When do tax amnesties work?

Abstract

Tax amnesties can raise short-term revenues but reduce long-term compliance by creating expectations of future amnesties and reducing tax morale. This paper uses a laboratory experiment in Bolivia to investigate the impact of a change in the duration of amnesties and other previously unexamined amnesty modalities. The experiment examines the behavior of 338 professional workers in order to compare tax compliance, debt payment, and revenues during and post-amnesty under four different amnesty rules: (1) a one-shot unannounced amnesty (one period in the middle of a 25-period game), (2) an extended amnesty (three periods in the middle of a 25-period game), (3) a change in the audit probability rule after an extended amnesty, and (4) deductions on tax debt for paying in the first or second rounds of the extended amnesty. Increasing amnesty’s duration reduces tax compliance and cannot outperform one-shot amnesty’s total revenues. Extended amnesties with debt deductions increase total tax compliance compared to extended amnesties without debt deductions with no effects on revenues. Effects on compliance come primarily from individuals with previously high compliance rates. Any kind of amnesty mostly unmotivates individuals with low compliance rates, thus impacting amnesties’ capability to raise revenues in the long term.