Paul L. Caron

Monday, February 8, 2016

Satterthwaite:  Random Tax Audits Cause Participants To Cheat More In The Future

Audit (2016)Emily Ann Satterthwaite (Toronto), The Dynamic Effects of Audits: An Experimental Assessment:

Governments and tax administrators around the world rely on the premise that random audits can be used to reduce tax evasion through two channels of deterrence: first, the ex ante indirect threat of audit will induce taxpayers to truthfully report their income and, second, that the lived experience of audit will deter audited taxpayers from cheating in the future (“direct deterrence,” Alm et al. 2009). This paper provides original experimental evidence of the failure of the direct deterrence channel. Contrary to the predictions of the standard economic model of tax compliance, I find that random audits are ineffective in fostering higher post-audit levels of compliance. Surprisingly, they induce experiment participants to cheat more in the tax periods after the audit (the “bomb crater” effect).

These findings contribute to the emerging dynamic audit literature as well as an older literature on endogenous audits by studying a new — and much more realistic — type of audit rule in the U.S. taxpaying context. Using a novel dataset (data collected from 198 U.S. resident participants on Amazon’s Mechanical Turk task completion portal), I test and compare random audits to an audit rule under which taxpayers are re-audited in the succeeding period if an audit finds them to be non-compliant. I find results consistent with a large bomb crater effect. Participants’ average compliance dropped significantly (8 percentage points) following a random audit. However, after an endogenous audit in which a participant’s detected evasion was known to “flag” her for one or more future audits, compliance rates increased significantly (3 percentage points). Overall, average compliance in the endogenous audit condition systematically and significantly dominated average compliance in the random audit condition. My results suggest that framing audits as “random” may not serve tax administrators’ intended purposes. Further, they shed new light on the importance of communicating about the nature of audit selection with the general taxpayer population as well as with those who have experienced an audit.

IRS News, Scholarship, Tax | Permalink