Descriptive Representation and Party Building: Evidence from Municipal Governments in Brazil

Abstract

This article highlights a new way in which descriptive representation enhances democracy through inclusive party building. We theorize that parties retain and promote incumbents based on gendered criteria, disproportionately incentivizing women to recruit party members. However, gendered resource inequalities lower women’s access to the patronage required for recruitment. Women respond by recruiting more women members, as it lowers recruitment costs, is role-congruent, and eases credit claiming. Using rich administrative data on party membership from 2004 to 2020 and a regression discontinuity design in Brazil, we find that, despite resource disparities, women mayors recruit new members at similar rates as men but reduce the gender gap in party membership. As expected, women are more likely to be promoted in constituencies where they most lower the gender gap in party membership. We also find that women’s increased membership improves party resilience. Our findings suggest that descriptive representation strengthens party building by including underrepresented citizens.

INTRODUCTION

Decentralization combined with local democracy has heralded a new era of democratic politics in low- and middle-income countries. Coined as “democracy’s quiet revolution,” these twin reforms have been profound in their reach and in their capacity to shift power from central to local authorities. This quiet revolution has enabled progress on two democratic ideals: descriptive representation and party building. Research in gender and politics shows that descriptive representation enhances democracy by increasing women’s political participation (Desposato and Norrander 2009; Goyal 2023), enhancing substantive representation (Barnes 2016), and improving democratic legitimacy (Clayton, O’Brien, and Piscopo 2019; Hinojosa and Kittilson 2020). At the same time, research on political party development shows that local politics offers national party leaders a tool to screen for talent that will promote the party’s interest, while holding local office offers politicians’ resources to build parties’ territorial organization, paving the way for resilient parties, which are the bedrock of a stable democracy (Levitsky, Loxton, and Van Dyck 2016; Samuels 2003; Sells 2020; Van Dyck 2014).

This article contributes by highlighting a new dimension alongside which descriptive representation enhances democracy: grassroots party building. To establish this link, we offer a gendered theory of incumbency and party building. We theorize that parties retain and promote incumbents based on gendered criteria, incentivizing women to recruit new party members to forge viable political careers. These incentives emerge in two ways. First, parties select, retain, and promote female candidates (and members) who are party loyalists; party loyalists are more likely to invest in party building. Second, female incumbents are also more likely than men to depend on local politics as a pathway to power. Yet, at the same time, gendered inequalities inside party institutions disproportionately lower women’s access to the patronage required for recruitment. Women respond strategically to these constraints by recruiting women party members, as doing so lowers recruitment costs, is role-congruent, and eases credit claiming.

By highlighting the gendered incentives of party building, our theory yields additional expectations for the quality of women’s recruitment and provides us with downstream implications for party resilience. First, because women members are more likely to be party loyal than men members (and have fewer outside options), women’s recruitment as party members increases party resilience. Second, because women incumbents have lower access to patronage and their recruitment strategies are less contingent on patronage, they are more likely to recruit members (men or women) who are also less likely to join for particularistic reasons and, therefore, are less likely to switch or leave parties, which also increases party resilience.

We test these predictions in the context of the municipal government in Brazil. Several reasons make Brazil a near-ideal empirical site for our investigation. Brazil’s municipal governments enjoy considerable discretionary power over budgets and job allocation. As Samuels and Zucco (2016) point out, municipal executive positions are politically appealing and a pathway to both remaining viable in politics and continuing in higher-level politics. Brazil is also a substantively important case where there is substantial scope for party development and has recently experienced greater party consolidation (Levitsky, Loxton, and Van Dyck 2016). At the same time, Brazil has some of the highest gender gaps on women’s representation in Latin America, and research points to weakly institutionalized parties as a major cause of women’s underrepresentation (Wylie 2018). Our theory points to a resolution out of this bad equilibrium, which is particularly important in Brazil, where gender quotas have failed to improve women’s representation (Sacchet 2011).

Empirically, we used administrative data on the party membership of all parties in Brazil in the last two decades, from 2004 to 2020. This uniquely fine-grained dataset enables us to provide descriptives about the gender gap in party membership at scale, which are otherwise hard to study and limit the scale of investigation. Despite the importance of party activism in the politics of the Global South, there is little comparative research exploring the gender gaps in party activist recruitment. Furthermore, to the best of our knowledge, this is the first article to quantify the gender gap in party membership at scale using administrative data in any setting. While we discuss Brazil’s party activism in detail in our context section, we find a stark gender gap in party membership. Women make up roughly 40% of all party members, and this striking gender gap is similar in magnitude across all of Brazil’s major parties (and marginally stronger in less institutionalized or right-leaning parties). We also find that women party members are half as likely as men to switch or leave parties, suggesting that correcting for the gender imbalance in grassroots party membership can simultaneously increase party resilience, at least as measured on this singular dimension.

To identify the effect of women’s incumbency, we rely on a close-elections regression discontinuity design to overcome the selection bias of women mayors (s)electing into places that are more conducive to party building. By using this design, this article joins several other studies that have used the regression discontinuity design to identify incumbency effects and enhances knowledge accumulation (Boas and Hidalgo 2011; Brollo and Troiano 2016; Frey and Santarrosa 2022; Sells 2020). Investigating close races, we find that men and women incumbents are equally likely to increase party membership. In light of evidence demonstrating that women mayors have substantially fewer patronage and financial resources that are important to recruit party members in Brazil (Brollo and Troiano 2016), we interpret this as women incumbents’ either exerting more effort or using the limited available resources to recruit new party members, supporting our theory. In line with the main argument of our article, we find that women mayors lower the gender gap in party membership recruitment, while male mayors have no effect on the gender gap relative to their party being in the opposition. Utilizing party switching as our measure of the quality of new member recruitment, we find that women mayors recruit party members who are less likely to switch or leave parties. Finally, we show that women incumbents have a higher likelihood of being promoted and that these effects are stronger in municipalities that had lower baseline levels of women’s party membership and where women most strongly lower the gender gap in party membership. Investigating the electoral cycle of party building, we find that women lower the gender gap close to the final term year, where among other motivations, the incentives for signaling party building are stronger relative to initial years in office.

