Lang : English
close× Call Us

Brief 09: Electoral Responses to Corruption Revelations

Ferraz and Finan assess the effects of the provision of information on corruption on election outcomes. They find that the release of audit reports prior to the election affects the performance of mayors seeking reelection.

Link to Full Study

Category: Corruption, Elections

Date of Publication: Friday, August 29, 2014

EGAP Researcher: Claudio Ferraz

Other Authors: Frederico Finan

Click to Download the Data

Geographical Region: South America

Research Question:

What are the effects of a national program that audits municipal governments and discloses information on corruption on the electoral outcomes of municipal elections?

Do voters hold their mayors accountable by punishing them when they try to seek reelection? How does the disclosure of the substantive findings of audit reports affect the reelection rate of mayors? Does the information environment, specifically the presence of local media, account for differential electoral responses to the release of these audits?

Preparer: Tara Lyn Slough



Much work on democratic accountability emphasizes the importance of information in allowing voters to be able to effectively hold their leaders accountable. However, this proposition is difficult to assess empirically because the information a voter has is often a function of a voter’s personal characteristics or their environment. In this study, the use of corruption audits that occurred in different municipalities as if by chance, that is randomly, allows for the identification of how the timing of exposure of corruption affects voting behavior.

In 2003, Brazil inaugurated a new program to audit municipal governments’ use of funds from the federal government. Each month, a number of municipalities are selected by lottery to be audited by the federal Controladoria Geral de União (CGU). Following inspection of accounts and completed public works projects, the CGU releases a report detailing their findings online as well as to the media. Between May 2003 and July 2005, the CGU conducted such audits in 669 municipalities; Ferraz and Finan exploit this randomized selection as a natural experiment and examine 373 of these audited municipalities that had a first-term mayor that was eligible for reelection in the October 2004 elections.

Research Design:

This study relies on the timing of the audits relative to the October 2004 municipal elections. The treatment group consists of the 205 municipalities that were audited before the election and the control group consists of the 168 villages that were audited between October 2004 and July 2005. The treatment is therefore the exposure of corruption (or lack thereof) within municipal governments before the election.

The content of the audits is measured as a count of the number of incidences of corruption (“irregularities”) described in the reports. On average, there were 1.74 such irregularities per municipality. The main measure of local media is the presence of an AM radio station within the municipality; these stations are present in 27% of the municipalities in the study.

The main outcome measured is the re-election rate, or success, of incumbent mayors.


Findings on Re-election Rates of Mayors: Mayors in the treatment municipalities, where the results of the audit were released before the election, were 3.6 percentage points less likely to be re-elected than mayors in control municipalities. However, this treatment effect is not statistically significant.

Findings on the Effects of the Audits by Degree of Corruption:Ferraz and Finan find a negative relationship between the number of incidences of corruption and the reelection rate of mayors in the treatment group. In the control group, where audits were conducted after the election, this relationship is not observed. The reelection rates of mayors are nearly constant and close to the overall average across all levels of corruption. This suggests that voter’s responses to the findings of the audit depended on the content of the report and the relation of these findings to their prior beliefs about corruption.

Findings on the Effects by Degree of Corruption and Presence of Local Media: The effect of the timing varies by the degree of corruption and the presence of a local radio station. In the treatment municipalities where the audit results were released before the election, radio appears to further deter electoral support for mayors in highly corrupt municipalities. This is depicted in the graph by the sharply falling reelection rates of mayors by number of irregularities in treatment villages with a radio station. This indicates that the media is one avenue through which voters gain information that they can use to hold elected politicians accountable.


Policy Implications:

In sum, this study lends support to theories of accountability that suggest that informed voters can hold politicians accountable by sanctioning poor-performing elected officials. Moreover, the presence of a local media can improve the reelection prospects for non-corrupt incumbents, suggesting that voters reward honest politicians. These findings continue to offer insights about how voters may respond to reports of politicians’ behavior at different levels of office and about how media coverage might condition these responses.

The policy implications from this study include ways to promote an informed electorate. Most centrally, this study emphasizes the importance of media alongside the availability of information on government performance as important in in this pursuit. To that extent, policies promoting transparency and exposing corruption (audits) play an important role in providing the electorate with information about the performance of their elected leaders. The findings of such audits should be advertised and made available to the general public. Moreover, the critical role of the media in providing this information to the public indicates that efforts must be made to ensure and protect the independence of media while promoting access to media.