Closing the Gap: How Mayors’ Individual Attributes Affect Gender Wage Disparities in Local Bureaucracies

The Role of Gender Representation in Decreasing Gender Inequities

The idea that representation facilitates the inclusion of historically underrepresented groups in public sector governance is the motivation behind efforts to establish diverse and representative organizations (Riccucci & Van Ryzin, 2017). While government organizations have often excluded such groups (Bishu & Alkadry, 2017), minority representation in various settings has resulted in outcomes aligned with minority interests (Bishu & Kennedy, 2020; Favero & Molina, 2018). Indeed, public organizations with women in positions of power create more equitable environments that help reduce gender-based discrimination (Alkadry et al., 2019).

According to existing theory, government personnel become passive representatives when they are grouped into the same social category as the people they serve (Bishu & Kennedy, 2020). Thus, women responsible for governance are passively representative of women affected by governance. In addition to mirroring shared characteristics, an intrinsic identification with group members can lead public servants to act in accordance with the group’s interests, leading to active or substantive representation. A sense of common identity is central to the process linking gender representation to gender-based equity.

Within numerous organizational settings, personnel are able to recognize the members of their respective group; in doing so, they distinguish others with whom they share a common, and often salient, trait such as gender (Ashforth & Mael, 1989). As Ashforth and Mael (1989) note in their discussion of social identity theory, identification as a member of a specific group acts as a heuristic individuals use to interpret various dimensions of their organization role. This includes an attachment with other women in the organization and these gendered attachments can influence various decisions and processes (Ashforth & Mael, 1989; Ely, 1994). However, gender is more likely to influence decision-making when women are given significant discretion in an institutional context where a gender-salient issue is at hand (Keiser et al., 2002).

The influence of gender-based identification reverberates through the ranks of a bureaucratic system. At the top, the link between women’s passive and active representation is strongest when women in leadership positions identify with other women in society and within her organization. Previous studies suggest this sense of identity is often present; for example, public opinion surveys find that women share more in common with other women than they do with men. Research on mass public opinion points to a significant gender gap in political preferences and voting behavior across numerous settings (Givens, 2004; Norrander & Wilcox, 2008). In the context of elected leadership, research finds women elected officials frequently express policy preferences that diverge significantly from the preferences of men (Barnes, 2016; Poggione, 2004). Further research suggests that women elected officials act on these preferences by prioritizing different government programs and allocating government expenditures differently than men. For example, women elected officials are more likely to prioritize spending on issues that disproportionately concern women, such as social policy issues (Funk & Philips, 2019; Holman, 2014).

These findings are important because they reveal that women in elected leadership often feel a strong sense of identification with other women, and this has significant implications for the impact these women may have on other women throughout their organization. If a woman elected official’s policy preferences are closely aligned with the preferences of other women, then she may be inclined to appoint or promote women to fill positions throughout the bureaucracy, especially positions with influence over important policy issues or resources. In this way, she can mitigate the well-known principal-agent dilemma between elected officials and bureaucrats and ensure that her policy agenda is implemented. Indeed, research finds women’s representation in elected offices is associated with women’s representation in appointed offices throughout various tiers of the bureaucracy (Meier & Funk, 2017; Whitford et al., 2007). Appointing more women to bureaucratic leadership roles, which are often accompanied with higher pay, thus helps to shatter the glass ceiling and reduce gender disparities that may exist throughout the bureaucracy.

Living as a woman in a patriarchal society also means that women elected officials have likely experienced gender inequities themselves. Thus, women elected officials should be more aware of the need to improve women’s representation and equality throughout the public sector, especially at higher tiers of the bureaucracy where gender disparities tend to be magnified. Prior research provides evidence that women elected officials are more likely to recognize women as an underrepresented group, view themselves as representatives of women, and act in the interests of women compared to their men counterparts (Bratton & Ray, 2002; Schwindt-Bayer, 2010). Further research suggests that improving women’s representation in leadership and managerial roles leads to smaller gender wage gaps (Cohen & Huffman, 2007; Funk et al., 2019; Reese, 2019) and increased emphasis on racial and gender diversity (Feeney & Camarena, 2021) in the bureaucracy.