This article establishes the link between women’s incumbency and its gendered effects on party building. By establishing this link, we highlight a new dimension alongside which descriptive representation enhances democracy. We extend the research linking descriptive representation to enhanced substantive representation and political participation (Desposato and Norrander 2009; Htun 2016), and present a new possibility alongside which the costs and benefits of descriptive representation can be evaluated. By doing so, we also echo and contribute to a growing scholarship in comparative politics that draws on feminist institutionalist theory to illuminate how gender relations shape women’s behavior inside legislatures and in political parties (Barnes 2016; Clayton and Zetterberg 2021; Htun 2002; Kenny 2014; Krook and Mackay 2011; Sacchet 2011; Wylie 2018). We contribute to the research on the gender gap in political career progression (Folke and Rickne 2016; Goyal Forthcoming; Kerevel 2019; O’Brien 2015), by highlighting that party building can boost women’s chances of securing promotions and renominations.

Our findings are of interest to scholars of political party institutionalization and development (Bohlken 2016; Hicken and Kuhonta 2014; Mainwaring and Scully 1995), especially in Latin America (Hagopian, Gervasoni, and Moraes 2009; Levitsky, Loxton, and Van Dyck 2016; Sells 2020). Nearly four decades since the onset of the third wave, political parties remain weak in Latin America, and most new party building efforts have failed, hindering the prospects of a stable democracy. Our findings show that descriptive representation can enable parties to build territorial organizations and establish more durable partisan attachments, sowing the seeds for long-term party stability. Party stability is in turn crucial to economic growth and democratic stability (Bernhard et al. 2020; Bizzarro et al. 2018).

The Critical Role of Local Leadership in Party Building

Political parties are the basic building blocks of representative democracy, and as Schattschneider and Aldrich famously noted, democracy is “unthinkable” and “unworkable” without political parties. The experience in Africa, Latin America, and Asia shows that successful party building is challenging in new democracies but not altogether impossible (Hicken and Kuhonta 2014; Mainwaring and Scully 1995; Riedl 2014). The scholars of party organization conclude that one of the key elements of successful party building is the construction of a territorial organization and the recruitment of party members, which ensures that party leaders can rely on local political support (Levitsky, Loxton, and Van Dyck 2016). Territorial organization enables parties to win votes and build partisan attachments. Parties use grassroots members and party activists to communicate party brands, build ideological or patronage linkages, and organize rallies and ground campaigns to mobilize voters. Because party members are more likely to stick it out and remain loyal to their party regardless of its electoral success, parties with a strong grassroots activist support base often survive past electoral success and failure.

Yet successful party organizations are not built overnight. A rich literature on party organization in Latin America underscores that successful party building is a resource- and time-intensive responsibility that requires considerable effort on the part of local embedded actors (Levitsky, Loxton, and Van Dyck 2016; Van Dyck 2014). Party membership recruitment requires that local politicians and party functionaries engage actively in grassroots mobilization to forge citizen–party ties. Local leaders schedule neighborhood events such as town halls, plant and recycling drives, meetings, knocks on doors, and organizing transport and buses for party members to attend meetings. In Brazil, local party officials in the PT (Partido dos Trabalhadores; The Worker’s Party) interviewed in Samuels and Zucco (2016, 345) note that, “If you don’t have local leaders who are willing to take on the job of organizing the party, there’s no way a DM [diretório municipal] will take root in that town.” Investigating the importance of local operatives in establishing the PT’s success in the North Eastern region of Brazil, Van Dyck (2014, 11) similarly concludes that, “To make electoral inroads in the NE, parties without strong local patron-client networks, like the PT, need local operatives who can travel deep into the interior, schedule events, hold meetings, claim credit for policies, and actively encourage individuals to support a candidate or ticket.” Personal interactions between candidates and voters have traditionally been crucial in Brazil, particularly at the local level, in growing party membership (Sacchet 2005).

In other Latin American settings, studying the relationship between slum dwellers and local political brokers in Argentina, Auyero (2000, 56) highlights the intricate grassroots processes of mobilization and problem-solving through which mayors and their affiliates, such as brokers and local party functionaries, organize citizens to attend party meetings. He highlights how the mayor relies on several local brokers, such as Norma, “Every month, at the party meetings, the mayor informs us [the brokers of the 140 UBs who usually attend the meeting] of the date when they are going to give out food at the municipality…We tell the neighbors.” Zarazaga (2014, 26) highlights the pyramidal and hierarchical structure through which mayors grow party membership in Argentina. “At the apex are the mayor and an inner circle of two or three people who help build and control the network; these usually include the municipal secretary of government and secretary of social development. Beneath them is a group of municipal delegates or council members who deal directly with the brokers.” Studies of grassroots recruitment in India and Ghana have discovered a similar pyramidal structure of party organization where local leaders are actively involved in growing the party membership base (Auerbach and Thachil 2018; Brierley and Nathan 2020).

Another way to understand the crucial and direct role of local leaders in expanding the party membership is to inquire of party members as to how they came to join parties. Despite the dearth of survey research that includes interviews with party members, a growing number of studies show that party members cite active party recruitment by local party leaders or their affiliates as one of the main reasons to join the party. Hardly a few citizens join parties on their own initiative but instead because they are asked to join a party. Conducting a survey of municipal party convention participants in the state of São Paulo in Brazil, Ribeiro (2015) finds that 74% of members were asked to join the party by another person,¹ while 11% joined due to their participation in a non-party organization such as a labor union or a neighborhood organization, and a mere 8% joined on their own initiative. Further highlighting how local incumbents can leverage their connections with bureaucrats, Frey and Santarrosa (2022) show that mayors in Brazil have an advantage in recruiting street-level bureaucrats into their parties. Outside of Brazil, Goyal (2023) conducts a representative survey with party activists in all major parties in Delhi, India, and shows that the majority of party activists credit active party recruitment as the reason for joining the party. Furthermore, they credit a local politician for connecting them to the party organization; hardly any women or men mention role model effects as a reason to join parties. In Ghana, Brierley and Nathan (2020) find that brokers in Ghana’s ruling party have significantly more upward ties to local elites than non-brokers, underscoring the top-down nature of their party membership.

