The dynamic model presented in this paper intends to account for the evidence, which appears to be particularly significant for Italy, showing that the incidence of tax evasion in a certain region is negatively correlated to the level of social capital existing in the region. Beside including social capital among the determinants of tax evasion, we extend the model so as to incorporate a mechanism whereby a high level of tax evasion depresses the formation of social capital, thus helping to explain how regional differences in the endowment of social capital and in the incidence of tax evasion co-evolve and why they tend to be highly persistent. The model seeks also to capture the fact that in a democracy the political determination necessary to effectively repress tax evasion depends on the voters’ propensity toward the phenomenon. Indeed, one should expect that–in areas where a relatively large (small) number of citizens are tax cheaters—the political consensus in favor of tough policies against tax evasion tends to be weak (strong) and short (long) lasting. Consistently with this intuition, the model shows that regions where social capital is relatively low and tax evasion is relatively high can do better in the long run (i.e., they can reach a steady state characterized by a higher level of social capital and a lower level of tax evasion) when tax-enforcement policies are determined at the national level rather than at the regional level. The opposite is true for regions where social capital is relatively high and tax evasion is relatively low.

Key words: Tax compliance, dynamic models, multiple equilibria, tax-enforcement policies.