If women elected officials are more likely to share policy preferences with other women and more likely to recognize the existence of gender disparities and have a vested interest in reducing these disparities (compared to men elected officials), then increasing women’s representation in elected offices should translate into smaller gender disparities in the bureaucracies over which these elected officials have influence.

Diversity within Gender Groups and Attitudes toward Women

Though women share many commonalities, we must also account for the reality that women compose a very large and diverse social group and interrogate what this means for gender disparities in the public sector. In addition to their gender identity, women possess various cross-cutting identities, attributes, and preferences. This means that women’s identities and preferences may differ from one another in a manner that influences organizational outcomes.

Research on intersectionality demonstrates how coinciding systems of disadvantage and inequality affect group members differently, causing certain members, particularly those marginalized on multiple fronts (e.g., gender and race or social class), to experience unique forms of discrimination, while others are advantaged (Crenshaw, 1991). Examples of how the differences among women have manifested throughout history are abundant. In modern times, the divisions among women’s attitudes towards issues such as childcare subsidies, abortion, and family roles provide useful illustrations of this point (Bolzendahl & Myers, 2004; Cassese & Barnes, 2019a). And while issues of gender equality are relevant to all women (Baldez, 2011), some women may not view gender disparities and sexism as fundamentally problematic (Cassese & Barnes, 2019b).

This is significant because it implies that certain types of women elected officials may view gender disparities as largely inconsequential or merely a function of market forces, while others may view it as an injustice in need of rectification. The types of women elected officials most likely to view gender disparities in the public sector as problematic, and express an interest in reducing such disparities, are those who hold progressive gender attitudes; that is, less sexist and more egalitarian views toward gender (Cassese & Barnes, 2019b). Compared to women who hold more traditional views of gender, women with progressive attitudes should be more likely to view gender equality, and hence reducing gender disparities, as an important endeavor. Thus, women elected officials with certain attributes, particularly attributes that are typically correlated with progressive views toward gender, should reduce gender disparities in the bureaucracy to a larger degree compared to women elected officials who hold less progressive, more traditional attitudes toward gender and gender equality.

We focus on three attributes that tend to be correlated with progressive attitudes toward gender: political ideology, age, and level of education. Theoretical explanations suggest that individuals are more likely to develop progressive attitudes toward gender if gender egalitarianism benefits their self-interests or if they are exposed to such ideals through the workplace, education, or political socialization (Bolzendahl & Myers, 2004). Previous research has found that women, and individuals who are younger, better educated, and those who express leftist ideologies are among those most likely to hold progressive gender attitudes across a variety of cultural contexts (Bolzendahl & Myers, 2004; Cassese & Barnes, 2019b; Inglehart & Norris, 2003), including Latin America (Morgan & Buice, 2013).

In Brazil, which serves as the context of this study, public opinion data suggests that women, along with left-leaning, younger, and better educated individuals, have more progressive views of women in society. Data from the 2014 AmericasBarometer survey presented in Table 1 suggests that a larger proportion of women than men disagree with sexist statements, like women are asking for too much in their demands for equality, gender discrimination is no longer a problem in Brazil, women are worse drivers than men, men should have priority when jobs are scarce, and women use more emotion than reason when discussing politics. Likewise, respondents who identified as leftists are more likely to disagree with these statements compared to those on the ideological right, as are respondents who are younger than 50 years old compared to those 50 or older, and respondents with some college education compared to those with a high school education or less.

OPEN IN VIEWER

Table 1.

Percent of Brazilians Who Disagree with Sexist Statements in the 2014 AmericasBarometer Survey.