Given the centrality of local leaders to growing party membership, decentralization and local democracy provide party leaders with a tool to build a robust territorial organization that increases their party’s odds of political survival. Studying local democracy worldwide, Bohlken (2016) finds that party building goals are a key explanation for why party leaders devolve power and extend local democracy in the first place. Party leaders (principals) would like local incumbents (agents)—who are imperfect party agents—to invest patronage and party resources in party building. At the same time, local incumbents are imperfect party agents who can divert these resources to serve their own ends, particularly in weakly institutionalized parties where ideological attachments and disciplining mechanisms are weak (Sells 2020; Van Dyck 2014), jeopardizing the party’s electoral and organizational survival (Novaes 2018).

To guard against this kind of opportunism, party leaders have many ways to lower agency costs, such as their power over re-nomination, promotion, patronage, ideology, norms, and other ways to keep discipline local leaders. Party leaders can use the information from local elections and campaigns to find out how much support there is for the party on the ground and to find talented people who want to help build the party. These tools are particularly effective with strong parties, as seen in many different settings. Sells (2020) explains that holding mayoral office bolsters party building in Brazil, but only in parties that are strongly institutionalized where party leaders have the capacity to discipline local incumbents. He finds that municipal incumbency increased membership recruitment only among centralized and programmatic parties that already had a strong territorial presence in the municipality, and it was ineffective for weaker parties.

We argue that the ways in which party leaders use rewards and punishments to get local leaders to work on building the party are not the same for men and women. Using a feminist institutionalist approach to “gender” party building (Kenny 2014; Krook and Mackay 2011), we argue that the incentives and sanctions that parties use to align local incumbents’ incentives to engage in party building and the resources that incumbents have for accomplishing these goals are not equally distributed between genders. To gender the process of party building means that gender relations influence the underlying logic incentivizing men and women to engage in party building and the ways in which men and women party members are incorporated into party institutions. Not only that, supply and demand lead men and women existing out in the society to be recruited into party institutions (Kenny 2014).²

A Gendered Theory of Party Building

Our argument has two parts. The first part highlights the disproportionate incentives that women incumbents have to engage in party building. The second part highlights how women incumbents lower access to resources that help with recruitment influences their strategy for engaging in party building, with implications for their career progression. In this section, we draw heavily but not exclusively on research on gender and politics from the Latin American context and descriptive data from Brazil in particular to illuminate the building blocks of our argument and provide support for the underlying assumptions of our theory. However, we expect our theory to be more generally applicable, and we revisit the scope conditions in the conclusion.

We theorize that parties re-nominate or promote incumbents based on gendered criteria, incentivizing women to recruit new party members for career progression.³ These incentives emerge in two ways. First, parties select, retain, and promote women who are more disciplined; that is, they are party loyalists who are more likely to toe the party line, which includes investing in party building. Gendered expectations that women will not deviate from party interests increase these tendencies. Second, having fewer pathways into politics, women incumbents are also more likely than men to depend on parties to retain their turf or rise up the party ranks, using party building as a signal to demonstrate their worth to political parties and local politics as a pathway to political survival and power.

We argue that this gendered selection of more disciplined women and gender norms that force women to follow the party line will make women more likely than men to invest in building the party. Party loyalists are more likely to invest in party building, and women are more likely to be party loyalists. Research in comparative politics in a variety of institutional settings, from PR systems to majoritarian systems, shows that women are more loyal to their party than men, doing what the party’s leadership wants, even if it goes against women’s interests. Investigating gender and party identity in Brazil, Htun and Power (2006) conclude that women prioritize party interests over gender interests and are able to achieve feminist objectives only when these coincide with party interests. Party leaders are also less likely to value women’s engagement in deviant behavior, and women face stronger repercussions for deviating from the party line (Morgan and Hinojosa 2018). In Latin America, research finds that women who enter politics through gender quotas may feel additional normative pressure to downplay gender interests to avoid stigmatization and backlash (Htun 2016). Outside of Latin America, conducting a survey with parliamentarians that included several measures of party discipline showed that women were more likely than men to express party discipline across 17 African legislatures. They argue that parties select more disciplined women and gendered expectations about proper behavior limit women legislators’ ability to act independently from their parties, establishing women as stronger party loyalists than men even within the same parties. Moreover, while parties may accept men switching from other parties, they may penalize women for such deviant behavior, further causing women to be more party loyal by restricting their options. In summary, women are more party loyal than men, and therefore we expect them to be more invested in building their party.⁴

In addition to selecting party-loyal women and disciplining them to be more loyal because women have fewer outside options and resources, they are also more reliant on party resources. A vast body of research on gender and political parties shows that women have lower financial capital, weaker informal connections, and sparser political networks than male politicians to begin with (Barnes 2016; Piscopo 2016; Schwindt-Bayer 2010; Wylie 2018). Consequently, women are more likely to rely on party brands and party organizational networks to recruit citizens and grow party membership. Stronger party identification and party-centered political networks further deepen the dependence and bond women have with their parties. Consequently, women are more likely to do the groundwork for reifying the party’s territorial presence. In contrast, having more outside connections and capital to demonstrate their worth to party leaders and mobilize voters, men can substitute away from making these party-specific investments. The criteria for promoting men and women will reflect these gendered dynamics and the sexual division of labor, with the result that women are more likely to be selected, re-nominated, or promoted for party building than men are for the outside connections and resources that they bring to party organization.

Finally, women are also more dependent on local politics to remain politically viable or to progress to higher political levels than men, which makes them more strongly invested in local-level activities such as party building. Worldwide, women are more likely to enter and remain in local governments (Berevoescu and Ballington 2021), suggesting that women may have a greater taste for holding local office, potentially because it offers women more flexibility to manage work-life balance and remain in politics. Despite a gender gap in political career progression, women incumbents in higher-level politics are still more likely than men to have a prior background in local politics (Goyal Forthcoming). Women aspirants rely more heavily on local politics to gain visibility and experience (Clayton and Zetterberg 2021; Folke and Rickne 2016). In contrast, men face fewer challenges and obstacles when contesting directly in higher-level politics, bypassing the grunt work involved in growing party membership and rising through the ranks of party organization altogether. In settings where parties and gender quotas have been ineffective in providing a mechanism for women’s representation, such as in Brazil, women may have an even stronger reliance on local politics to carve a political career (Htun 2002; Sacchet 2011).

