Sousa et al based on

Sousa et al. (2008), based on a model by Fisman and Wei (2004), use commerce data between Brazil and the United States and data on import tariffs in order to measure the impact of tariffs on tax evasion. They showed that higher tariffs imply higher degree of evasion. In addition, this relation is not linear, so the impact is meaningful only after some level of import aliquot. Siqueira and Ramos (2006) extends the A–S model and find a result that points to the opposite direction. They showed that an increase of the marginal aliquot reduces tax evasion and, in addition, that an increase of the probability of detection and of the penalty fine also leads to a GSK2606414 of tax evasion. The differences between these results may reflect the income and substitution-effect pointed out by Yitzhaki (1974). Richter and Boadway (2005) use the A–S model as well as Yitzhaki\'s in order to study the interaction between tax evasion and tax structure. Under Yitzahki\'s framework, the optimal tax design remained invariant with respect to the introduction of risks inherent to tax evasion. Under A–S framework, on the other hand, it showed a trade-off between tax distortion and the magnitude of tax evasion. Goerke (2003) studies what happens with the amount of labor in the market as the tax structure becomes more progressive. When opportunities to evade are introduced into the model, employment increases as taxes become more progressive. In particular, this result holds only when part of the penalty fine is dependent on the non-reported income, as in the A–S model. From these two papers, it is possible to conclude that tax evasion influences the tax design and its impact on taxpayers. Most of the papers focus on the individual decision-making. Schneider and Klinglmair (2004) estimate the size of the informal labor market in 110 countries and show that the size varies with the country. Sandmo (2005) shows that these variations cannot be explained by the magnitudes of the tax rates and fines alone. Cowell (1990) emphasizes that tax evasion requires a theory of social interaction, since it is a social phenomenon. Therefore, part of the evasion could be explained by factors related to the social interaction between agents. In the A–S model, the taxpayer gets to an opinion about the probability of detection also by observing the other agents and their probabilities of being audited. Then the taxpayer\'s subjective belief of being detected depends on his own evasion and the evasion of others. If he perceives that the non-reported income by others increases, his subjective belief of being detected is reduced and his non-reported income increases. On the top of that, there is a disutility from not reporting the true income, though this could be lower in case he perceives that many other do not report truthfully. In their study on corruption, Andvig and Moene (1990) also find the same pattern: the more corrupt the environment the individual is in, the harder hypotonic is for the individual to be honest. A dilemma extensively studied in the literature of tax evasion is the existence of people who declare fully their income if the expected value of the utility when the taxpayer do not report part of its income, is positive. According to Andreoni et al. (1998), there are moral and social factors that influence the decision to evade. Among these factors are the feelings of guilt and shame that agents feel by not declaring all their income. There is a disutility when the agent feels he did something wrong. Furthermore, the perception of fairness in the tax burden from the taxpayer also influences their decisions. If he perceives that its tax burden is unfair compared with the tax burden of others, or if it perceives that others do not fully declare their income and therefore is at a disadvantage, there is an incentive to evade his income. Another factor that influences the amount of reported income mentioned by Andreoni et al. (1998) is the satisfaction of the taxpayer with respect to government policies. The misuse of taxes by the government is another incentive to circumvent the system of tax payments. Another study that explains evasion as a social phenomenon is that of Barth et al. (2005) who consider the case of two people who receive the same income, the one working longer and having a lower remuneration and other working less but getting more for time worked, and both paying the same amount of taxes. The first group feels wronged and has an incentive to lie about its income. All these analyzes consider the interactions among taxpayers, not just individual motivations to explain tax evasion.