	Gender		Ideology		Age		College edu.
Statement	Women	Men	Left	Right	Under 50	Over 50	Yes	No
1. Women are asking for too much in demands for equality	38.17+	32.89	40.86***	32.69	36.49	33.98	49.69***	34.14
2. Gender discrimination is no longer a problem in Brazil	51.62**	46.31	52.38**	45.04	49.26	48.18	55.21**	48.26
3. Women are worse drivers than men	63.94***	47.25	58.46**	51.56	57.65**	50.24	69.33***	54.00
4. Men should have priority when jobs are scarce	57.45***	45.10	54.37**	48.52	54.04***	44.07	68.71***	49.21
5. Women use emotion over reason when discussing politics	37.09**	30.73	34.67**	30.43	35.41*	30.00	38.65**	33.38

Source. The 2014 AmericasBarometer by the Latin American Public Opinion Project (LAPOP), www.lapopsurveys.org.

Note. +p < .1. *p < .05. **p < .01. ***p < .001 in a one-tailed t test comparing each group on the 7-point disagree-agree scale. Respondents reporting 1, 2, 3 are coded as disagreeing with the statement. Respondents are coded as left-leaning if they report 5 or less on the 10-point left-right ideology scale, and as right-leaning if they report 6 or higher. Respondents who have more than 11 years of education are coded as having some college education.

For all of these reasons, we expect elected officials who are younger, better educated, and those from leftist parties to hold more progressive gender attitudes, and therefore be more likely to view gender disparities as problematic and act to reduce those disparities, compared to their older, less educated, and non-leftist counterparts.

Recent studies have also examined men’s role in representing women interests, finding that men make decisions that produce benefits for women as a whole under certain conditions, particularly when it aligns with their self-interests or worldview (Nugent, 2019; Valdini, 2019). We therefore offer a third hypothesis concerning this potential relationship. While men share a gender identity, important differences also exist among this gender group. These differences can cause certain types of men to be more aware of, and concerned about, gender inequalities that exist in the public sector. Like women, men elected officials who are younger, better educated, and from leftist parties may be more likely to hold progressive gender attitudes and view gender inequality as problematic and, as a result, reduce gender disparities in government organizations to a larger degree than men who are older, less educated, and those from non-leftist parties.

Comparing across genders, men with these attributes may reduce gender disparities in the bureaucracy to an even greater degree than certain types of women, particularly women who lack these attributes. Moreover, if these attributes are actually more important than gender in determining one’s concern for issues of gender equality, then men with these attributes should have as great an impact on reducing gender disparities as do women elected officials with these attributes, and an even larger impact compared to women who lack these attributes. However, if gender group identity is really the key factor, then men elected officials with varying attributes would be expected to reduce gender disparities in the bureaucracy to a lesser extent compared to women elected officials across the board.

Empirical Context: Brazilian Municipal Executive Bureaucracies

We test these hypotheses within the context of Brazilian municipalities. Specifically, we examine how mayors’ individual attributes affect gender wage disparities in the municipal executive bureaucracy, an organization that includes all those employed by the municipal executive branch.

Women are significantly underrepresented at all levels of government in Brazil. Despite electing a woman president in 2010 and 2014, and adopting a gender quota law in 1995, Brazil is among the countries with the lowest levels of women’s representation in the world. ² At the national level in 2020, women legislators made up only 15% of the Chamber of Deputies and 15% of the Senate. At the municipal level, only 12% of mayors and 14% of city councilors elected for the 2017 to 2020 term were women—and these are the largest percentages of women ever to be elected at the local level. Furthermore, gender discrimination and inequality continue to be widespread throughout Brazil. Women are not only underrepresented in political institutions, but they are also more likely to live in poverty, earn smaller wages, and experience greater violence compared to men (Barros et al., 1997; Waiselfisz, 2015).

Municipal governments play an important role in Brazilian democracy and are venues through which disparities in women’s representation and equality can be redressed. The 1988 Constitution recognizes municipalities as autonomous political units with the authority to create laws, collect taxes, and elect local leaders. Municipalities control a considerable amount of government resources and are responsible for delivering important public services, including education, healthcare, and sanitation.