The Brazilian case exemplifies women’s dependence on local politics to forge successful political careers. Data from Brazil’s federal-level congressional elections underscore the claim that holding prior local office again benefits women more than men (see Supplementary Tables O.1 and P.1). While holding local office is associated with a greater chance of winning congressional office for both men and women, women who lack local experience are three times less likely to win elections relative to men who lack local experience (see Supplementary Table M.1). However, women with local experience are as likely as men to win congressional elections. In other words, having local experience matters far more for women to win congressional office than men. Furthermore, echoing several studies on the gendered inequalities in campaign revenues in Brazil (Brollo and Troiano 2016; Sacchet 2011), women candidates for Brazil’s Federal Chamber of Deputies who have never held local office receive 37% less campaign donations compared to male candidates who did not hold local office, while women candidates who held the mayorship or another local office receive roughly the same amount of campaign donations as men (see Supplementary Table N.1). Local political office, therefore, matters because it improves women’s ability to attract campaign donations far more than it enables men. Exploring the subset of candidates’ campaign donations that come from either the party’s electoral committee or from other candidates, we find that female mayors who run for Congress receive greater financial support from their parties compared to both male mayors and female congressional candidates who did not previously hold mayoral office. These data suggest that parties do favor women with backgrounds in local office.

Yet Brazilian women may not only want to contest congressional elections but also re-contest as mayors. There can be several reasons for this. Women may have a greater preference for or taste for local politics. Local mayoral offices, particularly in large cities, are more prestigious than state or federal seats and, therefore, even more lucrative (Samuels 2003). However, because mayors in Brazil are limited to two consecutive terms, it is not uncommon for mayors to seek election to the Federal Congress to keep their network alive and rerun for mayor again. Supplementary Table O.1 shows that congresswomen in Brazil’s Federal Chamber of Deputies are 35% more likely than their male counterparts to end their congressional terms early in order to run for mayoral office, which suggests that they have a greater preference for local office.

On the one hand, women have stronger incentives to engage in party building; on the other hand, gendered constraints and inequality in the distribution of party patronage disproportionately lower women incumbents access to the patronage they may require to recruit party members relative to men. Research from across the world shows that women are less able to use political office to access patronage resources than men. Women incumbents are less likely than their male counterparts to accumulate wealth and campaign funds, and some of the strongest evidence for this pattern comes from Brazil’s municipal government. For example, Brollo and Troiano (2016) find that female mayors hire fewer temporary public employees than male mayors during the electoral year and tend to attract fewer campaign contributions when running for reelection. These gender inequalities are striking and on the rise with each election. Sacchet (2011) shows that while in 2002 the average percentage of funds raised by female candidates running for positions of federal deputy was equivalent to 51% of the amount raised by men, in 2014, it represented only 23%. She shows that women are also at a disadvantage because campaign finance is an even more important electoral resource for them. In addition to campaign revenues, providing party members paid opportunities in public employment is a key way in which politicians reward and attract followers in Brazil (Frey and Santarrosa 2022). On both these dimensions through which mayors attract supporters—the ability to secure campaign revenues and the ability to secure public employment for their supporters—women mayors are at a strong disadvantage relative to men.

These gendered constraints have implications for the quality and quantity of women’s party building. With regard to quantity, we expect that women are more likely than men to recruit women party members, as doing so lowers recruitment costs, eases credit claiming, and is role-congruent; this enables women to more effectively engage in party building.⁵ Using interviews to highlight the gender gap in clientelism in Argentina, Daby (2021, 232) notes how women brokers make up for the lack of resources through their effort: “If I see a woman [mina] and guy [tipo] with the same people [female and male brokers mobilizing voters], have no doubt that the woman is a beast.”

Due to several reasons, it is easier and, therefore, less costly for women mayors to recruit other women into their parties compared to men. First, women mayors can use their women-centric networks to recruit women citizens and build coalitions with women’s groups. Women politicians are more likely to conduct their recruitment and political activities in spaces and at times when they are most available and open to conversations, such as school management committees, social groups, religious spaces, and women’s unions, which can further increase their ability to recruit women. One of the first studies highlighting mechanisms underlying women’s grassroots party recruitment, Goyal (2023) shows that women candidates in local politics are more effective at recruiting women party activists because they can simultaneously lower household and party organization constraints to facilitate women’s entry into parties. They are able to recruit women from women-centric networks and spaces, as well as build relational support with household and community members to allow women to join parties and participate in party politics. Second, role model effects can lead women citizens to become more interested in politics and more willing to join the party when an opportunity presents itself (Jalalzai and dos Santos 2015). This can make it easier for women recruiters to persuade women to join the party. Finally, being seen with other women is compliant with women’s social roles, lowers reputation costs and violence concerns (Barnes 2016; Wylie 2018), and is mutually advantageous for both women politicians and party members. Women can also persuade women citizens on the basis of their shared gender identity, a strategy that may seem less credible when used by male incumbents.

Indeed, throughout Latin America, women have used coalition building across political parties and social movements as a means to advance their policy agendas (Daby and Moseley 2022; Sacchet 2005). Evidence from Brazil shows that women attempt to combat their marginalized status through the formation of the Bancada Feminina (the Women’s Bench or Caucus) (Wojcik and Mullenax 2017). Women’s ability to form these alliances at strategic times such as elections has been an important asset that has enabled them to emerge as successful political candidates (Macaulay 1996). In another study in Brazil, Wojcik and Mullenax (2017) found that female legislators not only have stronger networks with women but also tend to foster denser and more diverse networks than their male counterparts.

