Race data on taxpayers can be inferred by matching returns with census zip-code demographics

Evidence

I’d never ask taxpayers to identify themselves by race on their returns—it seems far too likely that individual taxpayers would fall victim to cases of implicit or explicit bias from federal employees, resulting in more audits—and today’s Social Security matching offers less data than it used to, thanks to new procedures that assign a number at birth or upon qualified entry to the country and don’t collect racial data. Fortunately, there are other ways to get this information. Bearer-Friend’s research proposes matching taxpayer data with nontax data sets, and using the results to gain race-based insights into taxation. For example: sort tax returns by zip code and run an analysis based on which returns come from majority-black versus majority-white neighborhoods reported in census data. In neighborhoods where the racial demographics are starkly disproportionate, the IRS could treat those tax returns as from white or black residents with a great deal of confidence that their analysis would be pretty close to accurate. — Dorothy A. Brown, The Whiteness of Wealth, 517