At the municipal level, political power is concentrated in the mayor’s office. Mayors are elected by popular vote every 4 years. Once elected, mayors have the ability to appoint civil servants to positions in the municipal executive bureaucracy. Unlike the federal level of government, wherein civil servants must meet high qualification standards and pass a rigorous entrance exam (concursos públicos) prior to being hired into the executive bureaucracy (Pacheco, 2010a, p. 194), municipal executive bureaucracies tend to be less professionalized. Patronage politics are commonplace in Brazilian municipal governments (Pacheco, 2010b, p. 278) and a position in the civil service, with its competitive salary, generous benefits, and prospect of tenure, is highly desirable. While mayors are limited in their ability to remove tenured civil servants, they can alter the composition of the municipal executive bureaucracy through the creation of temporary positions, by increasing the size of the permanent staff, and by promoting individuals to leadership roles, such as head of a municipal agency. In this way, mayors who are concerned about women’s underrepresentation and lack of equality can work to ameliorate gender disparities that exist within the municipal executive bureaucracy.

Measuring Gender Disparities in Local Executive Bureaucracies

Brazil has 5,570 municipalities, each with its own government made up of a mayor, a city council, and a municipal executive bureaucracy. While women are severely underrepresented in elected offices, they make up the majority of bureaucrats in the municipal executive bureaucracy. Across the three electoral terms included in our data (2001–04, 2005–08, 2009–12), the average municipal executive bureaucracy consisted of 62% women. However, a closer look at how women are distributed throughout the bureaucracy reveals that women are underrepresented in higher-paid positions—those tied to important posts like municipal agency head.

Adjusting for inflation, women bureaucrats’ average monthly salary across all municipalities in 2001 to 2012 was R$811.89 (Brazilian Reals) and men’s average salary was R$950.31, resulting in an average wage gap around R$138 per month. This difference may seem trivial, but it is actually quite large. Over the course of a year, this salary disparity results in men earning approximately R$1,656 (excluding any bonuses) more than women on average. This number equates to the average man earning roughly two extra months of salary compared to the average woman bureaucrat over the course of a year. Moreover, even after accounting for inflation, the difference in men and women’s average salaries has increased slightly over time.

Figure 1 plots average real wages for men and women bureaucrats over time along with the gender wage gap, calculated by subtracting women’s average wages from men’s average wages. The figure shows that while both men and women’s wages have fluctuated over time, men consistently earn more than women and this gap has generally widened over time. In 2001, the difference between men and women’s average monthly salaries was R$65.12, but the difference rose to R$159.66 in 2012, and exceeded R$200 in 2006 and 2009. These large gaps may indicate that women face a glass ceiling in the executive bureaucracy; however, the deidentified data used in this study does not include individual bureaucrat’s ranks or job descriptions, so we are unable to determine whether the wage gap is a problem of the glass ceiling (women holding lower ranks) or gender discrimination (women being paid less than men at the same rank).

OPEN IN VIEWER

Figure 1.

Average real wages and the gender wage gap over time.

Measuring Mayors’ Individual Attributes

The mayor’s gender is measured using self-reported data on whether the mayor identifies as feminino or masculino. While this measure is limited in that it dichotomizes gender into two distinct groups and does not capture the more fluid and multifaceted aspects of gender, it is the only measure currently available.

Political ideology is measured using the mayor’s political party affiliation, which was coded as left-leaning or not (Power & Zucco, 2012; see Table S1 in the Supplemental Appendix for a list of parties coded as left). Age is measured as the mayor’s age in years. Education is measured on a 6-point scale, where 1 equals incomplete primary education and 6 equals college degree or higher. In the Supplemental Appendix, we use alternative measures as robustness checks, including a latent measure created by combining political ideology, the inverse of age, and level of education via factor analysis; the left party variable on its own; and a dummy variable for whether the mayor is from the Worker’s Party (Partido dos Trabalhadores, PT), the left-leaning party considered to be the most ideologically disciplined historically.

