It is now well established that people differ with respect to their tax compliance behaviours. A recent stream of research has shown that individuals' tax compliance behaviours are affected by the behaviour of other taxpayers (see Alm et al2016}). For instance, individuals seem more likely to be (dis)honest the more the other taxpayers are (dis)honest in their declaration. Starting from the observation of taxpayers' heterogeneity in terms of honesty and of the existence of a perverse social dynamic leading the most honest to evade, in this study, we propose a simple mechanism to restore honesty. We use experimental method (i) to test the willingness of subjects to voluntarily contribute to a fund dedicated to finance the fight against tax evasion and (ii) to analyze whether the compliance with social norms may explain the contribution of individuals.

More precisely, we conduct two experiments in a between-subject design. In the first experiment, we ran a pure declaration game, without redistribution and so without public goods, with fixed groups of individuals who are fiscally identical and who cannot identify their other group members. At the beginning of each period, participants have to determine the amount of income they will self-report to the tax authorities, knowing the fiscal parameters. At the end of each period, they are informed about whether they have been audited, their net payoff and the average income reported by the other members of their group. This process is repeated over 20 periods. Because groups are fixed and taxpayers learn the average declaration of their peers, they may be influenced by their behaviour. Such information may induce a taxpayer to decrease his future declaration and may therefore undermine tax collection. To break this vicious circle, at the end of the 10th reporting period, they have the opportunity to contribute a fixed amount to a special fund dedicated to fight against tax evasion. When collective contributions reach the threshold determined exogenously by the experimenters, then the administration is supposed to have enough additional funds to increase the audit probability.

While Bazart & Bonein (2017) have shown that contributions may be explained by inequality aversion, in this study, we investigate whether such contributions could reflect some compliance with social norms. For instance, honest individuals may want to punish evaders because they do not comply with their norm of honesty. To test this conjecture, we conduct a second experiment to elicit subjects' perceptions of the appropriateness of contribution to the special fund, using the social norm elicitation task introduced by Krupka & Weber (2013).

Overall, 30% of subjects contribute to this fund. Using a probit regression analysis, we find that the level of declarations and the difference in declarations explain, at least in part, the contribution decision. Data obtained in the social norms elicitation task allow us to test the impact of social norms in the contribution decisions. Results of conditional logit models capture well the general qualitative properties of the data obtained in the declaration game experiment.