IRS algorithms in determining tax returns to audit, enforcement policy, and resource constraints result in significant racial disparity in returns chosen for a tax audit. A comprehensive study conducted by economists from Stanford University, the University of Michigan, the University of Chicago, and the Treasury Department resulted in a detailed report that concluded that the IRS is about three times more likely to audit Black taxpayers compared to other taxpayers. The study states, "despite race-blind audit selection, Black taxpayers are audited at 2.9 - 4.7 times the rate of non-Black taxpayers."  

The study found the most likely causes of this racial disparity in tax returns chosen to be audited are: 

IRS algorithms disproportionately flag returns that claim the Earned Income Tax Credit 

The IRS's computer algorithms for its selection process disproportionately flag tax returns that are more likely to have errors, specifically those that claim refundable tax credits like the earned income tax credit (EITC).  

In fact, according to the study, "the main source of the disparity in differing audit rates by race among taxpayers is the earned income tax credit."  

Resource constraints result in less complex returns being audited

Due to IRS resource constraints, resulting from continual budget cuts and the loss of roughly 40 percent of its experienced agents, the agency has focused on auditing less complex returns. Less complicated returns are more likely to go through a correspondence audit because these tax examinations are for a single year, involve no more than a few issues, and can be conducted (and resolved) by mail by reviewing a few relevant documents. 


Comment: 70% of IRS audits are done by mail, and 50% involve earned income tax credit claimants


Compared with labor-intensive field audits – which require an in-person examination at the taxpayers' place of business or their tax professional's office – correspondence audits of returns claiming the EITC are easier to flag, cost very little, and require minimal effort by IRS staff.  

The burden of correspondence audits on EITC claimants is more likely to fall on lower-income individuals whose tax returns are less complex and less likely to lead to challenges, including litigation.

No bias found at the agent level 

It is important to note that the study did not find bias at the individual tax enforcement agent level. IRS agents do not know the race of the individuals they are auditing. 

The IRS can modify selection methods without sacrificing revenue or selection accuracy

To better understand the audit selection bias, the study team modeled the racial impact that various alternative audit selection policies might have. The result demonstrates how the IRS might be able to change its confidential algorithm to reduce its racially disparate impact. 

The authors suggested that "the IRS drill down to understand and modify its existing audit selection methods to mitigate the documented disparity." The authors believe that the IRS can reduce racial audit disparities without necessarily sacrificing tax revenue or the accuracy of the audit selection.  

Why are Black taxpayers more likely to be audited? 

After finding that Black taxpayers were 2.9 - 4.7 times more likely to be audited than non-Black taxpayers, study authors considered possible reasons for that disparity. It was determined that the problem is primarily with the IRS algorithm's use of the Dependent Database, which flags a potential problem and generates an audit letter to the taxpayer.  

The vast majority of the racial disparity involved so-called "correspondence" audits done by mail, rather than more complex, in-person "field" audits. These correspondence audits, by their nature, involve less complicated returns that tend to claim refundable tax credits. 

The team also found that the IRS disproportionately audits people who claim the Earned Income Tax Credit (EITC) – a refundable tax credit targeted to assist low- to moderate-income workers.  

The disproportionately large number of audits focused on the EITC magnified an additional issue – that the largest source of disparity occurs among EITC claimants. While Black taxpayers accounted for 21% of EITC claims, they were the focus of 43% of EITC audits. 

How did the study determine taxpayer race?

Although there have been questions about whether the IRS uses its audit powers equitably, privacy concerns and the confidentiality of the IRS's algorithms in choosing the returns for audit made it difficult to determine. That changed when President Biden signed Racial Justice Executive Order 13985, requiring all federal agencies to assess how their programs impact racial and ethnic equity.  

Applying that order to the IRS tax return audit program, economists at the Treasury Department collaborated with the Stanford RegLab team to analyze over 148 million tax returns and approximately 780,000 audits for tax year 2014 (an overall audit rate of 0.54%). 

Even with all the data they received, the research team could not know the race of the taxpayers in the data set. They created an approach to predict whether a taxpayer identified as black.  

The disparate impact of the IRS audit selection algorithm 

As part of the study, the authors did not have access to the algorithms the IRS uses to select audits. Despite this, they modeled several possible explanations for the racial disparity in audit rates.  

Key study findings 

The authors determined that, even where audit selection processes are largely automated to ensure no intentional discrimination in audit selection, these processes can significantly impact racial disparity in audit selection. There are several reasons for this, most importantly that by designing algorithms that select for underreported refundable credits rather than the total amount of underreported tax, the racial disparity continues.  

If the algorithm focused audit selection on the amount of underreported tax, rather than specific issues like EITC, the racial disparity in audits would be significantly reduced. 

Next steps for the IRS 

The study's authors do not make formal recommendations for making the IRS audit selection algorithm more equitable. Instead, they document the likely effects of alternative policies, which provides the IRS with several potential alternatives for reducing the racial impact of its audit selection system, including:   

  • Predicting and focusing on the magnitude of taxpayers' underreported income rather than just the likelihood of it 
  • Viewing dollars as equal whether they are to be paid in refundable credits or received in taxes 
  • Using IRS resources to audit more complex returns rather than focusing only on the simpler ones that are much less expensive to audit 

Now that the equity implications of how the IRS selects audits are known as a result of this study, the research team suggests the IRS should tweak its confidential audit selection algorithm.  

"Racial disparities in income are well known, and what the IRS chooses to focus on has big implications for whether audits complement, or undercut, a progressive tax system," according to one of the study's authors.  To read more click here.

Comments powered by CComment