Abstract (Translated):
The distribution criterion of the Municipal Participation Fund (Fundo de Participação dos Municípios) has biased the population data for small Brazilian municipalities. The paper applies the test for manipulation of the running variable in discontinuity regression (McCrary, 2008) to Brazilian municipalities' data. It shows that previous censuses suffered the same distortion and it is getting worse. For the 2010 Census the paper estimates that 192 out of the 3565 municipalities were misclassified, generating a misallocation of more than US$100 million in 2010 in that year.