Inequality Aversion and Marginal Income Taxation

This paper deals with tax policy responses to inequality aversion by examining the first-best Pareto-efficient marginal tax structure when people are inequality averse. In doing so, we distinguish between four different and widely used models of inequality aversion. The results show that empirically and experimentally quantified degrees of inequality aversion have potentially very strong implications for Pareto-efficient marginal income taxation. It also turns out that the exact type of inequality aversion (self-centered vs. non-self-centered), and the measures of inequality used, matter a great deal. For example, based on simulation results mimicking the disposable income distribution in the US in 2013, the preferences suggested by Fehr and Schmidt (1999) imply monotonically increasing marginal income taxes, with large negative marginal tax rates for low-income individuals and large positive marginal tax rates for high-income individuals. In contrast, the often considered similar model by Bolton and Ockenfels (2000) implies close to zero marginal income tax rates for all.