Equal opportunity is the bedrock of American democracy, and our country’s diversity is one of our greatest strengths. However, in recognition of the fact that many Americans still face systemic barriers to equal opportunity and full participation in American life because of their race, on his first day in office, President Biden committed to an ambitious racial equity agenda, which included signing two Executive Orders and catalyzing a whole-of-government effort to redress disparities and inequities still faced by underserved communities. This Issue Brief summarizes social science research on the prevalence of discrimination by race in a variety of economic domains. It also highlights the role of racial bias, and presents some new analysis on the relationship between racial discrimination and racial bias.
Equal opportunity is the bedrock of American democracy, and our diversity is one of our country’s greatest strengths. The landmark Civil Rights Act of 1964, enacted 60 years ago, aimed to advance this equality in multiple domains. It outlawed segregation in public places—including courthouses, parks, restaurants, theaters, sports arenas, and hotels. The Act also outlawed employer discrimination on the basis of race, color, religion, sex, and national origin, and created the Equal Employment Opportunity Commission to enforce the law’s provisions.
Sixty years after the Civil Rights Act of 1964, the Biden Administration is working to break down barriers to equal opportunity. On his first day in office, President Biden signed Executive Order 13985, Advancing Racial Equity and Support for Underserved Communities Through the Federal Government and on February 16, 2023, the President signed Executive Order 14091 to further advance racial equity. These Executive Orders aim to address the racial discrimination that persists in America today. This Issue Brief focuses on this present-day discrimination.
Discrimination not only affects its direct targets, but it also hurts the entire economy by hindering a significant portion of the population from realizing their full economic potential. For instance, Cook (2014) finds that violence against Black Americans from 1870 to 1940 resulted in over a thousand lost patents and that “the economic impact of that decline was equivalent to the GDP of a medium-sized European country at the time.” Moreover, Hsieh et. al. (2019) found that reducing the misallocation of Black talent since the 1950s (e.g., Black Americans who became shopkeepers instead of doctors due to racial barriers) increased per-capita GDP growth by about 4.9%—the equivalent of approximately $1 trillion in today’s dollars, adjusted for inflation. This is consistent with one accounting exercise that estimates that gaps in economic opportunity between Black, Hispanic, and other racial minority groups and those of non-Hispanic white individuals cost the U.S. economy trillions of dollars per year. While it is difficult to pin down a precise dollar amount, the evidence makes clear that eliminating discrimination is not just a matter of fairness but also about economic efficiency and the prosperity of all Americans.
Since the 1960s, certain measures of explicit bias and racism have significantly decreased. For instance, in the 1960s, a majority of White Americans believed they had the right to exclude Blacks from their neighborhoods and opposed interracial marriage. Today, these beliefs are held by a small minority. However, both explicit and implicit biases (that is, thoughts or impulses that can manifest as prejudiced action or inaction without biased intent) still exist in subtle but pervasive and consequential ways. Indeed, as discussed below, social science research confirms that racial discrimination is present in various sectors, including banking, housing, and employment.
While racial differences do not always imply discrimination, social scientists have found significant racial disparities in key areas, such as access to capital, employment, and housing, and in settings where the only difference between individuals was race, or in settings where all other factors were irrelevant—indicating that differential treatment because of race is still a reality (as discussed below).
This document describes evidence of differential treatment of renters, workers, business owners, and everyday people that can only reasonably be attributed to race. This differential treatment by race has effects from childhood through adulthood, and limits racial minorities’ access to neighborhoods that promote upward economic mobility, access to employment with good pay, and access to loans which facilitate business ownership and wealth creation. The research paints a clear picture of decidedly unequal access to opportunity because of race. While much of the existing research focuses on Black Americans, we also highlight examples of discrimination faced by other racial and ethnic groups when evidence is available. Moreover, this document highlights how this differential treatment is directly related to measures of implicit or explicit racial bias developed by social scientists. The patterns underscore that that differential treatment is important, economically meaningful, and pervasive even today, and highlight that racial discrimination still exists, even in settings when the official rules are race-neutral on their face.
As used in this document, racial discrimination refers to different treatment of individuals or groups because of their race, whether the result of explicit or implicit bias. One method for demonstrating racial discrimination is to establish that (1) there are gaps (i.e., differences in outcomes) by race for some outcome, and (2) show that these gaps cannot be explained by reasons other than race.
