Fair and efficient taxation under partial control: Theory and evidence

There is clear evidence that fairness plays a role in redistribution. Individuals want to compensate others for their misfortune, while they allow them to enjoy the fruits of their effort. Such fairness considerations have been introduced in political economy and optimal income tax models with a focus on income acquisition. However, actual tax-benefit systems are based on much more information. We introduce fairness in a tax-benefit scheme that is based on several characteristics. The novelty is the introduction of partial control. Each characteristic differs in terms of the degree of control, i.e., the extent to which it can be changed by exerting effort. Two testable predictions result. First, the tax rate on partially controllable characteristics should be lower compared to the tax rate on non-controllable tags. Second, the total effect of non-controllable characteristics on the post-tax outcome should be equal to zero. We estimate implicit tax rates for different characteristics in 26 European countries (using EU-SILC data) and the US (using CPS data). We find a robust tendency in all countries to compensate more for the uncontrollable composite characteristic (based on sex, age and disability in our study) compared to the partially controllable one (based on family composition, immigration status, unemployment and education level). We also estimate the degree of fairness of tax-benefit schemes in different countries. Only the Continental countries France and Luxembourg pass the fairness test, whereas the Baltic and Anglo-Saxon countries (including the US) perform worst.