Empirical Models and Variables

We use generalized least squares with random effects to analyze how mayoral attributes affect gender wage disparities in local executive bureaucracies. The unit of analysis is the municipality-year, that is, individual municipalities in 2001 to 2012. The dependent variable is the gender wage gap in the local executive bureaucracy (the solid line plotted in Figure 1). The wage gap was calculated by subtracting women’s average wages from men’s average wages, adjusting for inflation. Thus, a large positive value would indicate that men bureaucrats earn significantly higher monthly wages than women bureaucrats, on average, while a negative value would indicate that women bureaucrats earn higher wages than men bureaucrats. In the first model, the main explanatory variables are the mayor’s gender, political ideology (left party or not), level of education, and age. In the second model, we include interaction terms between each of these four variables to examine how the combination of different mayoral attributes affects the gender wage gap.

The models also control for whether the mayor served in the previous term (incumbent) and whether it is the mayor’s first, second, third, or fourth year in office (mayor’s year in term). All mayors are elected concurrently to a 4-year term. Mayors are limited to one immediate reelection, but can run for office again after sitting out one term. We are unable to control for the mayor’s race and ethnicity because these data do not exist for 2001 to 2012. The Brazilian government began collecting data on the racial identification of local elected officials in 2016.

Finally, the models also control for characteristics of the municipal government and the municipality that likely shape women’s equality in the public sector, including the percent women on the city council, size of municipal population, and gross domestic product (GDP) per capita. We take the log of population and GDP because these variables are highly skewed. We also control for political patronage using two variables: first, the percent of workers in the municipality’s formal labor force that are employed by the municipal executive bureaucracy (patronage) and, second, the number of individuals employed by the municipal executive bureaucracy per 1,000 residents (bureaucracy size). Both measures capture the importance of the municipal executive branch in providing employment for municipal residents.

We control for gender gaps between men and women bureaucrats in terms of their average levels of education, age, hours worked per week, and total months of employment in the municipal executive bureaucracy since these gaps might explain gender gaps in average wages. Each of these variables were constructed by subtracting women’s average value from men’s average value, parallel to the calculation of the wage gap. The analyses also include a one-year lagged dependent variable and a linear time trend to account for unobserved effects over time. Summary statistics for all variables are presented in Table S1 in the Supplemental Appendix.

Results: Women Mayors Produce Smaller Gender Wage Disparities

Table 2 presents results from the first model. The results suggest the mayor’s gender is significant in explaining gender disparities in executive bureaucracies. Specifically, the model estimates that municipalities with a woman mayor will have monthly gender wage gaps that are an average of R$6.94 smaller compared to municipalities with a man mayor, even after accounting for gender differences in men and women bureaucrats’ average education, age, hours worked per week, months of employment, and additional municipal-level variables. This supports the first hypothesis that women mayors will reduce gender disparities to a larger degree than men mayors. ³ These results also suggest that the mayor’s partisanship, level of education, and age do not have a statistically significant effect on the wage gap.

OPEN IN VIEWER

Table 2.

Explaining Gender Wage Disparities in Local Bureaucracies.

	Base model
Woman mayor	−6.937* (2.777)
Left party	2.318 (1.820)
Mayor’s education	−0.423 (0.480)
Mayor’s age	0.075 (0.081)
Incumbent	−6.501*** (1.545)
Mayor’s yr. in term	−19.342*** (0.704)
Pct. women councilors	−0.165* (0.068)
Log population	−0.849 (0.885)
Log GDP per capita	50.281*** (1.691)
Patronage	−12.762** (4.537)
Bureaucracy size	−0.299*** (0.048)
Education gap	53.171*** (1.491)
Age gap	13.439*** (0.419)
Hrs. worked gap	4.537*** (0.261)
Months employed gap	0.268*** (0.038)
Lagged wage gap	0.554*** (0.004)
Time trend	1.500*** (0.292)
Constant	46.727*** (11.369)
Overall R2	0.463
Observations	58,380

Note. Model estimated via random-effects generalized least squares (GLS). Standard errors are in parentheses. The dependent variable is the difference between men and women bureaucrats’ average monthly real wages in constant 2012 Brazilian Reals (R$), calculated by subtracting women’s average wages from men’s average wages.

*p < .05. **p < .01. ***p < .001.

