Audit Uncertainty, Bayesian Updating, and Tax Evasion

We extend the standard, one-period model of tax evasion to an inter-temporal framework in which an expected-utility-maximizing taxpayer updates expectations about the probability of a future audit based on past audit experience. This framework provides a theoretical grounding for the empirical evidence indicating that tax evasion is affected by taxpayers' perceptions of audit probabilities and is influenced by taxpayers' prior audit experience. We show that for a variety of risk preferences, Bayesian updating increases present and expected future tax evasion and reduces tax payments, inclusive of expected fines. These findings call into question the usefulness of Internal Revenue Service secrecy about audit probabilities for raising taxpayer compliance and expected tax revenue.