Política de dividendos no Brasil: As instituições financeiras são diferentes?

Abstract

This study aims to analyze the determinant factors of the dividend policy of two sample groups being the first comprised of financial institutions and the second by all of the remaining publicly traded companies. This analysis is carried out in order to verify whether financial institutions are different from all of the other companies in terms of dividend policy’s determinants. Through Tobit Models with unbalanced panel data, this study has analyzed the Brazilian publicly traded firms listed in B3 between the years 2000 and 2016, totaling 3,167 firm-year observations. The results suggest that the main determinants of the dividend policy reported in the literature influence the dividend policy of the two sample groups differently. All of the variables of interest considered in the main econometric model have influenced the payout of the non-financial companies, while only the stability of the dividend policy (EPD) has influenced the payout of financial institutions in the majority of the specifications of this research. The EPD variable combined with the fact that the company is a financial institution has a negative differential effect on its payout. This indicates that the effect of this variable on payout has been significantly greater for the non-financial companies than for financial institutions, thus reinforcing that, at least in what refers to this factor, the sample groups are in fact different. However, by comparing the results of regressions of the sample groups to regressions estimated for the complete sample, it can be observed that the results of non-financial companies are similar to those of the complete sample in terms of signal and statistical significance of the variables’ coefficients. Thus results suggest that the exclusion of financial institutions from studies on dividend policy has not result in significant changes in the econometric estimations. In order to verify the sensitivity of results to the chosen period of analysis, the complete period has been segmented into five subperiods chosen according to variations of the Brazilian economic scenario. There was a reasonable variation in the determinant factors depending on the analyzed subperiods, indicating that different economic scenarios can influence the dividend policy of the sample groups differently, and that determinant factors are not constant in time.