Perfil do sonegador: um olhar sobre contribuintes inabilitados do cadastro do ICMS

Abstract

Tax Evasion makes the tax system unfair because it turns the tax burden heavier for some individuals, whereas others get economic advantage over market competitors. Increasing tax compliance by fighting evasion reduces the tax gap, which is the main goal of tax authorities. To accomplish that, one must classify taxpayers according to their tax behaviour. Modern studies demonstrate that tax compliance decisions are not solely based on maximizing economic outcome, but rather influenced by Economic, Psychological, Sociological and Demographic factors. This paper presents a study in which allthese factors have been taken in to account to builda mathematical model that is able to classify individuals from a population for which there has been few if any studies conducted so far: taxpayers of ICMS (Brazilian consumption tax) that have presented signs of business inactivity. For such, a sample of this population has been selected and used to identify which variables could best be used in a Logistic Regression Model to categorizeindividuals in one of two groups: taxpayers with signs of pure business inactivity and taxpayers with signs of evasion previous to business inactivity. The results show that the selected variables could be successfully used to categorizethe individuals from thesample with an Accuracy higher than 84%. Moreover, Goodness-of-fit tests show that the model fits the data very well and could be used in the whole population of the study. Thiswould allow Tax Authorities to respond to taxpayers according to their tax behaviour and that would be a possible mechanism to increase tax compliance and reduce tax evasion.