initiated by the CCP in November 2015 as a treatment to construct a DID design, another quasi-experimental method. Based on a panel data set of 12th NPC legislators, we regard male legislators as those being treated and the period 2016–17 as the post-treatment period,¹⁶ and we examine whether the DID effect exists. The unit of analysis is the legislator-year. Linear regressions are run on the number and the proportion of proposals concerning poverty alleviation in a certain year, respectively. The interaction between male legislators and the post-treatment period is regarded as the DID variable after controlling for variables indicating male and post-treatment or controlling for legislator and year fixed effects; the legislator’s age in a certain year and the number of proposals in a certain year are also controlled. Summary statistics are shown in Table A2 in the online appendix.

Results

Figure 2 shows the per capita number of proposals on different policy issues raised by female and male NPC legislators during their five-year term, counted from the reweighted data set. This indicates that during their five-year term in the 12th NPC, female NPC legislators put forward 0.311 proposals about women’s issues per capita, while male legislators submitted only 0.111 such proposals per capita. Female legislators proposed 0.640 proposals about children’s issues per capita, while male legislators proposed only 0.340. Female legislators per capita raised 0.411 proposals about family issues, but their male colleagues put forth only 0.210 such proposals per capita. Female legislators submitted 1.951 proposals on education issues per capita, while male legislators put forth only 1.277 such proposals per capita. Additionally, female legislators per capita raised 0.754 proposals about aging, but their male colleagues raised only 0.543 per capita. Moreover, female and male legislators, respectively, put forward 0.856 and 0.851 proposals on social security issues per capita and 1.978 and 2.028 proposals about health care issues per capita. Female legislators submitted 5.461 proposals about constituency interests per capita, while their male colleagues raised 5.771 such proposals per capita. In terms of poverty alleviation, female and male legislators submitted 0.829 and 0.963 relevant proposals per capita, respectively.

Figure 2. Per capita number of proposals on each policy issue submitted by male and female legislators during their five-year term in the 12th NPC, 2013–17. All covariates have been balanced by entropy balancing.

Tables 3 and 4 indicate the results of the regression models with the reweighted data set related to proposals about feminine issues. Models 1–7 in Table 3 are negative binomial regression models run on the number of proposals on the seven types of feminine issues. Models 8–14 in Table 4 are linear regression models run on the proportion of relevant policy issues.

Table 3. Negative binomial regressions on number of proposals about feminine issues

Notes: All covariates have been balanced by entropy balancing. Incidence rate ratios are reported and linearized standard errors are in parentheses. * p < .05; ** p < .01; *** p < .001.

Table 4. Linear regressions on proportion of proposals about feminine issues

Notes: All covariates have been balanced by entropy balancing. Regression coefficients are reported and standard errors are in parentheses. * p < .05; ** p < .01; *** p < .001.

The incidence rate ratios in Table 3 show that, on average, female legislators raised 2.8, 1.884, 1.96, 1.528, and 1.387 times more proposals on women’s issues, children, family, education, and aging than male legislators, respectively. However, a significant gender difference did not exist in the number of proposals on social security and health care.

Moreover, Table 4 demonstrates that female legislators raised significantly larger proportions of proposals on women’s issues, children, family, education, and aging than their male colleagues. The regression coefficient shows that female legislators raised 1.3 percentage points more proposals about women’s issues than male legislators. The proportions of proposals about children, family, education, and aging raised by female legislators were, on average, 1.8, 1.3, 2.6, and 1.5 percentage points larger, respectively. However, female legislators did not put forward significantly larger or smaller proportions of proposals on social security and health care.

Notably, 15 female legislators in the 12th NPC were leaders in charge of women’s affairs, including leaders of women’s federations and female directors of village committees or trade unions. These legislators may have been more active in putting forth proposals on feminine issues as a result of their positions rather than their gender. Thus, in the robustness test, we exclude these legislators from the sample, reweight the subsample data through entropy balancing, and rerun regressions on the number and proportion of relevant proposals to exclude the effects of female legislators’ positions as leaders in charge of women’s affairs. The results of the robustness test are shown in Tables A3 and A4 in the online appendix; they indicate that all the results are robust after these legislators are excluded.

