In this article I will argue against the view racial bias in the modern economy negatively impacts black people’s incomes, unemployment rates, educational opportunity, and ability to get a loan.
Differences in Income
Some think racism is necessary to explain group differences in income. But blacks actually make as much or more money than whites when education, cognitive skill, marital status and other confounding variables are controlled for. Such has been shown multiple times (Farkas 1996., Kanazawa 2005)
Similarly, it has been shown that people are willing to pay 3.5% more for a book when they are led to think that the book’s author is black suggesting that when comparing authors of equal talent black authors likely have a higher income (Weinberg et al., 2022).
Of course, racism may explain a group difference in one of these confounding variables, such as educational attainment, but it is important to note that group differences in income disappear or flip direction without directly controlling for any measure of racism.
Differences in Educational Opportunity
With respect to educational opportunity, Murray and Rueben (2008) calculated spending per pupil for US schools between the years 1972 and 2002. They found the following:
“In 1972, the ratio of nonwhite to white spending was .98; this trend had reversed by 1982, as spending per pupil for nonwhite students was slightly higher than for white students in most states and in the United States as a whole and has been for the past 20 years”.
Thus, since 1982, spending on non-white students has been greater than spending on white students. This issue was revisited by Richwine (2011) who found that spending on black students was 1% greater than spending on white students, while spending on Asian and Hispanic students was a few percentage points lower.
On top of all this, within school districts blacker schools receive more money and meaning these between-district analyses understate the degree to which black education is better funded than white education (Ejdemyr et al., 2017).
These difference in spending are reflected in school characteristics: on average, blacker districts have smaller classes, more experienced teachers who have more formal education, and teachers who receive more pay (Cocoran et al., 2003).
This information may surprise some people since media outlets like Vox, NPR, and the NYT, have spread the results of the “EdBuild” report which claimed to show that white districts are better funded but which utilized a dataset based on an unrepresentative subset of American school districts (see their methodology). Moreover, they failed to weight schools by population size and so got the numbers wrong even for the districts they analyzed (Safier, 2019). As we’ve seen, a more representative dataset show their conclusion to be false.
A second point of confusion comes from the fact that some people think school funding just comes from local property taxes and so minorities must have worse funded schools since they live in poorer neighborhoods. This narrative is misleading since only about a third of school funding actually comes from property taxes and minority schools have their funding boosted by federal money in order to make up this inequality (NCES, Cocoran et al., 2003).
Anyway, once high-school is complete, students apply to college. Based on aggregated data from 20 previous studies, we can estimate that when comparing people of equal qualifications, Black applicants are roughly 21 times more likely than white applicants to be admitted, while Hispanics are 3 times as likely, and Asians are 6% less likely or 59% more likely depending on whether we use the mean or median estimat1.
Similarly, it’s been estimated that the proportion of students attending selective colleges who are white would increase from 66% to 75% if admissions were based solely on test scores (Carnevale et al. 2019)
And once in college non-white students are more likely to receive grants and scholarships despite the fact that white students are no more likely to have their parents pay for their school (Kantrowitz, 2011; Brown, 2019)
As a whole then, resource allocation within the education system favors non-whites students over white ones. Obviously then, white racists have not inhibited black economic success by depriving them of the resources needed for educational success.
Difference in Unemployment
Previously, I noted that racial income differences can easily be flipped by controlling for obvious determinants of income. Of course, to have an income at all you first have to be employed and many think companies avoid hiring minorities for racist reasons.
There are obvious reasons to think this is not true. First, it is illegal to racially discriminate in hiring giving firms an inventive not to. Moreover, since the early 1990s it has been true that most US firms with 100 or more employees have some sort of affirmative action policy (Dobbin et al. 2006).
The idea that racism explains group differences in employment levels is also hard to square with the fact that there was no unemployment gap between races in the early 20th century when white people were far more likely to be racially biased against black people (Fairline and Sundstrom, 1999).
So proponents of the view that racism inhibits black people’s ability to get hired are arguing that around the time racial bias in hiring was made illegal, firms decides to secretly begin engaging in just this practice while pretending to go out of their way to hire minorities. On its face, this seems rather implausible.
It’s also worth noting that the unemployment gap between races that emerged in the second half of the 20th century seems to be, at least in part, voluntary.
As Williams (2011) reports: “During 1979-1980, the National Bureau of Economic Research conducted a survey in the ghettos of Boston, Philadelphia, and Chicago. Only a minority of the respondents were employed, yet almost as many said it was easy or fairly easy to get a job as a laborer as said it was difficult or impossible; and 71 percent said it was fairly easy to get a minimum-wage job.”