While showing gaps by race is straightforward, showing that gaps are caused by racial bias is more difficult. For example, because of changing demography in the United States, the average White American adult is 50.3 years old, while the average non-White American adult is 39.9 years old. Also, older individuals tend to have more years of experience and therefore earn higher wages (Census, 2013). As such, White Americans may earn higher wages than non-White Americans, on average, partly due to differences in experience rather than due to race per se. Even after adjusting for differences in age, one may also need to account for other economic, geographic, or demographic factors to assess whether the observed racial wage gap is due to racial bias. Indeed, in cases involving allegations of discrimination, courts routinely examine whether a policy or outcome resulting in disparities between racial groups can be explained on grounds other than race (e.g., Village of Arlington Heights v. Metropolitan Housing Development Corp.)
While not all racial gaps imply racial bias, there are some gaps that clearly do. That is, social scientists have gotten around the challenge on accounting for other factors by relying on settings where there are no differences across groups other than race, or settings where any other factors would have no influence on outcomes. In such settings, the only reasonable explanation for racial gaps is discrimination. We outline some of this evidence – showing clear evidence of discrimination against Black, Hispanic, and Asian Americans across several key domains that limit their economic prospects from childhood through adulthood.
One of the early determinants of economic well-being is one’s neighborhood during childhood. Notably, the neighborhoods where children grow up significantly influence their long-term educational and economic outcomes (Chetty and Hendren 2018; Chyn and Katz 2021). These neighborhoods determine children’s access to clean water which is free from pollutants (Huynh et al., 2024), high performing schools, and availability of job networks (Fee 2020). The difference between growing up in a high-mobility neighborhood versus a low-mobility one has a strong causal effect on children’s long-run outcomes. Chetty et al. (2018) find that, even when comparing outcomes of siblings from the same low-income families, growing up in a high- versus a low-mobility neighborhood can increase adult earning by over 30 percent. As such, any discrimination that limits racial minorities’ access to high-mobility neighborhoods limits minority children’s prospects for upward mobility—that is, unequal opportunity because of race.
There are large racial gaps in neighborhood traits and amenities. Due to redlining—defined as the systematic denial of mortgages and other financial services because of the racial or ethnic characteristics of the residents of the neighborhood in which the property is located—many people of color still live in neighborhoods they were previously consigned to, which are in close proximity to oil and gas wells. Also, racial minorities are much more likely to be exposed to air pollution than White Americans, more likely to live within a mile of a hazardous waste site, and less likely to have access to lead-free drinking water. In rural areas, they are also less likely to be served by a hospital. There is compelling evidence that racial bias in access to neighborhoods plays a key role in where people live—impacting outcomes for multiple generations. Consistent with this notion, Chetty et al. (2020) isolates the causal effect of neighborhoods on upward mobility (defined as growing up in a median income household and then having adult earnings in the top quintile) and finds that predominantly White neighborhoods are three times more upwardly mobile than predominantly Black neighborhoods. While past discrimination is partially responsible for these gaps, here we detail evidence that current discrimination plays a role.
In 2022, there were a total of over 33,000 fair housing complaints across the country and there are very recent high profile examples of sellers who do not want to sell to Black families. Using a more quantitative approach, social scientists test for racial bias in housing using correspondence studies in which researchers distribute identical fictitious responses to rental ads, which by design, differ only in the implied racial identity of the applicant. Absent any bias, an equal share of applicants in all groups should be responded to, and the time to response would be the same. Since differences in response times or gaps in callback rates cannot be ascribed to application differences, gaps must therefore indicate differential treatment, or bias.
Using this approach, several studies have found that emails from non-White applicants have slower response times and are less likely to receive a response at all. Carpusor and Loges (2006) randomly assigned White, African American, and Arab names to emails and found that “African American and Arab names received significantly fewer positive responses than White names, and African American names fared worst of all.” Similar results are found for African American names by Hanson and Hawley (2011).[1] Similar to correspondence studies are audit studies or undercover tester studies that send actual applicants (trained actors) with near identical profiles from different groups. Using undercover testers of different races with similar financial profiles, a 2019 audit study from Newsday analyzed over 5,700 house listings in Long Island NY (a high-opportunity and highly segregated suburban neighborhood) and found discrimination against Asians, Hispanics, and Blacks: 19, 39, and 49 percent of the time, respectively. Discrimination captured various forms of disparate treatment by agents, including refusing to provide house listings or home tours to minority testers unless they met financial qualifications that weren’t imposed on their White counterparts, directing Whites and minorities into differing communities (to match the demographics of the neighborhood), and showing White testers more listings than other testers. Other audit studies comparing equally qualified Asian Americans and Pacific Islanders (AAPIs) and White Americans in rental and sales markets. AAPIs were told about 9.8 percent fewer available rental properties and were shown 6.6 percent fewer units than their White counterparts. The effects were even larger in the sales market, where AAPIs were told about 15.5 percent fewer available properties and were shown 18.8 percent fewer properties.
