Comparative Returns on IRS Audits by Income Groups

Featured in print Digest

This figure is line graph titled, Audit Costs and revenue Generated. The y-axis is labeled, dollars per audit and ranges from 0 to 90 thousand, increasing in increments of ten thousand.  The x-axis is different groups of income distribution, increasing in increments of 10 until it reaches the 90th percentile. The groupings then range from 90 to 95, 95 to 99, 99 to 99.9 and greater than 99.9 There are two lines, one representing revenue and the other cost. Both lines follow similar paths as revenue and cost hover around 5 thousand from the 0 to 70th percentile groupings. Cost then slowly increases until it reaches about 15 thousand in cost for the 99.9 or greater percentile grouping. Revenue increases at a slightly greater rate as it reaches close to 20 thousand for the 95 to 99 percentile grouping but then experiences two major spikes, reaching over 90 thousand for the 99.9 or greater grouping.  The source line reads, Source: Researchers’ calculations using data from the IRS.

The US Internal Revenue Service (IRS) estimates that in the middle of the last decade, underreporting of income on individual income tax returns accounted for roughly $300 billion in unpaid taxes. Additionally, research suggests that more than half of this total is associated with taxpayers in the top 10 percent of the income distribution. Audits of tax returns are an important tool for collecting unpaid taxes. Audits of high-income taxpayers may have greater potential revenue yield, but they are also more resource-intensive and incur higher labor and overhead costs.

William C. BoningNathaniel HendrenBen Sprung-Keyser, and Ellen Stuart, in A Welfare Analysis of Tax Audits across the Income Distribution (NBER Working Paper 31376), examine the results of in-person individual income tax audits between 2010 and 2014. They also utilize internal accounting data from the IRS that enable them to compute the cost of performing each audit. For each audit, they calculate both the upfront revenue associated with the audit and also the deterrence effect that the audit has on future tax revenue paid by the person who is audited. The researchers find that the average upfront revenue per audit was $14,283, compared with an average cost, including exams, appeals, and collections, of $6,418. The average revenue yield was therefore $2.17 per dollar spent on audit resources.

Audits of taxpayers in the top 0.1 percent of the income distribution returned more than $6 in revenue for every dollar spent in audit resources, much more than audits of lower-income taxpayers.

This upfront return to audits varies significantly across the income distribution. Audits of taxpayers in the top 0.1 percent of the “total positive income” (TPI) distribution — the distribution of the sum of positive components of IRS-defined adjusted gross income — produced nearly triple that upfront return, yielding $6.29 for the IRS per audit dollar spent. The average cost per audit was higher for these audits than for the average audit, but the average revenue yield was greater by an even larger margin. The researchers also find that audits of tax returns in the bottom half of the TPI distribution generated an average revenue yield of $0.96 per dollar spent, or less than breakeven.

The researchers also explore whether an expansion of IRS audits would yield similar returns. To do so, they study the 40 percent decline in the overall audit rates of tax returns between the 2010 and 2014 tax filing years. By comparing the returns to audits of 2014 returns to those of 2010 returns, the researchers argue they can infer whether there is evidence of “decreasing returns” to additional audits. In practice, they find that the upfront returns to audits in 2014 are similar to returns to audits in 2010. They also note that some of the overhead costs would likely not increase if the IRS were to expand audits. This suggests the IRS could likely obtain these high returns if they were to return to the 2010 audit rates.

Finally, the authors explore how being audited can deter taxpayer noncompliance in the years following an audit. In a random sample of audits conducted between 2006 and 2014, they find that being audited leads to a persistent increase in future taxes paid. Summing over the 14 subsequent years they observe in their data, they show that this future deterrence revenue is three times as large as the upfront revenue collected during the audit. They find the same across taxpayers’ positions in the income distribution. Combining the revenue gains from the initial audit with those from future deterrence raises the ratio of audit return relative to cost. For a taxpayer with TPI between the 90th and 99th percentiles, the deterrence-inclusive return is in excess of $12 per dollar spent on auditing. In contrast, they estimate a return of $5 for audits of taxpayers with below-median income.

— Leonardo Vasquez