Principal investigator: Pedro Jorge Holanda Alves
Article title: ELECTORAL ACCOUNTABILITY AND CORRUPTION: EVIDENCE FROM THE AUDITS OF LOCAL GOVERNMENTS
Article authorsClaudio Ferraz and Frederico Finan
Location of the Intervention: Brazil
Sample Size: 476 observations
SectorEconomic Policy and Governance
Type of InterventionThe effect of institutional electoral politics on corruption.
Main Variable of InterestTotal amount of resources related to corrupt activities.
Evaluation method: Experimental Evaluation (RCT)
Policy Problem
When we study economics, we begin by understanding that goods are scarce and that we need to make decisions that best satisfy our respective desires. When we introduce the public sector, we add the detail that the State must make the decision that best satisfies the population that will benefit from these policies. Democracy enters as the political factor that provides the population with the right to define which agent will represent society in deciding these policies.
In reality, we know that this is not always the case. In many instances, we encounter situations of abuse of political power, corruption, and any other factor that becomes a threat to modern democracies. Underdeveloped countries, in particular, serve as examples because there are numerous cases in which political elites divert public funds intended for health, education, and security to obtain private gains. It is believed that both the extent of corruption practices and the means of correction are associated with electoral systems, since the way the electoral system is determined or the decisions adopted by the population can be decisive factors in whether a politician chooses to become corrupt or not.
For this reason, Ferraz and Finan (2011) seek to examine the effects of electoral accountability on corruption in local governments in Brazil. For their analysis, the authors used reports from the Anti-Corruption Program, which randomly audits Brazilian municipalities, conducting analyses related to fraud by municipal managers. The objective is to verify the incentives for reelection by comparing mayors who are in their first term with those who are in their second (and last) term, and evidence of corruption among these mayors.
Assessment Context
Since the 1988 Federal Constitution, Brazil has adopted policies aimed at both institutional structure and state fiscal oversight. In 1997, the possibility of reelection was introduced, and in 2003, the Comptroller General of the Union (CGU) created the anti-corruption program with the objective of randomly auditing Brazilian municipalities. The combination of these two policies allows for a comparison of corruption levels between 2001-2004 for mayors in their first term with those in their second term.
Even though a mayor can only hold office for two consecutive terms, there is the possibility of returning after a one-term hiatus. However, only 12% of politicians who completed their second term in 2004 decided to run for election in 2008 after the hiatus, indicating a low probability of returning to political office in the future and showing that the average mayor treats their second term as if it were their last term.
Policy Details
Created in May 2003, the anti-corruption program, implemented by the CGU (Brazilian Federal Comptroller General), aims to reduce the misuse of public funds among municipal public administrators and encourage society to participate in monitoring public spending. In its first year, the program audited 26 randomly selected municipalities, each in a different state of Brazil.
Through lotteries conducted by Caixa Econômica Federal in Brasília, the program expanded and audited 50 municipalities in the second period and then another 60 municipalities, all of them with fewer than 450.000 inhabitants. To ensure that the audit will work, the press and local population are invited to witness the lottery and certify that all 10 to 15 auditors from the CGU will examine and record the information as reliably as possible.
After a week of inspections, a detailed report describing all irregularities is sent to the CGU (Brazilian Comptroller General's Office), which forwards it to the TCU (Federal Court of Accounts), the Public Prosecutor's Office, and the municipal legislative branch. The main conclusions of the audit are published online and disseminated in the media. It is these reports that Ferraz and Finan (2011) used as a basis for measuring corruption.
Methodology Details
Based on the coding of the reports, Ferraz and Finan (2011) defined the total amount of resources related to corrupt activities as their main variable of interest. They also report two other corruption indicators: the number of irregularities related to corruption and the sharing of service items related to corruption. To measure non-visible violations, they used a mismanagement indicator, which represents the number of violations divided by the number of services audited.
The main objective is to analyze whether incentives for reelection affect political corruption in municipalities. Ideally, to estimate these results, municipalities would be randomly assigned the possibility of reelection, and then the difference between the two groups of municipalities (first and second term) would be measured. Even if the selection of municipalities by the CGU (Brazilian Federal Comptroller General) is random, the experiment would be biased due to the nature of the data. To solve this problem, the authors use a Regression Discontinuity (RDD) calculation, which allows for an approximation of a randomized experiment, comparing only mayors who won reelection by a very small percentage with those who lost by a very small margin.
This model allows for the comparison of two groups that, for some exogenous (uncontrollable) reason, are very similar but had different outcomes. To adapt the idea using Regression Discontinuity, we define mayors who, in their second term, lost or won in the 2000 elections by a very small percentage. Using this finding allows us to consider that the two individuals are similar and that the only difference between them occurs due to small or negligible electoral factors.
For municipalities that narrowly won, the treatment would involve mayors who are in their second term, while those that narrowly lost would be replaced by a new mayor, who constitute the controlling group, as they are in their first term.
Results
The results show that mayors with incentives to run for reelection are less corrupt than those without such incentives. Mayors in their first term have, on average, a 27% lower share of misappropriated funds than mayors in their second term. Considering that municipalities receive, on average, US$2 million in federal transfers, mayors in their second term steal US$55 more, and the incentive to run for reelection reduces corruption by US$150 million throughout Brazil, equivalent to half of what the federal government spent in 2002 on the Bolsa Escola income transfer program.
It is also evident that incentives for reelection vary according to differences in local institutional contexts, for example, the local public and media. In general, in municipalities without a local media presence, incentives for reelection reduce political corruption, while, on the other hand, mayors who won by larger margins can afford to be more corrupt.
Lessons in Public Policy
Corruption is a recurring and important theme in empirical studies. Resources that could have been allocated to schools or hospitals are diverted and used for personal gain. The abuse of power by elected politicians is a central issue in a large number of countries, making it necessary to seek measures that reduce corruption and generate better investments for the population.
Corruption is linked to several factors, ranging from issues related to democratic institutions to the absence of popular oversight and punishment by the justice system. In this work, Ferraz and Finan (2011) highlight how the structure and electoral accountability, related to the possibility of reelection, can affect the discipline of politicians and control their behavior in diverting public funds.
Their findings are important because politicians in their first term are associated with less corruption, indicating that the second term, being the last, makes mayors more likely to commit irregularities.
Reference:
FERRAZ, Claudio; FINAN, Frederico. Electoral accountability and corruption: Evidence from the audits of local governments. American Economic Review, vol. 101, no. 4, p. 1274-1311, 2011.