Recruiting women is a win–win strategy as it enables women mayors and their affiliates to claim credit more exclusively as well as rely on the backing of female supporters to sway party decisions. It is, therefore, not surprising that Jalalzai and dos Santos (2015) find that women are more likely than men to rely on support from party brokers and this support comes into question under times of crisis. By enlisting their friends, relatives, supporters, and clients into their party, a politician signals the depth of their grassroots support to party leaders and gains valuable allies within the party who will support them in the party’s local conventions. As Mainwaring (1999, 257) notes, these local conventions play a central role in internal competition and career advancement within Brazilian parties at all levels:

Politicians make every effort to ensure that they retain control of their bases and expand to new areas. The battle for control of local directorates and executive committees, and especially of the delegates chosen for the state convention, is crucial for ambitious politicians. Local conventions determine who the prevailing local political bosses will be, and they also affect the political strength of national political figures.

Seeing a large number of women attend local conventions and take part in party meetings and campaigns is an easily attributable signal of women’s party building effort. Insofar as social and political networks are gendered, men may find it harder, if not impossible, to claim credit for recruiting women members. At the same time, women are also more likely to support women leaders in local conventions and party events, as they are more likely to perceive women as more credible sponsors of their talent (Preece, Stoddard, and Fisher 2016), and as more supportive of women’s substantive interests (Htun 2002). Further investigating the positive consequences of stronger networks, Wojcik and Mullenax (2017) show that Brazilian legislators who had the largest, most diverse networks tended to receive a larger share of the votes at election time. Because women faced greater barriers to election but won re-election at rates similar to their male peers, they conclude that the robust and diverse networks forged by women helped them to put pressure on party leaders and retain their positions in politics. Sacchet (2005, 11) notes that stronger women’s networks also help women secure greater political clout in Latin America, and in Brazil in particular.

“The parties most likely to respond to gender-related demands are those in which women tend to be well organised and able to carry out concerted and resourceful collective actions, and in which they have links to other supportive organizations and to civil society which afford them higher political clout. The support of political parties in Latin America for gender-related initiatives has been achieved through constant processes of coordination among women and the strategizing of their action.”

One illustrative example from Brazil underscoring these mechanisms is the nomination of Luiza Erundina, “a woman from the bases,” as the Workers’ Party (PT) candidate for mayor of São Paulo in 1988 (Macaulay 1996). The national and local party leadership favored the mayoral candidacy of PT congressman Plinio de Arruda de Sampaio, who had helped found the party in 1980. However, the former city councilor and grassroots activist Luiza Erundina managed to win an upset victory over Sampaio in the nomination vote among the women-heavy rank-and-file PT members (including several lower middle class housewives) in São Paulo, thanks in large part to her strong record of participating in local social movements in São Paulo, including the housing movement and the women’s movement. She organized mass rallies addressing women in the working-class neighborhoods of the periferia, and her party supporters relied on boca de urna (pollside leafleting) to convince the undecided as they went to cast their votes.⁶ Despite her lukewarm support among the party leadership, Erundina went on to win the mayoral election later that year, and she became the first woman to hold the mayorship of Brazil’s largest city. Her local grassroots base, which was heavily female, propelled her to victory against all odds. Although research investigating these mechanisms is sparse in Brazil, Luiza Erundina’s rise to power elucidates these mechanisms.

This leads us to our first two interrelated hypotheses. We expect that female politicians’ greater incentives to invest in party building and their comparative advantage in the recruitment of other women can compensate for the greater resource constraints that they face.

H1: Women incumbents recruit new party members at a higher or same rate as men.

H2: Women incumbents are more likely to recruit women party members relative to men incumbents, therefore, shrinking the gender gap in party membership.

Our theory also yields additional predictions for the quality of membership recruitment, with downstream implications for party resilience. Here, by party resilience, we mean specifically the fact that the party retains its member footprint over time and fewer of its members leave the party. By lowering the gender gap in party recruitment and recruiting more women relative to men, women increase party resilience. Party resilience increases because women party members are more likely to be loyal to parties and less likely to switch or leave parties than male members. Furthermore, women’s incumbents recruitment is less contingent on patronage (due to a relative lack of resources) but more likely to be focused on party ideology, policy, or gender rhetoric. Therefore, party members recruited by women incumbents—regardless of their gender—are also less likely to join for particularistic reasons and, therefore, are more likely to remain affiliated with the party for longer. For instance, while reliance on material incentives may help a party recruit a larger activist base in the short term, this may make the party’s activist base more fragile over the long run. In particular, members who were recruited into the party mainly through particularistic promises may be more likely to leave once the party is no longer in office and can no longer keep those promises. Investigating party switching in Brazil’s municipal government, Novaes (2018) finds that disloyal mayors who join for rents and patronage are more likely to switch parties, sowing the seeds for party fragility.⁷ By contrast, members who join their party for ideological or identity reasons or careerists who join parties to realize their political ambitions may be more loyal to the party over the long term. Party loyalists may also be more likely to extend effort toward deepening members party identification, which can set off a virtuous cycle that binds members more closely to the party network, sowing the seeds of party stability.

H3: Women incumbents are more likely to recruit party members who are less likely to leave or switch parties.

Our theory raises an ancillary hypothesis that we test empirically. We expect that because women benefit more from local politics and are more reliant on building a career in local politics to access other political positions, and because parties select and reward women for party building with party nominations, they are more likely to gain these benefits in constituencies where they engage more in inclusive party building. Note that in Brazil, we find that women progress in their political careers by either using local politics as a springboard to move to state or federal office or using their move to these other offices as a revolving door to return to local politics (see Supplementary Tables M.1 and O.1). This leads us to our ancillary hypothesis, which is as follows:

H4: Women incumbents are more likely to progress in their careers in constituencies where they engage in more inclusive party building.

Party Building and Women’s Representation in Brazil

During the reestablishment of democracy in the 1980s, Brazil developed a famously inchoate and fragmented party system characterized by many non-ideological, weakly organized, and predominantly clientelistic parties and frequent elite-level party switching between so-called partidos de aluguel, or “parties for rent” (Mainwaring 1999). Consequently, Brazil comes across as a relatively less likely case for any theory of party building, raising the question of why would candidates—men or women—engage in any party building at all. However, recent developments in Brazil’s party system and rich party membership sketch a more tenuous picture with vast subnational and intra-party variation. In fact, of the four party systems that Mainwaring (1999) classified as “inchoate,” only Brazil’s has strengthened over the last two decades, and of the 11 successful cases of party building in Latin America since 1978, 3 are Brazilian parties (PT, PDSB, and DEM) (Levitsky, Loxton, and Van Dyck 2016).