Tables 5 and 6 show the results of regression models related to proposals concerning issues of political stance. Models 15 and 17 in Table 5 are negative binomial regressions on the number of relevant proposals, while Models 16 and 18 are linear regressions on the proportions of these proposals. Moreover, Models 19–22 in Table 6 offer the DID estimates for the number and proportion of proposals about poverty alleviation for male and female legislators. Models 19 and 20 are random-effects linear regressions in which the interaction between male and post-treatment is regarded as the DID variable and variables indicating age, number of proposals made this year, male, and post-treatment are controlled. Models 21 and 22 are legislator and year fixed-effects regressions in which the control variables for male and post-treatment are dropped.

Table 5. Regressions on the number and proportion of proposals about issues of political stance

Notes: All covariates have been balanced by entropy balancing. For negative binomial regressions, incidence rate ratios are reported and linearized standard errors are in parentheses. For linear regression, regression coefficients are reported and standard errors are in parentheses. * p < .05; ** p < .01; *** p < .001.

Table 6. Linear regressions with DID estimates on the number and the proportion of proposals about poverty alleviation

Note: Regression coefficients are reported and standard errors are in parentheses. * p < .05; ** p < .01; *** p < .001.

Contrary to our hypotheses and fieldwork findings, there were no significant gender differences in the number and proportion of proposals about both policy issues of political stance. Models 15 and 16 in Table 5 illustrate that female legislators did not submit significantly more or fewer proposals on constituency interests, and the proportion of such proposals that they raised was not significantly larger or smaller than that of male legislators. Additionally, Models 17 and 18 demonstrate no significant difference in either the number or the proportion of proposals about poverty alleviation between female and male legislators.

Furthermore, Models 19 and 20 in Table 6 show that after the “fight against poverty” was initiated, NPC legislators raised a significantly larger number and proportion of proposals concerning poverty alleviation. However, Models 19–22 all demonstrate that the DID estimates for the number and proportion of relevant proposals between male and female legislators are insignificant. This means that both male and female legislators cater to the CCP’s politically prominent policies, and female legislators are not less likely to do so than their male colleagues. These results reveal that gender differences do not exist concerning political stance in NPC legislators’ proposal making.

Conclusion and Discussion

The effect of gender on policy preferences in single-party regimes has been an understudied research topic in the literature of gender and politics because of the dearth of available data (Joshi and Thimothy 2019; Shalaby and Elimam 2020). This article addresses this research gap by examining the link between gender and proposal making in China’s national legislature using an innovative data set recording 38,383 proposals introduced during the 12th NPC (2013–17). Our research confirms that a gender difference in policy preference exists in China’s single-party legislature despite restraints imposed by the CCP. Female legislators are more likely to introduce proposals that directly address women’s issues. Meanwhile, as suggested in prior research on liberal democracies, gender differences are also present in feminine policy areas associated with women’s roles in private and public spaces. We show that female legislators are more likely to put forward proposals addressing issues related to children, family, education, and aging.

However, we did not find gender differences in legislators’ political stance in proposal making, which signals their overall political loyalty and proactivity to align their proposals with national interests and the CCP’s policy preferences—operationalized by policies concerning constituency interests and poverty alleviation. Even though most of the legislators we interviewed insisted that female legislators’ political stance is lower than that of their male counterparts, the data do not support such statements.

These findings contribute to the literature on both gender politics and authoritarian politics. First, despite the differences between liberal democracies and authoritarian regimes, women in both regimes seem to have shared life experiences that drive their overall similar policy preferences. Additionally, the restraints of the CCP still leave room for gender differences to emerge in these areas. There are two explanations for these findings. On one hand, such an interpretation corresponds with recent studies on women’s substantive representation in authoritarian regimes, where ruling elites might have no interest in monitoring these policy areas (Forman-Rabinovici and Sommer 2019), as these policies pose no threat to regime survival (Jiang and Zhou 2022). On the other hand, it is also possible that allowing gender differences to emerge serves the interests of rulers. Like their counterparts in liberal democracies, single-party regimes need to gauge public preferences and solicit information to legitimize their rule (Manion 2015; Truex 2016). Having female legislators propose suggestions specific to women’s needs serves as such an information channel, as female legislators help identify and collect information on women’s grievances using a bottom-up approach. These grievances, if not addressed, could evolve into regime-threatening social unrest. In the case of China’s Anti-Domestic Violence Law, for instance, female legislators presented the NPC with survey results, hotline records, and vivid details of domestic violence cases (Jiang and Zhou 2022). For the CCP, keeping this channel open is critical to the longevity of its rule.

