IRS Audited Black Taxpayers More Frequently Reportedly Due to Algorithm

A study by economists from Stanford University, University of Michigan, University of Chicago and the Treasury Department analyzed 148 million tax returns and 780,000 audits primarily from 2014. The research found that Black taxpayers are audited at rates between 2.9 and 4.7 times higher than non-Black taxpayers, even when accounting for differences in return types. The disparity is not due to individual agent bias, as agents do not know taxpayer race, but rather stems from computer algorithms the IRS uses for audit selection. The algorithms disproportionately flag returns with potential errors in claiming certain tax credits like the Earned Income Tax Credit (EITC), which low-income workers are more likely to claim. Black taxpayers account for 21% of EITC claims but 43% of EITC audits. The researchers used taxpayer names and neighborhood demographics to predict race, validating their approach with North Carolina voter registration data. The study found that an audit approach focused on maximizing detection of underreported taxes would eliminate the racial disparity, but the current system appears to prioritize easier-to-conduct audits over higher-value ones. The IRS does not disclose its audit selection methodology, but the researchers identified that correspondence audits of EITC claimants are preferred because they cost only $29 per audit compared to $385 for business income audits.