SUMMARY
This study seeks to identify the effects of the mandatory dividend in determinants of companies listed on the BM&FBovespa to pay dividends, demonstrating that the methodology proposed is a step forward in the pursuit of these determinants, mirroring the reality of actual willingness to pay. To accomplish this was used a tobit model with instrumental variables, separating the sample into three groups: the Complete Sample, ABOVE group and MINIMUM group. In total, were analyzed 1531 dividend distributions from 2007 to 2013. It was found that there is strong empirical evidence that companies paying dividends above the mandatory feature different determinants of those who only pay this. Thus, the existence of companies in the sample who pay only the mandatory dividend influence on the regression coefficients, resulting in significant biases for dividend determinants in Brazil. Because companies that pay above mandatory are the real makers of the decision to pay dividends, as companies only pay the mandatory simply comply with the legislation.