Abstract: The canonical model of Allingham and Sandmo (1972) predicts that firms evade taxes by optimally trading off between the costs and the benefits of evasion. However, there is still no consensus about whether real-world firms react to audits in this way. We conducted a large scale field experiment in collaboration with Uruguay’s tax authority to shed light on
these issues. We sent letters with information about audits to small and medium firms, and measured the effect of these signals on their subsequent perceptions about the auditing process and on the actual taxes they paid. We first confirm a standard finding in the literature: providing information about enforcement increases tax compliance. The change in tax compliance, however, is not consistent with the predictions of Allingham and Sandmo (1972): it is not a function of non-deceptive changes in the information provided about audit probabil-
ities and penalty rates, even though the information significantly affected firms’ perceptions. Audits may deter taxpayers from evading taxes in the same way that scarecrows deter birds from eating crops, which is consistent with models of risk-as-feelings.