The level of party membership in Brazil is also quite high relative to European democracies. According to official membership statistics reported by Brazil’s electoral court, the Tribunal Superior Eleitoral, just under 16 million Brazilian voters were affiliated with a political party at the beginning of 2020, which corresponds to 10.6% of Brazil’s electorate; these figures correspond with active membership reported by citizens in the 2006 World Values Survey (10%) and in the Latinobarometer (11.8%).

Party membership growth in Brazil is driven primarily by demand-side factors, including politicians’ struggles for control over municipal party conventions and local party organizations. As we described earlier in the theory section, as is the case in a variety of contexts, citizens who join a Brazilian party are recruited by another member, politician, or local leader from that party. A significant portion of Brazilian party members also participate in party activities, including municipal conventions, party meetings, and campaigns (Ribeiro 2015).

Turning to gender gaps in party membership, Figure 1 shows that women have consistently been underrepresented in party politics in Brazil. Figure 1 shows the gender distribution of the members who were recruited into Brazilian parties between 2005 and 2020. The plot shows that most parties recruited women at significantly lower rates than men, and this gender gap was largely consistent across Brazil’s major parties. Even institutionalized, leftist parties that are ideologically supportive of women’s involvement in politics—such as the Workers’ Party (PT) and the Communist Party of Brazil (PCdoB)—recruited more men than women in over 75% of municipalities.

Figure 1. Party Membership Recruitment of Women in Brazil by Major Parties, 2005–20

Previous research suggests that this gender gap is due to a variety of factors, including the weak institutionalization of most Brazilian parties, the barriers that women face in advancing to party leadership positions and elected office, and the tendency of parties to hold meetings at night, when women are expected to be at home and with their families (Wylie 2018). Furthermore, in contrast with other countries in the region, quotas have not been an effective mechanism for increasing women’s representation in Brazil due to the design of the quota law, Brazil’s use of an open-list PR system in legislative elections, and the persistence of gender inequalities in campaign finance (Janusz, Barreiro, and Cintron 2022; Sacchet 2011). Therefore, women’s entry into local governments as a pathway to power offers a key means of improving women’s representation.

Our argument is based on the assumption that the women who join a party are likely to be more loyal to their party over time than the men who join a party. Women may be less likely to leave parties for reasons such as party loyalty or because they have fewer outside options—both of which are consistent with our theory. Table 1 provides empirical support for this assumption. While 14.6% of male rank-and-file party members switched to a different party at a later date, the rate of party-switching among women was considerably lower at only 8%. Likewise, female candidates for elected office were less likely to switch to a different party compared to male candidates at all levels of government. All of these differences in the rate of party-switching between men and women are statistically significant at the 0.01 level. These gendered differences in party-switching among party members suggest that correcting for gender-imbalance in party membership may increase the ability of the party to retain its grassroots membership and increase party resilience, sowing the seeds for stronger parties.

Table 1. Rates of Party Switching among Brazilian Party Members by Gender, 2005–20

Data and Empirical Strategy

Our main dependent variables are the rate of activist recruitment and the gender gap in recruitment. We measure both variables using individual-level data on party membership in Brazil provided by Brazil’s electoral court, the Tribunal Superior Electoral (TSE). This dataset contains a unique identifier, name, and the year of affiliation for all party members, and we code the members’ gender based on Portuguese naming conventions for given names. The data used in our analyses are aggregated to the municipality-term-party level, and it spans the four mayoral terms between 2005 and 2020. Appendix A of the Supplementary Material reports summary statistics for this dataset.

We operationalize the total recruitment rate as the number of new party members that the party recruited per one thousand voters in a given municipality over the course of the 4-year mayoral term. Following convention, we operationalize the gender gap in recruitment as the difference between the recruitment rate of men and the recruitment rate of women in a given municipality. We prefer this measure because it is a closer match for our theory (which is about relative recruitment), and it also requires fewer parameters to interpret.

$$ {\displaystyle \begin{array}{l}GenderGap=1,000\times \left(\frac{MaleRecruits}{MaleVoters}-\frac{FemaleRecruits}{FemaleVoters}\right).\end{array}} $$

Figure 2 shows the distribution of gender gap in party membership. The average gender gap is 0.79, which indicates that recruitment is biased toward men. Women members were recruited at a lower rate than men in 73.5% of municipalities.

Figure 2. Gender Gap in Party Membership across Brazil’s Municipal Constituencies, 2005–20

We estimate the effect of having a female mayor in office using a regression discontinuity (RD) design in close municipal elections. We restrict our sample to mayoral elections in which one of the two main candidates was female, while the other was male. We estimate the following model within a bandwidth of $ \pm h $ , where h is selected using the data-driven bandwidth selection method developed by Calonico, Cattaneo, and Titiunik (2014).

$$ {Y}_i={\displaystyle \begin{array}{l}{\beta}_0+{\beta}_1M{V}_i+{\beta}_2M{V}_iFemaleMayo{r}_i\\ {}+\hskip2px \delta FemaleMayo{r}_i+{\epsilon}_i.\end{array}} $$

Our running variable, $ M{V}_i $ , is the margin of victory of the female mayoral candidate, and positive values indicate that the female candidate won the mayorship. The main quantity of interest is $ \delta $ , which represents the change in the outcome variable at the 0% margin of victory treatment threshold. We discuss how our design departs from previous research that has used an RD design to study gender effects in Appendix C of the Supplementary Material. Furthermore, in Appendix J of the Supplementary Material, we also examine whether our results are robust to a difference-in-differences specification, and we find that this alternative model yields strikingly similar conclusions as the RD model about the differences between female and male mayors.

