Audit Courts, Secret Voting, and Mayoral Accountability in Brazil
audit courts; municipal councils; secret voting; accountability; Brazil.
This study examines how the interaction between state audit courts and municipal councils shapes mayoral accountability in Brazil, in a setting marked by the 2010 “Clean Record Law” and the 2016 Supreme Court ruling (Topic 157), which
restored to municipal councils the final authority to judge mayors’ accounts. Using a dataset of 37,984 prior opinions issued by eight state audit courts, covering 3,103 municipalities (55.7% of all Brazilian municipalities), the first stage analyzes, for the 2010–2019 period, whether the jurisprudential shift changed courts’ propensity to recommend rejection and the relative importance of the Fiscal Responsibility Law (FRL) as a basis for their decisions.
In the legislative stage, these opinions are linked to council votes on mayors’ accounts, yielding an unbalanced panel of 387 municipalities in the states of Tocantins, Pernambuco, Espírito Santo and Rio Grande do Sul, with up to 33 fiscal years per jurisdiction and 1,437 observations with full information on controls. Linear probability models with municipality and year fixed effects are estimated, alongside logit and probit specifications with the same set of covariates.
Preliminary results indicate that the Supreme Court ruling is associated with a discrete but precisely estimated increase in the probability of adverse prior opinions. Time-series evidence suggests a partial substitution of outright rejections by approvals with qualifications, together with an increase in the share of rejections explicitly grounded in FRL violations, a pattern consistent with weaker enforcement of fiscal rules and a deterioration of municipal public finances. In the legislative stage, fixed-effects linear models and non-linear specifications do not, so far, detect a statistically robust effect of secret voting on the probability that councils reject mayors’ accounts: estimated coefficients are small and sensitive to specification, implying no conclusive evidence that the voting modality systematically alters the translation of the audit courts’ technical signal into political sanctions.