Tax evasion, behavioral microsimulation models and flat-rate tax reforms. Analysis for Italy

It is sometimes argued that a flat-rate tax reform can reduce tax noncompliance. The argument is, however, inconsistent with the so-called Yitzhaki' s puzzle of the classical expected utility (EU) model. The latter predicts an increase, rather than a reduction, in tax evasion following a cut in the tax rates resulting from a flat-rate reform. We study the impact of a flat-rate tax in a microsimulation tax-benefit model of Italy which allows us to analyse various hypotheses of tax evasion behavior. In addition to the EU model, we analyse expected utility with rank dependent probabilities (EURDP) and the model of reference dependent (RD) preference, the most favourable to overturn Yitzhaki' s puzzle. Our simulations show that a flat-rate tax would barely reduce overall evasion in Italy in all models considered. Redistributive effects are in all cases large.