RESULTS

Women’s Party Membership Recruitment

The central prediction of our argument is that although women and male incumbents are likely to have similar effects on party membership recruitment in the aggregate, female incumbents are more likely than male incumbents to invest in the recruitment of women. Figure 3 illustrates the effects of a woman mayoral candidate’s electoral victory on membership recruitment into her party over the course of the next term. The points depict binned sample means, while the curves depict quartic global polynomials that approximate the population-conditional means for the treatment and control observations. Panel A shows the effect on her party’s overall recruitment rate in that municipality, and there is no evidence that having a woman mayor in office either boosted or depressed membership recruitment in the aggregate. However, Panel B suggests that a woman mayor can significantly affect the gender composition of her party’s recruits. The sharp drop in the gender gap at the 0% margin of victory threshold indicates that the party’s recruits were more evenly balanced between men and women in the municipalities where the female candidate won the mayoral election. This is consistent with the hypothesis that having a female politician in executive office can lead her party to cultivate a more gender-balanced activist base.

Figure 3. The Effect of Female Mayors on Membership Recruitment into the Female Candidate’s Party

Figure 4 compares the effects of women and men mayors on recruitment into their respective parties. Panel A shows that regardless of the gender of the party’s mayoral candidate, winning the mayoral election had no significant effect on the party’s total membership recruitment during the next term. This implies that although women mayors may have less access to patronage resources compared to male mayors (Brollo and Troiano 2016), this does not lead female mayors to recruit new members at a significantly lower rate compared to male mayors.

Figure 4. The Effect of Incumbency on Membership Recruitment, by Winner’s Gender

Note: Coefficients depict RD estimates of the effect of incumbency on membership recruitment, based on a local linear regression specification with triangular kernels. The comparison cases are municipalities where candidates of the same gender lost the mayoral election. Additional details on the model specification and results can be found in Table C.1 in the Supplementary Material.

By contrast, Panel B of Figure 4 shows that there is a significant difference in how the two types of mayors affect the gender composition of their parties’ recruits. Having a female mayor in office significantly decreased the recruitment bias toward men by around one member per one thousand voters relative to when her party was in opposition, while male mayors had no effect on the gender gap of their parties’ recruits relative to their party being in opposition. This suggests that the gender gap results shown in Figure 3 do not simply reflect an incumbency effect in general but rather the effect of a female mayor in particular. It also provides further evidence for this article’s main claim that female politicians are more likely than male politicians to build inclusive party organizations.⁸ In Appendix J of the Supplementary Material, we show that a difference-in-differences specification yields nearly identical conclusions about the difference in the effects of female and male mayors: the gender of the mayor had no effect on aggregate membership recruitment, while female mayors reduced the gender gap by around one member per one thousand voters relative to male mayors.

Quality of Women’s Recruitment

We also expect that women and men recruit members in qualitatively different ways, and their contrasting recruitment strategies can affect the quality of the members that they bring into their parties. One way to figure out how well the party is recruiting new members is to look at how long the new members stay with the party. If the member joined the party for particularistic reasons, then we might expect them to disaffiliate or switch to a different party during a future term when their original party is no longer in office. On the other hand, people who join a party because of its ideas or programs may be more likely to stay in it for a long time.

Figure 5 shows the effect of female and male mayors on the probability that the members who were recruited into their party during their mayoral term chose to leave or switch to a different party during any subsequent term. Members who are recruited under women mayors are significantly less likely to switch or leave the party in a future term, and this effect is particularly strong for male members. This suggests that women mayors are able to select or instill loyalty among male members, while male mayors have no such effect on member loyalty. Because women members are less likely to switch or leave their parties regardless of who recruits them but are more likely to be recruited by women, and men members recruited by women are also less likely to switch or leave parties, this suggests that on net, having a woman mayor would increase the resilience of her party’s membership.

Figure 5. The Effect of Incumbency on Membership Attrition, by Gender of Mayor and Members

Note: Coefficients depict RD estimates of the effect of incumbency on membership attrition for the mayor’s party, grouped by the gender of the party members who were recruited during the mayor’s term in office. Additional details on the model specification and results can be found in Table C.2 in the Supplementary Material.

Supplementary Figure L.1 breaks down these results by the party of origin. Given that we are slicing close races on multiple variables, our estimates are noisier and should be interpreted with caution, but they suggest that the increased loyalty among male members recruited by women is driven by lower attrition in the PSDB and MDB but not the PT, that is, parties that are relatively less institutionalized than the PT. Most estimates are, however, statistically insignificant and underpowered due to the sparsity of data on close races split by party and gender. Yet these findings provide preliminary support for the claim that increasing women’s representation can help make weakly institutionalized parties more resilient.

Our results are in the expected direction. However, it is important to emphasize that we only observe party registration and not member inactivity. Some party members who have become inactive may choose not to go through the bureaucratic process of de-registering from the party, or they may forget that they were affiliated in the first place. Consequently, an alternative interpretation of our finding that party members recruited by women are less likely to leave parties is that women may simply be recruiting more inactive members who do not de-register themselves. However, we do not have any reason to believe that the costs—administrative or psychological—of member de-registration vary with the mayor’s gender. Existing research on party members activities suggests that party members are very active in party life (Ribeiro 2015).

Additionally, we investigated whether there are electoral cycles to party building in Appendix Q of the Supplementary Material. Supplementary Figure Q.2 shows that women’s recruitment does vary with the election cycle and has positive implications for their career progression, suggesting that women are not simply recruiting inactive members who are unlikely to bring them this leverage. Furthermore, the role model effect theory suggests that the effects of seeing women win elections are strongest closest to the election year and fade over time (Wolbrecht and Campbell 2007). However, we do not find that women mayors lower the gender gap in the first year of the election; instead, the gender gap lowers in the third year of their term, where most municipal conventions take place. Together, this evidence gives further support to the critical role of local leaders in recruiting active members into the party.

Party Building and Political Career Pathways

According to our theory, holding local office would be more consequential for the political survival and careers of women politicians compared to men politicians because the former are unlikely to advance far within a men-dominated party organization without the campaign funds, experience, visibility, and network-building capacity that local office provides. By contrast, male politicians may have other pathways to continue in politics or improve their future career prospects.