These results are robust to several different estimation strategies (see Table S2 in the Supplemental Appendix). The variable for woman mayor continues to be negative and statistically significant across all model specifications, except for the municipal fixed effects model. This is likely due to the lack of variation in the woman mayor variable within a single municipality over time. Between 2001 and 2012, 82.25% of municipalities never had a woman mayor. Thus, the majority of the variation in the woman mayor variable is between municipalities rather than within municipalities. The mayor’s level of education and left partisanship also remain statistically insignificant across model specifications. However, the mayor’s age reaches statistical significance in both the fixed effects and hierarchical models, suggesting that younger mayors might produce smaller gender wage gaps, but this finding is not robust across all model specifications.

To test how the combination of the mayor’s gender with their other attributes affects gender wage disparities, we estimate a second regression model that includes an interaction between the mayor’s gender, partisanship, age, and level of education. The interaction allows us to examine the effects of women and men mayors with both similar and different attributes. In particular, it allows us to test hypotheses 2 and 3 and whether the mayor’s gender has a statistically significant effect on gender wage gaps only when certain attributes are present. We dichotomize both age (under 50 = 0 and over 50 = 1) and level of education (high school or less = 0 and at least some college = 1) for ease of analysis. Because the results of the four-way interaction model are difficult to interpret based on regression coefficients alone (for more, see Kam & Franzese, 2007), predicted values of the dependent variable given different combinations of mayoral attributes are presented in Table 3 and Figure 2. Regression coefficients for the full interactive model are presented in the Supplemental Appendix (see Supplemental Table S3).

OPEN IN VIEWER

Table 3.

Average Predicted Wage Gap under Different Types of Mayors.

Mayor’s gender	Mayor’s attributes	Predicted wage gap	Lower estimate	Upper estimate	Obs.
Woman	Left party, no college, over 50	84.12	47.16	121.09	90
Woman	Non-left party, college, under 50	90.29	81.53	99.04	1,604
Woman	Left party, no college, under 50	102.48	74.03	130.92	152
Woman	Non-left party, no college, under 50	105.75	92.93	118.56	749
Woman	Non-left party, no college, over 50	115.11	99.66	130.56	515
Woman	Left party, college, under 50	117.22	101.96	132.48	528
Man	Non-left party, no college, under 50	137.59	134.31	140.88	11,381
Man	Non-left party, no college, over 50	138.87	135.28	142.45	9,555
Man	Left party, no college, under 50	140.77	134.69	146.86	3,318
Woman	Non-left party, college, over 50	141.63	129.17	154.08	793
Man	Non-left party, college, under 50	144.83	141.74	147.91	12,919
Man	Non-left party, college, over 50	158.03	154.10	161.95	7,973
Man	Left party, no college, over 50	160.41	152.30	168.51	1,873
Man	Left party, college, over 50	170.20	162.88	177.52	2,297
Woman	Left party, college, over 50	177.37	153.12	201.63	209
Man	Left party, college, under 50	185.79	180.63	190.96	4,608

Note. The lower and upper estimates correspond to 95% confidence intervals. Obs. is the number of mayors in each group between 2001 and 2012.

OPEN IN VIEWER

Figure 2.

Average predicted wage gap with 95% confidence intervals.

Table 3 is organized according to the type of mayor predicted to produce the smallest gender wage gap to the largest gender wage gap. By these estimates, the type of mayor predicted to produce the smallest wage gap (R$84.12) is a woman mayor from a left party, who is over 50 years old, and lacks a college education. The mayor predicted to have the largest gap (R$185.79) is a man from a left party, who is under 50, and also college educated. Perhaps most striking about the predicted values presented in Table 3 is that women mayors tend to be clustered toward the top of the table, indicating that women mayors—regardless of their other attributes—are predicted to have smaller gender disparities in their executive bureaucracies compared to men mayors. Only women mayors who are over 50 and college educated (from both left and non-left parties) are predicted to have larger wage gaps than particular types of men mayors. Predicted values based on the mayor’s gender alone indeed confirm that the wage gap is smaller under women mayors in the interactive model (see Figure S1 in the Supplemental Appendix). Moreover, the marginal effect of the woman mayor variable produced by the interactive model (−4.87, p-value = .093) is very similar to the marginal effect of mayor gender in the non-interactive model.