These behaviors have real effects on where people live. Recent studies find large Black-White response gaps and show how this behavior blocks Black families out of particular neighborhoods. One 2023 study finds that these gaps are larger in places with high-quality schools and few Black residents – reinforcing racial segregation and depriving Black renters access to high quality schools. Similarly, another recent study analyzed over 25,000 email interactions with landlords across the 50 largest US cities and found that African American and Hispanic renters often face discrimination and that the extent varies by region. The study also matches evidence on actual rental outcomes to show that this discrimination was likely a driver of increased segregation and intergenerational income gaps—underscoring that this racial bias leads to racial gaps in access to high-mobility neighborhoods with amenities that promote positive outcomes.
One paper uses these same data to estimate the welfare loss due to having limited choices. First, the paper documents that discrimination limits applicants’ access to neighborhoods with higher rent, better schools, and lower exposure to toxins. The authors then estimate a model of willingness to pay for these amenities and conclude that discrimination resulted in “lost choices that these groups would be willing to pay significant sums to avoid.” They conclude that discrimination—by limiting housing options—imposes damages between 3.5% and 4.4% of annual income for renters of color on average, and as high as 7% of income at income for African Americans making above $100,000 per year.” The authors note that this accounting of damages may be incomplete, and the approach may not include the potential intergenerational effects. However, the key takeaway is that the welfare implications are sizable. More generally, by impacting where families are able to live, these discriminatory actions impact access to local amenities, including access to schools, jobs, quality health care, a toxin-free environment, and transportation. Chetty et al. (2018) finds that some neighborhoods have a profound causal effect on the likelihood that a child from a low-income family is able to be a high earner as an adult: housing discrimination blocks minority children form taking advantage of these neighborhood-based opportunities. These factors impact the health and well-being of adults (Chyn and Katz 2021) and impact the short- and long-run outcomes of their children.
While neighborhoods shape economic outcomes from an early age, a key determinant of economic well-being in adulthood is having a stable job with good pay. Thus, racial bias in employment has far-reaching implications for the well-being of those harmed. Based on the observed gaps, a typical Black worker in 2023 made about 12 percent less and was 2 percentage points less likely to be employed than a typical White worker of the same age, gender, education, and living in the same Census region. There are similar but somewhat smaller gaps for Hispanic workers.[2] As detailed below, research shows that these gaps for Black and Hispanic workers are, in part,due to discrimination.
As in housing, researchers test for racial bias in hiring using correspondence studies, in which researchers distribute identical fictitious resumes or job applications which, by design, differ only in the implied racial identity of the applicant. In one study published in 2004, resumes were sent out in response to help-wanted ads in Chicago and Boston newspapers. The authors randomly assigned some otherwise identical resumes with very White sounding names (such as Emily Walsh or Greg Baker) and others with very African American sounding names (such as Lakisha Washington or Jamal Jones). They find that the White sounding names “receive 50 percent more callbacks for interviews.” More recently, Kline et al. (2022) conducted a similar experiment with over 83,000 job applications sent to large Fortune 500 firms and find that callback rates were 9 percent less for the Black-sounding names, with much larger gaps from some specific employers. This basic result has been replicated by several researchers in a variety of settings, consistently finding a callback gap between 9 and 50 percent.[3] Looking systematically and over time, a recent meta-analysis of correspondence and audit studies (which send actors of different races to apply to jobs with identical qualifications) that also include Latino-Americans, finds an average Black-White callback gap of 36 percent and an average Latino-White callback gap of 24 percent. Looking at trends over time they find no change in the level of discrimination against African Americans since 1989, although they do find some suggestive evidence of declining discrimination against Latinos.
Looking beyond the callback stage, a recent summary of multiple audit studies found considerable additional discrimination. In their data, majority applicants with very similar qualifications received 53% more callbacks than comparable minority (e.g., Black, Hispanic, Middle Eastern) applicants, and 145% more job offers than comparable minority applicants – indicating that the gaps in callbacks understate gaps in the likelihood of receiving a job offer. The fact that minority applicants are less than half as likely to be offered a job than similarly qualified White applicants mechanically leads to elevated racial gaps in unemployment rates. It also reduces the bargaining power of Black and Hispanic workers relative to White workers, resulting in lower relative pay. The evidence clearly conveys that discrimination exists, and plays a key role in generating these racial gaps.