Figure 6 shows the effect of winning the mayoral election on the candidate’s chances of renomination in the next mayoral election (conditional upon being eligible for a second term) and the candidate’s chances of being nominated for state or federal office in some future term. Mayoral election outcomes had no significant effect on renomination for either gender. However, female candidates who won their mayoral election were 8 percentage points more likely to be nominated for state or federal office in a future term, while winning the election had a slightly negative and insignificant effect on the career advancement of male candidates. Appendix M of the Supplementary Material shows that local office indeed provides an important pipeline to congressional office (with close to 40% of congressional candidates having background in local offices). Appendix P of the Supplementary Material shows that women are also more likely than men to use congressional office as a revolving door that enables them to continue in local politics after their term as mayor has expired.

Figure 6. The Effect of the Election Outcome on the Candidate’s Career Trajectory

Note: Coefficients depict RD estimates of the effect of winning the mayoral election on the probability that female and male mayoral candidates were renominated as mayor or were nominated for Federal or State office in a subsequent election. Additional details on the model specification and results can be found in Table C.3 in the Supplementary Material.

This asymmetry between female and male candidates is consistent with the prediction that mayoral office is important in and of itself and serves as a vital launching pad for female politicians’ political careers but is relatively less important for male politicians, who may have other means of attaining the benefits of local office. Our findings speak to the example of the rise of Luiza Erundina, who embarked on the political stage based on the support of working-class women and her collaboration with the women’s and health movements and continued to thrive in politics thereafter.

Appendix O.1 of the Supplementary Material shows that by advancing to state or federal office, women can overcome the two-term limit rule and are more likely to return to being mayors than men. These results suggest that women mayors are successfully using local politics to have thriving political careers. We also expected that women mayors would be more likely to leverage mayoral office for future career advancement if they governed a municipality where women’s party membership was historically low and, therefore, where there was thus greater room for improving the gender balance of party membership.

Figure 7 shows the effect of winning the mayoral election on nomination for higher office, grouped by whether the share of women who were affiliated with any party was above or below the median value prior to the start of the term. Winning mayoral office had a substantive and significant effect on women’s chances of advancing to higher office only in municipalities where women’s party membership was previously low. As expected, in districts where women’s baseline membership recruitment was low, the gender gap in recruitment also lowered significantly. In other words, women progressed upward in politics in constituencies where the scope to lower the gender gap in party member recruitment was higher. Women also lowered the gender gap in recruitment the most in these same constituencies. In the light of findings from Supplementary Table O.1, we interpret these findings as women being able to use their advancement to congressional office as a revolving door to return to mayoral office in constituencies where they engaged in building a more inclusive party.

Figure 7. The Effect of the Election Outcome on Nomination for Higher Office, by Women’s Party Membership in Previous Terms

Note: Coefficients in Panel A depict RD estimates of the effect of winning the mayoral election on the probability of renomination for mayor and nomination for higher office, grouped by cumulative past membership recruitment of women into political parties in the candidate’s municipality; the sample is partitioned at the median value for this variable. Panel B depicts RD estimates of the effect of winning the election on the gender gap in recruitment during the current mayoral term. Additional details on the model specification and results can be found in Tables C.4 and C.5 in the Supplementary Material.

CONCLUSION

In our article, we show how inclusive party building is a new way in which descriptive representation enhances democracy. To establish this link, we offer a gendered theory of incumbency and party building. We theorize that parties promote incumbents based on gendered criteria, incentivizing women to recruit new party members for career advancement. Using rich administrative data in the substantive setting of Brazil’s municipal government, we show that despite having lower access to resources required for recruitment, women incumbents recruit party members at the same rate as men and lower the gender gap in party membership. Further, in line with our expectations, we find that women party members are less likely to switch parties in general, and that women mayors recruit party members who are also less likely to switch or leave the party. In turn, we find that women politicians can improve their own chances for career progression when they are able to narrow the gender gap in their party’s membership base.

Brazil has been our theoretical focus. However, we expect our findings to generalize to other settings where patronage and clientelism remain important for forging electoral and political success but remain less accessible to women. The gender inequalities that underpin our theoretical framework and link descriptive representation to party building, such as women’s greater party loyalty, the dependence of women on local politics, and the gender gap in access to resources and campaign finance, are present, naturally with variation, in most low- and middle-income democracies. Similarly, in most low- and middle-income countries, party elites rely on local politicians to build parties, as evident in the burgeoning research on party activist recruitment. While our results are internally valid for close races, our theory is less contingent on the levels of electoral competition, and we, therefore, expect our results to generalize to non-close races. Research from India is reassuring, where Goyal (2023) finds that women are more likely than men to recruit women as party activists in non-close but instead same-gender (reserved) races. We expect variation in electoral systems, degree of decentralization, party system institutionalization, and the prevalence of grassroots politics to shape our underlying mechanisms—the incentives of women to toe the party line and use local politics to progress in politics—and, therefore, to be important moderating variables. Future comparative research can vastly improve our understanding of how these conditions moderate the link between descriptive representation and party building.

Our work raises several additional extensions. A key question is whether changing the composition of the party’s grassroots has implications for how women legislate in parliaments or whether they demand fairer treatment from party leaders and selection committees. Existing research has mainly focused on how critical mass in legislatures affects women’s behavior within legislatures; that is, the relationship has been analyzed at the same political level (Barnes 2016). But we know less about how changes in descriptive representation at the party’s grassroots affect women’s ability to be effective politicians. In other words, conditional on having a grassroots support base, are women politicians more likely to rebel from their parties? Are women more likely to raise progressive gender issues that conflict with party interests? This question is particularly important, as women face more party discipline and tend to sacrifice gender interests over party interests (Htun and Power 2006). Another interesting area for research is the process of gendered recruitment across different parties and the nature of partisan identification among women party members. Our quantitative approach and sparse data on close races split by parties limit our ability to delve into important party-specific dynamics and to answer questions about the qualitative process through which women are recruited to become party members and the role they play in party life. By exploring these questions in a comparative research agenda, we can gain a better understanding of the gendered nature of representation and political party development. Finally, our findings show that in a setting where gender quotas have been ineffective, women’s party building can be a critical pathway to improving descriptive representation.