Rank dependent utility, tax evasion and labor supply

In the simple Allingham-Sandmo portfolio model of tax evasion an expected utility maximizer will cheat more than what is estimated in empirical studies. Two main types of explanation have been suggested as solutions to this puzzle: (1) Tax payers act according to some non-expected utility theory, and (2) Individual ethical norms and social stigma induce people not to cheat. In the present study we test two hypotheses within these broad explanations: (1) Tax payers are weighting subjective probabilities of being penalised according to the rank dependent utility theory, and (2) Tax payers' beliefs about social norms have an effect on their decision to evade taxes. Our model is characterized by a simultaneous determination of tax evasion and labour supply, including the effect on tax payers of a social norm of not cheating. Using Norwegian survey data our hypotheses are corroborated. Our estimates imply that if the objective probability of being penalized is, say 3 %, the weighted probability is about 23 %. Our study provides an independent confirmation of the rank dependent expected utility theory. The model explains data 53% better than pure random choices and predicts hours worked in the regular economy, among tax evaders as well non tax evaders, rather precisely. The model is an example of a two sector choice model and the results indicate that an overall wage increase may shift labor supply away from the irregular part of the economy towards the regular.