Wealth is arguably a more robust measure of long-run economic well-being than income. Access to credit and capital markets is important for business creation and building wealth. A large literature studying the determinants of entrepreneurship finds strong evidence that insufficient capital and access to credit markets serve as one of the main barriers to entrepreneurship. For example, sudden increases in wealth, through bequests or lottery winnings, and increased access to bank financing through financial deregulation, increase entry into entrepreneurship. As such, differential access to capital by race can lead to racial gaps in business ownership in addition to other measures of wealth.
A 2022 report from the U.S. Congress Joint Economic Committee found that “Black and Hispanic households are more likely than White households to be denied or not receive as much credit as requested when applying.” While racial gaps in business ownership and homeownership (both strong predictors of wealth) have narrowed in the past three years under President Biden due to a rise in minority-owned business formation, gaps remain. Looking at raw gaps in 2022, only about 6 percent of Black and Hispanic households owned a business—compared to about 9 percent of White households. Homeownership rates in 2022 were 46 percent for Black households, 51 percent for Hispanic household, and 63 for Asian households – all below the 73 percent for White households. Also, the Black-White wealth gap has been between a massive 80-85 percent in recent decades (Derenoncourt et al., 2024). As we summarize below, evidence indicates that present day discrimination in credit access plays a role for Black, Asian, and Hispanic borrowers.
Using the audit study approach and matching similar applicants of different races, one study conducted in 2017 finds that, compared to White business loan applicants, Black business loan applicants were asked to provide more financial and personal information, including marital status, which can be a violation of fair lending law. In addition, Black applicants were less likely to be offered to complete an application, offered a future appointment, or provided help in completing a loan application. These results have been replicated across different studies.
Recent research based on the Paycheck Protection Program (PPP) program also found strong and compelling evidence of private-sector lending discrimination between 2020 and 2021. In an effort to support small business jobs, private lenders administered PPP loans that were federally guaranteed. The federal guarantee essentially eliminated a lender’s risk because the government repaid lenders if a business defaulted. Even so, one study showed that Black business owners were less likely to receive PPP loans compared to White business owners with similar application profiles (including similar credit histories, education, age, and business profile).
Finally, one 2023 study leverages a policy change to isolate evidence of discrimination by race in PPP lending. First, the study documents (as others do), that minority borrowers in general and Black-owned businesses in particular were less likely to receive PPP loans from small and mid-sized banks in settings where subjectivity (and thus bias) was most likely to influence lending decisions. Again, these racial gaps were found despite the fact that there was a federal guarantee largely eliminating risk to the lender. However, the study also found that these racial gaps (for Black, Asian, and Hispanic business owners relative to White business owners) shrank when processing procedures such as income and payroll verification were computer-automated rather than conducted by a bank employee. That is, when there was less scope for human decision-making, and all applications were treated equally irrespective of the race of the applicant, racial gaps in approved loans decreased.
The previous section presented much evidence on racial gaps in situations where the only possible explanation was race itself. The compelling evidence of real-world racial disparities and discrimination has led many researches to explore and identify potential causes tied to underlying racial bias, either explicit or implicit. Social scientists often explore this possibility by examining whether racial disparities are more pronounced in regions with higher levels of racial bias. The theory that areas with stronger prejudice would exhibit larger racial disparities was proposed by Becker (1957) and is supported by empirical data.
We use measures of implicit and explicit bias that are used by researchers. Researches have employed varying measures of racial bias, including public opinion questions such as “Would you object to sending your kids to a school that had [a certain fraction of] Black students?” or “Do you agree that White people have the right to keep Black people out of their neighborhoods?” Other measures of bias include measures of local housing segregation, the number of hate crimes against various groups, racially-biased Google searches, and implicit and explicit bias tests. For purposes of the analysis here, we use two of these established measures of bias. One is a measure of implicit bias that comes from Implicit Association Test (IAT) scores (Greenwald, Nosek, and Banaji, 2003). This measures subconscious biases that individuals have towards various groups of people. It is based on the speed at which respondents are able to associate positive words and concepts with Black versus White faces or the word “Black” versus “White.” We take geographic averages of this measure. Our second measure is a measure of explicit bias based on geographic variation in the prevalence of racially charged web searches (Stephens-Davidowitz 2014).