Comparing women mayors to one another, no clear-cut patterns emerge with respect mayoral attributes beyond gender. Women mayors from left parties do not systematically reduce gender disparities to a greater degree than women from non-left parties. There are slightly more consistent patterns with respect to age and education: younger women mayors tend to produce smaller wage gaps compared to older women mayors, as expected, and women mayors who lack a college degree tend to produce smaller gaps compared to women with college degrees, contrary to expected. However, there are several exceptions to these patterns.

Comparing men mayors to one another yields very similar results. Men mayors from left parties have gender wage gaps that are just as large, or even larger, than men from non-left parties. Indeed, the four types of mayors (three men and one woman) predicted to have the largest gender disparities in their executive bureaucracies are all from left-leaning parties. Like women mayors, men mayors who are younger and lack a college education also tend to produce smaller predicted wage gaps compared to older men and college-educated men, respectively, but again there are exceptions to these patterns.

Figure 2 presents these predicted values graphically, with men and women mayors that share similar attributes plotted next to one another. Across each combination of attributes, pairwise comparisons tests confirm there is a statistically significant difference in the predicted wage gaps produced by men and women mayors in all but one case: mayors from left parties, who are over 50, and college educated. ⁴ Women mayors with these attributes are predicted to produce wage gaps that are similar in size to men mayors with these attributes. For all other cases, women mayors are predicted to produce smaller wage disparities compared to men mayors with similar levels of education, age, and ideology. At the low end, women mayors are predicted to decrease the wage gap by R$23.76 more than men mayors (comparing non-left, no college, over 50). At the high end, women mayors are predicted to decrease the gap by R$76.26 more than men mayors (comparing left, no college, over 50). Given that the average wage gap is R$138 per month, this means that certain types of women mayors may reduce the gender wage gap by more than half each month.

The predicted values presented in Table 3 and Figure 2 again offer support for the first hypothesis that women mayors will reduce gender disparities to a greater degree than men mayors. Comparing mayors with both similar and different attributes reveals that women mayors are predicted to almost always have smaller gender wage gaps in their executive bureaucracies compared to men. With respect to the second and third hypotheses, we do not find any clear-cut evidence that mayors who are left-leaning, younger, and college educated consistently produce smaller wage gaps compared to their non-leftist, older, and non-college educated counterparts. These results also indicate that men mayors who are left-leaning, younger, and college educated do not systematically reduce gender wage disparities to a greater degree compared to women mayors who are right-leaning, older, and without a college education.

In terms of the control variables, results from both the non-interactive and interactive models suggest that incumbent mayors and those in the later years of their term have smaller wage gaps in their executive bureaucracies. Similarly, mayors with larger bureaucracies (measured using the patronage and bureaucracy size variables) also produce smaller wage gaps. However, municipalities with greater GDP per capita have larger wage gaps. The wage gap is also larger in municipalities where large gender gaps exist between men and women bureaucrats’ average level of education, age, weekly hours, and months of employment. The positive coefficients on the lagged wage gap and the time trend variable indicate that the gender wage gap has increased over time, echoing Figure 1.

Finally, municipalities with more women on the city council have smaller wage gaps in the executive bureaucracy than those with few women councilors, suggesting that advancements in women’s representation on local legislatures can spillover into the executive bureaucracy. The models estimate that increasing the percent women on the city council by one percentage point is predicted to decrease the gender wage gap by an average of R$0.17. Given that the typical city council is made up of nine members, this means that electing one additional woman to the city council, and thereby increasing women’s representation by 11.11%, is expected to decrease the wage gap in the municipal executive bureaucracy by an average of R$1.89 each month.

Discussion and Conclusions

In democracies across the globe, shifting political and demographic landscapes underscore the need for inclusive governance. Perhaps now more than ever, it is essential that scholars examine the ways in which government personnel shape the experiences of society’s marginalized groups, including within government bodies. We devote our attention to this pursuit by studying the relationship between representation and the gender wage gap in Brazil’s public sector. In doing so, we answer the call of recent scholars (e.g., Bishu & Kennedy, 2020) to push the literature on representation in government bureaucracy beyond its current geographic emphases and popular policy domains.

