Transparência da gestão executiva municipal: um estudo dos Portais eletrônicos como promotores da accountability

Abstract

This study aims to identify which socioeconomic indicators are related to the level of fiscal transparency in municipalities. A non-intentional and probabilistic sample of 360 municipalities was stratified according to the Brazilian state. Through the multiple linear regression technique, it was verified which variables would be related to public transparency. Using an internationally recognized model as a mechanism to improve the transparency of public accounts, the transparency index was determined, the variable being used as the dependent one. The explanatory variables were used to apply the linear regression technique: total income, net current income, real assets, total population, literate population, IDH-M, per capita GDP and per capita income. Through the observation of the electronic portals of the municipalities, the average transparency was 3.76, a lower index than the state average. The low transparency index corroborates the assumptions of the public choice theory: to the manager, poor or poorly informed voters "facilitate" the acceptance of a manager by the population, allowing the manager to obtain the trust of these unsuspecting voters. The regression model explains 11.9% of the variation in transparency index. When sectioned by quartz interval based on the population, the regression model explains 16%, 18.1%, 15.5% and 15.9% of the variation in the transparency index, respectively. Based on the results discussed, it can be assumed that the initial assumption of this research that there is a relationship between the transparency index and explanatory variables is correct when considering the entire sample. When analyzing the intervals, the general hypothesis is rejected in municipalities with a population equal to or less than 5,419 (quartile 1) inhabitants or with a population equal to or greater than 10,550 inhabitants (quartiles 3 and 4). For the municipalities with population between 5,475 and 10,514 inhabitants (quartile 2), the initial hypothesis was accepted. In this situation, the transparency index is related to all variables, when analyzed together. However, the influence of each variable in isolation is not significant.