written by Connor Zahler:
Introduction
As the old adage goes, two things are unavoidable: death and taxes. Indeed, taxes are part of just about everyone’s lives. Every April (or hopefully before then), millions of Americans sit down and try not to commit felonious fraud (or pay someone to avoid fraud for them) as they report their incomes to the U.S. government on their tax returns. Tax returns are subject to audit for accuracy, but is everyone at the same risk of being caught with an accidental or an intentional error? CBS news reports that wealthy taxpayers, perhaps due to their ability to fight a lawsuit, are incredibly unlikely to ever face an audit. The New York Times reports another dimension of inequity in taxation, race, in the February 2023 article headlined: “Black Americans Are Much More Likely to Face Tax Audits, Study Finds.” Why might that be? Today, we’re going to dive into the New York Times article and the study it is referencing.
The New York Times Article
The New York Times article is built on information from an original study, additional quotes from the study’s authors, and references to previous IRS-related news. After repeating the research study’s claim, the NYT author, Jim Tankersley, elaborates that the racial disparity in audit rates is more suggestive of issues with the algorithms used to select individuals to audit, rather than individual racism on the part of auditors (who do not know the race of their subjects). A process with this kind of inherent bias is the very definition of systemic racism.
Tankersley’s main argument can be summed up by this excerpt, located near the beginning of the article:
“In effect, the researchers suggest that the I.R.S. has focused on audits that are easier to conduct and as a result, finds itself disproportionately auditing a historically disadvantaged group rather than other taxpayers, including high net-worth individuals.”
Tankersley provides additional basic information about the research study. It came from a partnership between the U.S. Department of Treasury and Stanford’s RegLab that used tax filings, census reports, and other state documents to infer the race of tax filers. After attaching race to returns, researchers identified a disparity in audit rates that was initially attributed to more Black Americans claiming the Earned Income Tax Credit. But even among those who claimed the EITC, Black Americans were still more likely to be audited than White Americans.
Notably, the NYT article does not directly link to the research study. It provides enough information to track the study down, but I only found the link from a separate Law360 article. Though the study was not cited, its authors were consulted for the NYT article, and two of them provided quotes. It seems that the research study had not been officially published when the article hit the presses, which likely explains the missing citation.
The Study Itself
The research study, from Stanford’s Institute for Economic Policy Research, is titled “Measuring and Mitigating Racial Disparities in Tax Audits.” Its authors include researchers from Stanford, UChicago, and Michigan’s Evelyn Smith. The main findings of this study seem to be accurately conveyed in the NYT article. Here’s how the research study authors phrase their main point in the discussion section:
“In this paper, we have presented evidence that Black taxpayers are audited at higher rates than non-Black taxpayers, and that this disparity is primarily due to differences in the audit rate between Black and non-Black EITC claimants…we find that the objective of the predictive model underlying audit selection, as well as operational considerations relating to the complexity of audited tax returns, can be critical drivers of disparity.”
Tankersley echoes this finding well in his NYT article and doesn’t make any major extrapolations. While the study includes a lot of background information and minute details of US tax policy, the article still provides the necessary nuances to understand the main thrust of the paper. The researchers also mention the constraints of their work, noting that audit selection is a small slice of tax policy and that they cannot account for all the constraints that the IRS faces. This is one thing not included in the NYT article. In total, though, there are no major differences between the findings of the study and Tankersley’s presentation of them.
One thing the NYT article leaves out is the novelty of the research paper. Previous attempts to analyze racial disparity in tax data has been hamstrung by the fact that taxes do not have associated racial information. The researchers were able to overcome this hurdle by using Bayesian Improved First Name Surname Geocoding, which involved matching census records with tax audit records and generating a probability that a given respondent was Black. General Bayesian Improved Surname Geocoding is “widely applied in academic studies and is recommended when race is missing by the National Academy of Medicine.” The study authors also developed their own algorithm to produce an estimated disparity from their race predictions, although the actual methodology is a bit complex for this article. In any case, this study is a landmark in creative methodology for studying tax policy.
The QMSS Connection
Students may remember geocoding as an essential part of Geographic Information Systems, or GIS. We cover the basic concepts of GIS in QMSS 301, but dedicated courses at the University cover geocoding extensively and even involve students doing some of their own geocoding. The application of GIS to a social problem such as racial disparities in a data-driven system like tax auditing is quintessentially QMSS. This Stanford study is an example of scholars applying innovative methodology to a massive body of data in order to solve a social ill.