Uma avaliação de impacto do programa nacional de apoio à modernização administrativa e fiscal dos municípios brasileiros (PNAFM) sobre a arrecadação de tributos (1997 A 2022)

Abstract

The objective of this dissertation is to evaluate the impact of the National Support Program for Administrative and Fiscal Modernization of Brazilian Municipalities (PNAFM), managed by the Ministry of Finance, with resources from the Inter-American Development Bank (IDB), on tax collection in Brazilian municipalities. in the period from 1997 to 2022. To this end, two econometric methods were associated, in a first stage, Propensity Score Matching (PSM) was adopted, with the objective of finding in the cluster of non-participating municipalities, those that are similar to the municipalities treated in all relevant characteristics before joining the Program. Then, aiming to obtain a smaller and more qualified sample, Nearest Neighbor Matching (NNM) is applied, assigning weights to each training between PNAFM participants and non-participants, thus each treatment unit is paired with a treatment unit. control with the closest score, testing pairings with the algorithms, one-to-one with replacement, one-to-one without replacement and NNM with 2 nearest neighbors, and then applying the diff-in method. diff (DiD) in multiple periods of time, which is based on the fact that, in the occurrence of an exogenous event, it becomes possible to evaluate from two groups, treatment and control, the impact of the event via double difference estimation, the first being between groups and the second between periods, assuming parallel trends that potentially remain only after conditioning on the observed covariates. It focused on DiD with staggered adoption, that is, on DiD configurations such that, once treated, the units remain treated in subsequent periods. The results found indicate that the effect of the program on the tax revenue of the participating municipalities is null for the three phases of the Program. As a way of testing the credibility of our empirical strategy, zero impacts were estimated.