Principal investigator: Viviane Pires Ribeiro
Article title: What happens when a woman wins an election? Evidence from close races in Brazil
Article authors: Fernanda Brollo and Ugo Troiano
Location of the intervention: Brazil
Sample size: 723 elections
Sector: Genre
Type of Intervention: Analysis of the relationship between the gender of government officials and policy outcomes.
Primary variable of interest: Gender of rulers
Evaluation method: Discontinuous Regression
Evaluation Context
The Brazilian federal presidential system operates within a multi-party system. The layers of political and administrative organization in Brazil are the federal government, the states, the federal district, and the municipalities. Municipalities are smaller federative units with autonomous local government, governed by a mayor, directly elected by citizens for a four-year term, and a legislative body, also directly elected by voters.
In municipalities with more than 200.000 voters, mayors are directly elected through a second-round system, while in municipalities with fewer than 200.000 voters, mayors are directly elected through a plurality system. Elections for president, governors, and members of Congress occur simultaneously every four years, while municipal elections are staggered by two years and also occur every four years. Before 1998, Brazilian mayors could not run for reelection, but after 1998, mayors were allowed to run for a second term.
Intervention Details
Brollo and Troiano (2016) use a microdata set from 5.567 Brazilian municipalities to analyze whether the gender of the elected official affects policy outcomes and whether men and women respond differently to local electoral incentives. To this end, the authors analyzed two terms of municipal administration in municipalities with fewer than 200.000 voters: 2001-2004 and 2005-2008. Thus, only contests with two candidates of opposite genders were considered, totaling a sample of 723 elections, representing 7% of the contests in each term. The electoral data are from the Superior Electoral Court.
This study focuses on discretionary transfers allocated to infrastructure projects, which represent approximately 15% of total municipal infrastructure expenditure. These transfers are related to budget items involving the construction of buildings and bridges, road paving, improvements to water and sewage systems, the purchase of ambulances, etc. Data on infrastructure transfers self-reported by the municipal administration were obtained from the National Treasury portal – FINBRA database.
For the analysis of health outcomes, data from the Live Birth Information System (SINASC) were used. This dataset contains monthly information on the number of prenatal medical visits and the gestational period (weeks). The variable used in the analysis is the proportion of pregnant women without prenatal medical visits and deliveries not considered premature (at least 37 weeks).
The data on corruption comes from random audits of municipal governments. The main categories of irregularities described in the audit reports are: illegal bidding practices; fraud; overbilling; and embezzlement. The definition of corruption is a dummy variable that indicates whether at least one episode of any of the types of irregularities was detected by the auditors.
Methodology Details
To analyze the gender effect of municipal leaders on the outcomes of Brazilian policies, Brollo and Troiano (2016) used a large dataset in a Regression Discontinuity design of close electoral contests.
The authors focused on political outcomes that depended on local government efforts and for which administrative data exists at the municipal level. First, they examined discretionary infrastructure transfers from the federal government, because the mayor's effort is a significant determinant of the amount of transfers municipalities receive from the Union. Second, they analyzed health outcomes, focusing particularly on health services related to prenatal care. The public health system in Brazil is decentralized; spending is primarily financed by the federal government, but municipalities are responsible for all decisions regarding resource allocation. Thus, municipal policies are a significant determinant of health outcomes. Third, they analyzed gender differences in corruption, using data from random audits of municipal governments that allowed them to construct an administrative measure of corruption. Fourth, they analyzed turnover in temporary public sector employees, since temporary hiring is a widely used instrument of political patronage (exchange of goods and services for political support) in Brazil. Finally, they analyzed whether the decision to run for reelection and the probability of reelection differ between female and male mayors elected after close races of both genders.
To analyze whether incentives for local elections motivate candidates differently, the research focused on two heterogeneous treatment effects that can capture the incentives provided to politicians by municipal elections. First, it sought to study term limits, since the ability to run for reelection can affect the policies implemented by local politicians. Mayors in Brazil can only run for one consecutive term once, therefore, the authors compared gender differences between mayors in their first and second terms. Second, it analyzed whether mayors behave differently as elections approach. Municipal elections are held every four years, so the authors sought to compare behavioral differences between both genders in the pre-election years (the last two years of the term) and the non-election years (the first two years of the term).
Results
The results show that there are significant gender differences in terms of policy outcomes. First, female mayors were found to attract twice as many discretionary transfers from the federal government as male mayors. Second, the results indicate that having a female mayor leads to better health outcomes related to prenatal care. In particular, municipalities with a female mayor experienced a 61% decrease in the number of women without prenatal medical appointments and a 1,3% increase in the number of regular (i.e., non-premature) births. These positive health effects are concentrated among mothers with lower levels of education, who are more dependent on public health services. Third, female mayors were found to be less likely to engage in administrative irregularities. The probability of observing an episode of corruption is lower (33% to 28%) in municipalities with female mayors than in municipalities with male mayors.
Analysis of gender differences in public employment indicates that male mayors hire approximately 50% more temporary employees to work directly in municipal administration than female mayors. However, no significant gender differences were observed when analyzing permanent public employees. On the other hand, when analyzing re-elections, no gender differences were observed in the probability of running for re-election. The results show, however, that female candidates elected after a contested election have a lower probability (50%) of being reelected compared to their opponents.
The study suggests that politicians of both genders may respond differently to local electoral incentives. First, it was found that in their first term, when they can run for reelection, male mayors tend to hire more temporary staff to work directly in municipal administration compared to female mayors. This finding suggests that male mayors may be promoting greater political patronage to be re-elected than female mayors. It was also observed that differences in prenatal care visits between municipalities with female and male mayors only emerge for politicians in their second term, which is consistent with the idea that "lame-duck" mayors may perform worse than their female opponents. Finally, it was found that male mayors tend to hire more temporary public employees (compared to female mayors) in the two years prior to the election, rather than the first two years of their term.
Lessons in Public Policy
Brollo and Troiano (2016) contribute to the literature on women's participation in politics, highlighting how local electoral incentives can affect political decisions and outcomes in a competitive environment. Local elections can increase accountability, helping to align politicians' actions with voters' preferences, but they can also generate incentives for politicians to behave strategically. That is, the effect of local elections on politicians' behavior can differ between genders – men and women respond differently to electoral incentives.
The analysis indicates that municipalities governed by female mayors receive more discretionary transfers and have better health outcomes. On the other hand, mayors tend to hire more temporary public employees before municipal elections and are more likely to engage in corruption.
Given that the research adopts a scenario where there is fierce electoral competition between candidates of both genders, the authors emphasize that it cannot be stated that the results found also apply to a scenario with quotas reserved for women in politics. Therefore, an interesting direction for future research would be to understand whether policies aimed at increasing female participation through quotas that restrict competition between both genders have different implications than policies aimed at increasing the number of women competing in open elections.
References
BROLLO, Fernanda; TROIANO, Ugo. What happens when a woman wins an election? Evidence from close races in Brazil. Journal of Development Economics, vol. 122, p. 28-45, 2016.
