October 18, 2021
October 18, 2021
As the 2020 Decennial Census begins to release public data, the question of how the U.S. Census Department defines race and ethnicity is drawing attention. In particular, there is growing concern that as the “Some Other Race” category becomes increasingly popular, it is obscuring important differences in the racial identities of Americans. In the 2020 release, the number of census respondents identifying as “Some Other Race” outpaced “Black or African American,” becoming the second largest category of race, second to “White.” Race is a social construct, so broadening race and ethnicity categories and evolving sentiment about racial identity was bound to lead to this outcome. However, the marked changes are leading to a growing sense that the Census categories for race and ethnicity no longer reflect the demographic make-up of the United States, and a failure to address these data quality concerns could have significant implications for racial equity in government and policy.
Race and ethnicity data from the Census Bureau are used for a host of government decision-making, including redistricting, calculating funding formulas, monitoring compliance, and setting policy goals. Data from the 2020 Census will be used to distribute $1.5 trillion in federal spending each year and is critical for evaluating policy successes and failures. Given the IRS does not collect data on tax filers’ race and ethnicity, census data are one of the only tools tax policy analysts have to understand the racial impact of our nation’s tax laws. ITEP has used it to build, for example, a tax microsimulation model capable of disaggregating the incidence of certain kinds of taxes by race and ethnicity.
The Census has changed the way it asks questions in the past and can choose to do so again in the future. As the Biden administration makes data a central part of its plan to achieve greater racial equity, it has an opportunity to implement research-backed changes that will improve our understanding of race and ethnicity in the United States, and in turn, our ability to draw meaningful conclusions about how our tax laws impact tax filers of different races.
In the most recent Census release, nearly 50 million respondents identified as Some Other Race, either alone or in combination with other race categories, an increase of 129 percent since 2010. Roughly 28 million of these respondents chose it as their only race. Research shows that the majority of respondents who identify as Some Other Race are Hispanic or Latino.
In part, this is because the current census survey design surrounding race and ethnicity is guided by an Office of Management and Budget standard set in 1997, which requires a distinction to be made between race categories and ethnicity categories. This presents a challenge for survey respondents who either do not understand the distinction between race and their ethnicity, or, more likely, who view their race and ethnicity as one and the same.
More broadly, there are longstanding concerns that the OMB race categories are too broad and should be further disaggregated. Race categories like “Black or African American” and “Asian” apply the same label to groups of people whose identities, historical contexts, and lived experience may be very different. This loss of granularity is compounded by the fact that racial and ethnic groups outside of non-Hispanic whites have historically been undercounted– a problem exacerbated by disruptions due to COVID and litigation surrounding the 2020 Census. This loss of detailed and accurate data has important racial equity implications.
In the tax policy world, efforts to understand how our tax code impacts racial equity rely heavily on this census race data. The IRS, which collects administrative data on taxes, does not collect any race or ethnicity data on U.S. taxpayers. Researchers interested in questions surrounding the racial impacts of tax policy must instead use census data to fill in these gaps.
This is important because, while the IRS does not collect information on race, racial inequities are baked into the U.S. tax system. We know, for example, that favorable tax treatment of investment income over income from work disproportionately favors white taxpayers, who have benefitted from historical policies that helped build white household wealth and the white middle class while systematically excluding or subjugating Black, Hispanic, and Indigenous people.
Detailed data on race and ethnicity can provide empirical evidence that the tax system has racialized impacts, that it has contributed to a breadth of socioeconomic disparities, and that it can be a powerful tool in narrowing some of these divides. But data that obscures race makes parsing racial effects of tax policy, and evaluating potential solutions, that much more challenging.
Data Collection is a Policy Choice
How the government chooses to collect and report data is not set in stone, and the census is constantly evolving to capture the changing social, demographic, and economic landscape of the United States. Race and ethnicity in particular are defined by the social and political context of a given time and have undergone many iterations since the first census in 1790. It wasn’t until 1960 that respondents could self-identify their race at all—prior to that, race was determined by census-takers.
From a social perspective, the new Census categories are a challenge. The United States has a four-century history that began with creating racial distinctions and it continues the racist practice of exclusion, discrimination and social hierarchy to this day. We certainly should aspire to dismantle remaining vestiges of this social construct. But we must do so in a way that doesn’t diminish the valuable policy insights we can glean from race-based data.
As early as 2008, the Census Bureau recognized that improvements to race and ethnicity reporting, particularly for Hispanic and Latino Americans, was an area that needed methodological attention. After years of research, the 2015 National Content Test recommended combining the race and ethnicity questions to better capture these respondents. Empirical tests of a combined race/ethnicity question found that 70 percent of Hispanic or Latino respondents chose Hispanic or Latino as their only racial identity, and the number of people checking “Some Other Race” dropped to less than one percentage point. In other words, a combined race and ethnicity question is more reflective of how respondents identify themselves and offers richer racial/ethnic data for Hispanic and Latino respondents in particular.
The 2015 National Content Test recommended other changes as well, including the addition of a Middle Eastern and North African (MENA) category. This, too, reduced the number of respondents who had previously answered “Some Other Race,” and offers greater insight into how people identify their race and ethnicities.
While the Trump administration chose not to implement these changes prior to the 2020 Census, the Biden administration can ensure that future household surveys adopt these recommendations for collecting more complete data on race in the United States. The Biden administration has already acknowledged that collecting data is a first step toward achieving greater equity and has made clear its commitment to identifying data gaps in the federal government. Ultimately, the decision to improve race and ethnicity reporting is a political choice with real policy consequences.
As the concepts of race and ethnicity continue to evolve, and the U.S. population becomes increasingly diverse, the future of racial equity depends on data that accurately captures the nuances of race in the United States. With richer data on race, researchers and public officials will be better positioned to understand the nature and magnitude of racial disparities, to target government efforts to eliminate them, and to ensure accurate monitoring and accountability.