The Social Multiplier of Tax Evasion: Evidence from Italian Audit Data
Giulio Zanella (University of Sienna)
Abstract
In this paper we investigate the role of social interactions in tax evasion. In our model the probability of being audited decreases with local tax evasion, because of a congestion effect. This is a parsimonious way of introducing social interactions, and is a testable assumption. Identification of the structural parameters exploits a novel method based on comparison of within- and cross-group variances of individual behavior (Graham, 2007). Such method is particularly convenient when dealing with scant administrative datasets lacking relevant contextual and individual information. This also allows to mitigate the problem of group and individual unobservables that usually plague cross-sectional studies of social interactions. We estimate the parameters of the behavioural equation by using a unique dataset of tax audits to about 80,000 self-employed individuals in Italy. We find that the social multiplier is 2.7, meaning that the cumulative response to a shock--such as tougher or looser enforcemen--that induces a variation in reported income is almost three times the average individual response.
For more information please contact Dr. M. Anufriev