On the other hand, the null effect of gender on our proposed political stance variables is informative in at least three respects. Prior research on feminist institutionalism has proposed arguments regarding institutional effects on gender differences. According to Stensöta, Wängnerud, and Svensson (2015), there are two types of institutional logics: logics that enforce individual and/or group-specific qualities and logics that suppress them. For example, compared with legislatures, bureaucracy is less susceptible to gender differences regarding issues such as corruption because the institution by its nature enhances impartiality and suppresses differences and individual preferences (Stensöta, Wängnerud, and Svensson 2015). The same distinction might explain our findings on political stance. Instead of between institutions, the different logics we have observed exist between different types of issue areas: politically prominent issues versus politically unimportant issues. Political stance, by its very definition, touches on the CCP’s current policy priorities, and thus gender differences in relevant issue areas are discouraged. While legislators are allowed to articulate local interests and raise various policy issues accordingly, male and female legislators in China are equally motivated to be concerned with national interests and offer suggestions on the CCP’s prioritized policies.

The discussion of political stance also speaks to differences across regime types. Existing research shows that one of the biggest differences between liberal democracies and authoritarian regimes is that legislators in authoritarian regimes are ultimately constrained by the nature of authoritarian politics and therefore are not able to propose issues that politically challenge their rulers (Boix and Svolik 2013; Malesky and Schuler 2010; Williamson and Magaloni 2020). This study identifies another behavioral pattern of legislators in single-party regimes: they are incentivized to actively cater to their rulers by signaling their support and recognition of the party’s priorities and interests. Surely, doing so has practical benefits—promotion opportunities,¹⁷ a higher probability for proposals to be addressed, and more funds. Still, this behavior has a great deal to do with the regime’s focus on “unity”; as repeatedly emphasized in the CCP’s official meetings, the party needs to “unify thoughts, unify wills, and unify actions” (tongyi sixiang, tongyi yizhi, tongyi xingdong) (Yang 2023). Given the party’s tightened grip on both local and national politics since Xi’s accession to power and the resultant convergence of policy priorities under his rule (Blanchette 2020; Chan, Lam, and Chen 2021), incentives for NPC legislators to showcase their high political stance are expected to persist, if not become stronger.

Another potential implication of this finding stems from the conflict between legislators’ reports of gender differences in proposal making and our analysis results. The idea that women have a lower political stance might reflect deep-seated gender stereotypes of women as apolitical and indifferent to politics, originating from the association of women with private spaces. Despite a recent improvement in women’s representation in Chinese politics,¹⁸ Chinese politics remains male dominated. The inclusion of women often reproduces instead of challenges this stereotype that women are indifferent to politics, because women are typically assigned to feminine bureaus. As a result, when asked their perceptions of gender differences in proposal making, legislators might build their answers on this stereotype and incorporate it with the most trending political buzzword at the time instead of relying on their actual experiences working in legislature.

Finally, this research provides several pathways for future explorations of gender and politics in authoritarian regimes. First, in this article, we find that single-party regimes may allow gender differences to emerge in politically unimportant issue areas but discourage such differences in politically prominent issue areas. However, what counts as a politically prominent issue varies against different authoritarian backgrounds, and future research might extend this theoretical framework to different authoritarian countries. Second, while this article operationalizes the concept of political stance, it does so with certain limitations. Many have noted that ideological closeness is a crucial part of the concept, but operationalizing ideological closeness during our research remained challenging. We therefore suggest that scholars continue working on ways to measure ideological closeness, possibly through surveys of legislators, and to test gender differences in this regard. Finally, by confirming the effect of gender on policy preference, our research paves the way for future tests of links between women’s descriptive and substantive representation in China, for example, a test of congruence in policy preference between female legislators and female citizens. Not only would such studies add another layer to the continuing debate on the NPC’s transition from “rubber stamp” to “iron stamp,” but they also offer a test of whether the “iron stamp” has different implications for Chinese men and women.