How effective is tax enforcement as means of raising government revenue? Does the return on additional tax enforcement vary when auditing high- or low-income taxpayers?

In this paper, we provide new evidence on the costs, revenues, and welfare impacts associated with each audit of individual taxpayers in the U.S. between 2010 and 2014. We do so using comprehensive internal accounting information from the audits division of the IRS combined with audit-level enforcement logs.

Our analysis proceeds in four steps. First, we estimate the average cost of each audit conducted in the U.S. and the revenues each audit raises. Here, we provide new comprehensive estimates of audit costs that include not only the cost of an auditor’s time, but also the associated management, space, IT, and other indirect costs that support enforcement. Second, we use the large recent decrease in audits in the U.S. to study how marginal revenues and costs from expanding audits might differ from average revenues and costs. Third, we study the deterrence effects of audits. We use the IRS’s random audits program to estimate one particular channel of deterrence: the specific impact of being audited on one’s future tax payments. In the final step of our analysis, we use our estimates to assess the efficiency of additional audits of high- and low- income taxpayers and the relative efficiency of raising revenue through additional audits as opposed to increased tax rates. We do this by nesting the results of our first three steps into the Marginal Value of Public Funds (MVPF) framework. Our results suggest that expanding audits of high-income taxpayers raises revenue at lower cost than expanding audits of low-income taxpayers or raising rates on either high- or low-income taxpayers.