Principal investigator: Pedro Jorge Holanda Alves
Article title: NORMS VERSUS ACTION:WHY VOTERS FAIL TO
SANCTION MALFEASANCE IN BRAZIL
Article authors: Taylor C. Boas, F. Daniel Hidalgo and
Marcus André Melo
Location of the Intervention: Brazil
Sample Size: 3.200 comments
SectorEconomic Policy and Governance
Type of InterventionThe effect of political infraction on
voter behavior
Main Variable of Interest: Votes for mayor
Evaluation method: Experimental Evaluation (RCT)
Policy Problem
Within the global political landscape, the democratic system prevails in most countries. However, the system suffers in some situations because it allows political agents, instead of generating benefits for society, to corrupt the system. This can be done through bribery or embezzlement of public funds, or by using their authority for private gain. Others may not be directly involved in corruption, but they violate the laws, ignoring budget targets, failing to pay employees, and so on.
One solution to these problems is to increase voters' level of information about how local government is implementing public policies. In various scenarios, studies on the subject have shown that politicians accused of corruption are penalized in their reelection bids when local or national media cover the scandals. Thus, the population of the region will have the opportunity to punish politicians who break the law. Under pressure, it is expected that politicians will be less corrupt in their decision-making.
Another point worth highlighting is the change in voter behavior when considering hypothetical and real scenarios. Hypothetical psychological experiments are suitable for analyzing voter behavior, since this type of analysis allows us to see how these oversight norms translate into actions – or the lack thereof.
Assessment Context
In Brazil, since the 90s, efforts have been made to create laws and institutions capable of punishing corrupt actions by public entities. Among these institutions, the Federal Court of Accounts (TCU) and the State Courts of Accounts (TCEs) stand out, responsible for monitoring government behavior based on the application of laws related to the budget and public administration. The punishments resulting from audits conducted by these institutions have serious consequences for politicians.
The evolution of oversight institutions has shown progress. For example, in 2010, if the TCU (Federal Court of Accounts) or TCE (State Court of Accounts) rejected the mayor's public accounts, the Clean Slate Law allowed candidates to be barred from running for office for 8 years. Even though the law has been weakened years later, it still holds great importance in combating corruption.
For an empirical application of the problems presented, Boas et al. (2018) examined the effect of information about the bad faith of the incumbent mayor on the behavior of Brazilian voters. Based on the implementation of a field survey in the state of Pernambuco (PE), the authors were able to apply an experiment in which voters are informed about the irregularities of mayors running for reelection in 2016. Simultaneously, voters are also asked what behavior they would adopt if a hypothetical mayor had their accounts rejected by the Federal Court of Accounts.
Policy Details
This research was conducted in conjunction with the TCE-PE (Court of Accounts of the State of Pernambuco), which, in addition to providing assistance and support in the research, also indicated the number of municipalities that had their accounts rejected. For the application of the hypothetical experiment, the authors used a scenario in which the interviewee must imagine a city similar to their own where the mayor has done a good job, but the State Court of Accounts has rejected the mayor's accounts in 2013 due to serious problems in budget management. Based on this hypothetical situation, the interviewees were asked, on a scale of 0 to 4, what the probability was of them voting for the mayor of that municipality.
The experiment is divided into two groups, where one group receives information about irregularities by the mayor and the other group does not receive this information. To avoid the contamination effect between the comparability of field and hypothetical experiments, the authors examine the hypothetical effects of the treatment only in municipalities that had accounts rejected but never received a pamphlet with this information.
In order to obtain the best results, the authors presented information related to irregularities only 2 to 3 weeks before the election, a period in which candidacies had already been declared and voters were likely deciding which was the best option for their vote. Opting for this strategy seems ideal, since information at a time of choice has a greater impact than an accusation disseminated by the media several years before the election. Another way to obtain better results was due to the care taken in formulating the questions and details, in order to generate the best possible understanding and not generate campaign propaganda.
Methodology details
The respondents were selected randomly with the same probability of selection within the municipality's census tract, separating them into three groups: a treatment group that received information about the approval or rejection of their mayor's accounts, a pure control group that received no information, and a treatment group that received information about the performance of the municipality's schools. The application of randomness allows for simple and consistent estimations using the Average Treatment Effect as differences between means, controlling only for the fixed effects of census tracts, since randomness eliminates any problem of inconsistency in the estimates.
The research on mayors' accounts comprises 3.200 adult voters in 47 municipalities in the state of Pernambuco. The database includes municipalities where the mayor ran for reelection in 2016 and the Pernambuco State Court of Auditors (TCE-PE) had already judged the 2013 accounts. Therefore, all seven municipalities where the mayor's accounts were rejected but who still opted to run for reelection are included, and another 40 municipalities where accounts were approved were randomly selected. Overall, 40 voters were interviewed in municipalities where accounts were approved, and between 80 and 416 voters were interviewed in municipalities where the mayor's accounts were rejected, totaling 1.600 respondents for each group.
A second round of research was conducted between 2 and 4 weeks after the election, interviewing 2.577 individuals who were asked whether or not they voted for the current mayor. To reduce response bias, the authors used printed ballots, which respondents were asked to deposit in an envelope with their answers.
Results
Hypothetical results show that telling respondents that the hypothetical mayor's accounts were rejected reduces the likelihood of voting for him by 44 percentage points. However, in the field experiment, informing respondents about the approval or rejection of their mayor's accounts has no significant effect on voting behavior. Although the result indicates the expected direction, these effects are substantially small and statistically insignificant. In particular, the estimated effect of handling account rejection on voting for the incumbent is almost exactly zero.
But why do rules regarding punishment affect the hypothetical scenario but not influence the behavior of those elected on election day? The authors list two main factors. First, even if voters consider corruption an act of bad faith, these same people who directly experience it consider corruption a low-priority concern compared to health, job creation, and other issues that politics affects. Second, political factors related to partisanship can cause voters to choose their vote based on party identification, limiting the potential for information to change electoral behavior.
When individuals are confronted with hypothetical scenarios, the results indicate that these voters are willing to punish incumbent politicians who encounter improbable public accounts; however, when this hypothetical scenario is compared to the real situation, the results do not hold true, indicating that informing voters about the acceptance or rejection of the public accounts of the mayor of the city where the individuals live does not alter the effect on the decision to vote for the mayor's reelection. The authors highlight that the divergence between the scenarios may be attributed to concerns related to job creation and health services.
Lessons in Public Policy
From this experiment we can draw some lessons. On the one hand, field experiments clearly reflect the best option for understanding how voter information affects a real election. On the other hand, the hypothetical experiment offers clues as to how people would act in more general cases, where the individual would act independently of the source, partisanship, different types of corruption, or other factors. Factors imposed in hypothetical situations are impossible to manipulate experimentally in the real world. In other words, even if the electorate wants the mayor to be punished for some government irregularity, this does not always happen in the real world. The existence of factors related to partisanship or political needs means that even knowing about the irregularity, the individual becomes indifferent.
Reference
BOAS, Taylor C.; HIDALGO, F. Daniel; MELO, Marcus André. Norms versus action: Why voters fail to sanction malfeasance in Brazil. American Journal of Political Science, vol. 63, n. 2, p. 385-400, 2019.