Previous research shows that places with higher levels of these two measures of prejudice have larger racial wage gaps, racial mortality gaps, and fewer loans for Black-owned businesses. In what follows, we show that these measures predict geographic variation in the prevalence of discrimination.[4] Given the much larger research base for Black-White gaps, much (but not all) of this analysis is on Black-White gaps.
Looking at discrimination in housing, Figure 1 plots rankings of callback gaps (the difference between the callback rate for minority and non-minority applicants) for housing applications found in Christensen et al. (2021). The paper considers 50 large Core Based Statistical Areas (CBSAs) and we order these places from largest to the smallest callback gaps. Lower numbers indicate larger racial callback gaps, with higher numbers indicating smaller gaps. The authors create callback gaps separately for Black applicants and Hispanic applicants. For this analysis, the CEA pulled information from their paper and linked each CBSA to the local measure of implicit and explicit bias described above, so that each dot in Figure 1 reflects a CBSA. The left panel in Figure 1 shows that going from an area with low explicit bias (10th percentile) to one with high bias (the 90th percentile) is associated with a CBSA being ranked about 18 places worse (i.e., a lower number) in terms of Black callback rates. The right panel in Figure 1 shows that going from an area with low to high implicit bias is associated with a CBSA being ranked about 14 places worse in terms of Hispanic callback rates. Put differently, going from a place with low to high explicit bias would reduce the callback rates of Blacks by about 14 percent relative to Whites, and going from a place with low to high bias implicit bias would reduce the callback rates of Hispanics by 7 percent relative to Whites. These patterns are clear: landlords are much more likely to treat inquiries for a rental differently by race in locations that exhibit a greater prevalence of both implicit and explicit racial bias.
Figure 2 plots the racial callback gap for job applications from Kline, Rose, and Walters (2022), using the two measures of bias. Job applications are pooled by geography so that each dot reflects a Designated Media Area (DMA). Figure 2 plots the estimated job callback gaps against DMA average levels of racial bias. Importantly, measures of both explicit and implicit bias are associated with larger Black-White job callback gaps. The left panel in Figure 2 shows that going from a low to high explicit bias area (as defined above) is associated with about a 1 percentage point increase in the racial callback gap (or about a 11 percent increase from a baseline gap of 9 percent); the right panel shows that going from an area with low to high implicit bias leads to a similar increase in the gap. Put differently, going from the least to the most biased DMAs (6 standard deviations apart) is associated with a 2-percentage point difference in the callback rate, representing a 23 percent increase in the callback gap. What this means is that if Black and White applicants submitted identical applications, there would be 23 percent more callbacks among the White applications than the Black application in high versus low bias areas. That is, in places where people tend to exhibit bias against Black Americans (either in web searches or in how long it takes to associate a Black face with a positive word), employers are more likely to treat otherwise identical resumes with Black and White sounding names differently.
Looking at business access to financing in Figure 3, the CEA used information in Tables 3 and 5 from Howell et al. (2023) to understand how the same two measures of implicit and explicit bias impact being approved for a business loan. Recall that there were racial gaps in PPP lending even though they were federally guaranteed, thus largely eliminating credit risk as a factor in explaining differential lending by race. Also recall that these authors study the change in lending patterns by race after certain banks automated evaluation of loan applications.
The left panel in Figure 3 shows the percent change in PPP loans going to Asian, Black, Hispanic, and White business owners, after automating parts of the loan evaluation process. When human discretion was reduced, the racial gaps in lending fell. That is, the fraction of PPP loans going to Asian, Black, and Hispanic applicants rose and the fraction going to White applicants fell. Most notably, lending to Black business owners almost doubled.
The difference between the loan approval rate before and after automation across different groups is a measure of differential treatment by race due to human judgment. If these differences reflect bias-driven discrimination, one should expect that the effect of automation of payroll and income criteria would be largest in areas with more bias. This is precisely what the authors document. The right panel in Figure 3 shows how automation increased the fraction going to Black business owners overall and compared to areas with high implicit and explicit bias. Indeed, this panel shows that when banks automated processing of loan applications, Black businesses received 93 percent more loans in places with the least amount of bias, 134 percent more in places with high implicit bias, and 120 percent more in areas with high explicit bias.[5] In sum, as with employment and access to housing, racial disparities that can only be reasonably attributed to differential treatment because of race are most pronounced in geographic locations with greater levels of measured explicit and implicit racial bias.
