Risk, entitlements and fairness bias: Explaining preferences for redistribution in multi-person setting

Researchers frequently studied the casual relationships of other-regarding preferences by applying experimental methods in bilateral settings (e.g., dictator game and ultimatum game). We use a framed experiment on taxes to study preferences for redistribution in a multi-person setting. We find presence of heterogeneous preferences with a substantial share of tax rate choices in line with both payoff maximization and other-regarding preferences. Notably, our data is not consistent with inequality aversion but points to other forms of other-regarding preferences, as fairness and altruism. By manipulating how subjects are assigned to a given level of pre-tax income, we vary the individual entitlements. We find a difference in the willingness to redistribute income when comparing the treatment where pre-tax income is assigned by relative performance in a production task (a general knowledge quiz) to the treatment where pre-tax income is assigned by luck. We do not find any significant difference in comparison to the intermediate treatment where pre-tax income is assigned by a combination of luck and performance. The perception of a fair tax is different depending on whether subjects' pre-tax income is below or above average, which is in line with a fairness bias. Finally, subjects not knowing whether their pre-tax income is below or above the average when choosing the tax rate behave as if they were more other-regarding.