Despite its place among the world’s largest democracies, relatively little research has examined the inner workings of representation within Brazil’s public sector. And while most research focuses on the relationship between personnel traits and outcomes for service-recipients, our study demonstrates that women’s representation within the leadership of an organization is important in the pursuit of a more equitable working environment for the group as a whole. Specifically, we find that Brazilian municipalities with women mayors have smaller gender wage gaps in the municipal executive bureaucracy compared to those with men mayors. We also find that municipalities with higher levels of women’s representation on the city council have smaller wage gaps in the executive bureaucracy, despite city councilors lacking formal influence over this institution. One explanation for this finding could be that women city councilors encourage mayors (explicitly or implicitly) to reduce gender inequities in the executive branch. Women city councilors might also work to enact policies that facilitate more gender equality throughout the municipality, creating positive spillovers into the bureaucracy.

We also find no consistent patterns with respect to the mayor’s gender combined with attributes that are typically correlated with progressive attitudes. Younger, better educated, and left-leaning mayors do not systematically reduce gender disparities in the bureaucracy to a greater degree than their older, less educated, and center- or right-leaning counterparts. If anything, men from left parties have larger disparities than their non-left counterparts, providing evidence that leftist parties do not always outperform other parties on issues of gender equality (Funk et al., 2017). However, results presented in the Supplemental Appendix also suggest future research should take a closer look at the relationship between leader gender and progressive attitudes. An interaction between mayor gender and a latent variable for progressiveness generated by combining the mayor’s partisanship, age, and education via factor analysis shows that the gender wage gap under a man mayor is predicted to be relatively constant across all levels of progressiveness, but the wage gap under a woman mayor is predicted to decrease as she becomes more progressive, providing some limited support for hypothesis 2 (see Figure S3 in the Supplemental Appendix). Finally, incumbents and mayors further in their term are predicted to produce smaller wage gaps, indicating that mayors may need sufficient time or experience in office before they can effectively reduce gender disparities in the bureaucracy.

In sum, we find that the mayors most likely to hold progressive attitudes still have large wage disparities in their executive bureaucracies, while women mayors with differing attributes tend to produce smaller gender disparities. These results indicate that a key strategy for reducing gender wage disparities in the public sector may be increasing women’s representation in public offices. Research suggests that increasing women’s representation in both elected and bureaucratic leadership roles produces better outcomes for women in society (Keiser et al., 2002; Meier & Funk, 2017; Meier & Nicholson-Crotty, 2006; Wilkins & Keiser, 2006). At the local level of government, increasing women’s presence in elected and bureaucratic leadership roles could lead to improvements in the delivery of social services, the adoption of women-friendly policies, and increases in women’s numeric representation in other venues as well (Feeney & Camarena, 2021; Holman, 2014; Meier & Funk, 2017).

Finally, we highlight three limitations of this study that can guide future research on the relationship between representation and disparities within public organizations. First, this study was unable to determine whether the gender wage gap was due to women holding lower ranks than men (i.e., the glass ceiling) or due to women being paid less than men at the same rank (i.e., gender discrimination). Future research could aim to better identify the root causes of the wage gap. Second, we theorize two causal mechanisms might explain why women elected officials reduce disparities to a greater extent than men: they share policy preferences with other women, and promote women bureaucrats to higher offices because of this, and/or their lived experiences as women cause them to be more aware of, and interested in reducing, gender disparities. Future research could further disentangle and test these causal mechanisms. Third, we examine elected leaders’ gender, partisanship, age, and level of education, but several additional attributes and contexts could shape whether and how representation affects organizational disparities. Future research should continue to investigate how the attributes of elected leaders, including their gender, race, social class, attitudes, and preferences, affect disparities in the bureaucracy (in terms of wages, representation, and other outcomes) and also whether these effects vary across contexts.

Supplemental Material