Policy Wonk
WorkedGovernance & participation

Brazil's random audits of city governments

Headline result
−7 pp
re-election chance, mayors with two corruption violations exposed pre-election
Quasi-experimental
Implemented by
Controladoria-Geral da União (Office of the Comptroller General), Brazil
Type
National government
Where
Brazil
Region
Latin America & Caribbean
When
2003–present
Evidence
Quasi-experimental

The challenge

Corruption in local governments' use of federal funds was widespread and rarely exposed.

The case

In 2003 Brazil's federal comptroller began choosing municipalities by public lottery for detailed audits of how they spent federal transfers, and published the findings. Researchers compared mayors whose audits were released before an election with those audited after. Where corruption was revealed before the vote, mayors were significantly less likely to be re-elected, especially where local radio spread the news. Where audits reported two corruption violations before the vote, the policy reduced the incumbent's likelihood of re-election by seven percentage points, or 17%, versus the control group.

anti-corruptionauditstransparency

Why it worked

Random selection made audits credible and hard to manipulate. Publishing results gave voters information they used, but only where local media carried it.

Sources

More on governance & participation

All governance & participation case studies