The evidence is clear that certain minority children are less likely to live in neighborhoods that have good schools, clean drinking water and clean air, and positive causal effects on upward mobility. As adults, many Americans from minority groups are less likely be called back for a job and attain good-paying employment, and less likely to gain access to much-needed capital to start a business or launch a new career than an otherwise identical White person. Furthermore, while not all evidence is experimental, there is compelling evidence of discrimination in many other important settings.
Ethnic minority home owners pay higher taxes than White homeowners in the same cities with similar homes, and they receive lower bids when selling items in online markets or selling their home. Black and Native American children are also more likely to receive substantiation and out-of-home placement decisions made by Child Protective Services, and Black children receive less attention from their teachers and instructors in online learning settings. Moreover, even when the teachers are randomly assigned, White teachers tend to rate Black students’ misbehavior more harshly than Black teachers. When stopped for speeding, Black and Hispanic drivers are less likely to have a ticket written for the lower speed threshold than White drivers, and more likely to be stopped when their race is more easily observed. In criminal justice, not only are there substantial sentencing gaps by race, but conditional on defendant traits and charges, research finds that incarceration rates are higher and sentences are longer in jurisdictions with a higher fraction of Black residents; the research concludes that confinement rates would fall by 15 percent if all jurisdictions adopted the sentencing of the most-White jurisdictions. Furthermore, there are large and important racial gaps in health. Even after accounting for socio-economic factors, Black women have a higher risk of maternal mortality than White women, life expectancy is shorter for Black Americans, and there is evidence that differential treatment by race plays a role. This is only a partial list, but the evidence is clear; an individual’s race still matters in myriad ways.
To combat explicit and implicit biases, several companies and organizations have put guardrails in place. For example, having managers voluntarily take diversity training can help reduce racial gaps in hiring. Direct recruiting on colleges can increase the hiring of minorities. Mentoring programs are effective at increasing the diversity of managers. Research has also shown that inviting non-managers to diversity and inclusion workshops can help organizations better identify points of conflict and possible resolutions. Working in teams as equals with co-workers of different ethnicities can lead to more equitable opportunity and promotion. In government, a new HUD policy requires certain lenders to allow borrowers to request a re-assessment of the appraised value of their property if they believe that the appraisal was inaccurate or biased. In sum, there are several ways for organizations to improve opportunity for less represented groups.
While challenges remain, Americans have made considerable progress since the 1960s in reducing many measures of overt racism and racial gaps in income, wealth, and educational attainment (Donohue and Heckman, 1991; Center for Education Policy Analysis, 2017; Wolff, 2022; Kent and Ricketts, 2024). Even so, sizable discrimination from both explicit and implicit bias still exists today. The existence of contemporary discrimination motivates President Biden’s commitment to racial equity. As we show above, these biases are associated with important racial differences in a variety of domains that cannot be explained away by other factors and can only be reasonably attributed to racial bias. These data show that, in many ways, the structures and forces that necessitated civil rights legislation and a broader focus on racial equity still exist, and those same structural barriers continue to necessitate remedial action. Despite progress over the last sixty years, the evidence is clear that race continues to be a significant determinant of economic well-being in the lives of many Americans.
It is important to highlight that racial bias leads to racial gaps precisely in settings where the rules are race-neutral. Discrimination from bias is often prevalent whenever people have discretion, such as in calling back a job applicant or deciding whether to give a break to a speeding driver.[6] In such cases, the rules are race-neutral, but the application of the rules is not. Even if individuals making decisions may not have demonstrated explicit biases, their implicit biases may still impact their decisions. Indeed, in many of the examples above, measures of implicit bias are as predictive (and often more predictive) of discriminatory gaps as measures of explicit bias. The prevalence and effect of implicit bias shows the limitations of relying solely on race-neutral tools to advance equality of opportunity. In various contexts, the research suggests that race-conscious action may be necessary to truly achieve equality of opportunity for all Americans irrespective of race, creed, and color.
[1] Some studies indicate that these biases are most prevalent for Blacks who are not of high social status and for Hispanics who are portrayed as recent immigrants.
[2] Due to sample size constraints, we do not report such analysis for Asian Americans, but raw averages suggests that Asians, on average, earn more than their White counterparts. However, there is considerable heterogeneity within among Asian populations.
[4] We obtained job application microdata from the authors of Kline, Rose, and Walters (2022); we impute housing application estimates from Figure SM3.2 of Christensen et al. (2021); and we use PPP estimates from Tables 3 and 5 from Howell et al. (2023).
[5] Places with more bias defined as two standard deviations more biased.