Many are likely to respond to this sort of argument by noting that black people are less likely than white people to get called back when they submit a job application that is identical in every way other than the race implied by their name.
To be specific, Quillian et al. (2017) meta-analyzed the research on hiring discrimination and found that black applicants received 36% fewer call backs than white applicants. According to the BLS, a person on average has to send out 6 applications to get one interview implying that black people may have to send out two more applications than whites per interview they get.
However, there is no good reason to assume that the need for these two extra applications is due to unfair racial bias. The basic problem is this: blacks and whites differ on many traits which predict job performance only a few of which are directly measured by items on a resume and each such trait represents a non-biased reason for which employers might prefer white applicants to black ones.
The correlation between a trait and job performance is often called its “validity coefficient”. The validity coefficients for the things most often found on resumes are 0.07 for work experience, 0.10 for education, and 0.26 for reference checks. These correlations are not very strong and for many of these items, like work experience, there isn’t a large racial gap among applicants to begin with. By contrast, there are larger racial gaps in variables like general mental ability, situational judgement, emotional intelligence, etc., and each of these predicts job performance as well as or better than the best items on a resume. Much of this data can be seen on this chart from Sackett et al. (2021).
One objection to this argument is that even if resumes don’t directly measure all these traits variables like education and job experience are proxies for things like general intelligence and job knowledge. To some extent this is likely true, but the fact that these psychological variables predict job performance better than do resume items shows that resume items fail to capture much of the predictive validity of these traits.
This is not surprising. Suppose that the distributions of job performance among blacks and whites consist of two overlapping normal distributions, like this:
Now suppose that qualifications on a resume require a certain level of skill and ability to obtain such that those who would be bad employees cannot easily acquire them. As is hopefully evident in the below example, there is no possible threshold for job performance, or any other relevant trait (e.g. job related knowledge, cognitive ability, self discipline, etc.,) where the white mean is above the black mean in general, but not still above the black mean among those who exceed that threshold and so have the relevant qualification.
This becomes even more true if we make it easier for black people to acquire a given qualification than it is for a white person. In this scenario, among applicants with any such qualifications, white job performance will exceed black people’s job performance among those with equal qualification even if there is no mean difference in job performance between black and white people in the total population. Moreover, in this scenario whites who lack the qualification will still often perform better at work that blacks who do have the qualification.
This situation is exactly what happens when a society institutes affirmative action.
These theoretical considerations should be sufficient to show that these callback experiments are invalid measures of racism, but if you’d like empirical evidence to substantiate this consider the results of a massive study carried out by the federal government to measure people’s work-related cognitive abilities in terms of things like everyday math skills, writing skills, and the ability to efficiently use information taken from a document.
As can be seen, the general trend was such that black respondents were outscored by white respondents who had lower levels of education attainment than the black respondents did.
Similar results are found in Neill (1990), a paper which shows the mean AFQT percentile scores of black and white men aged 19-21 by education level for the years 1953-1958 and 1980. (The AFQT is a test designed by the military to measure cognitive skills relevant to job performance such as reasoning ability, mathematical ability, and reading ability.)
In 1980, Black people who had completed 3-4 years of college came in, on average, at the 49.7th percentile, or slightly below the average score unconditional on education. By contrast, the average percentile scores were 80.2 for whites with 3-4 years of college, 65.8 for whites with 1-2 years of college, and 46.5 for whites with 3-4 years of high school. This was even more extreme in the 50s, when black people with 3-4 years of college completed scored lower than whites with 3-4 years of high school.
Similar disparities are seen within occupations, with white people significantly out scoring black and Hispanic people on IQ tests when comparing people within the same industries (Murray, 2021).
Thus, it is clearly rational for employers to prefer the average white applicant over the average black applicant even if they are the same on paper in terms of things like work experience and educational attainment.
Ideally, we’d want to test this by comparing the rate at which black and white applicants are hired after controlling for the qualifications that show up on a resume as well as direct measures of the job performance related abilities that are not directly captured on a resume. To my knowledge, the only paper to do this is Ho (2005) who found that race did not predict whether an applicant was hired once such variables were controlled for.
The most important evidence on this question comes from Roth et al. (2003) who meta-analyzed data from 19 previous studies and found that black employees scored 0.30 standard deviations lower than white employees on measures of job performance even when they were working the same job at the same organization. This suggests that what is being required of applicants in terms of actual job performance, rather than on paper qualifications, is lesser for minorities.
This is consistent with most firms engaging in affirmative action in hiring and, because they are invalid measures of racial bias, this is also consistent with black applicants receiving 36% fewer call backs than white applicants.
Some might argue that this explanation is rude, racist even, because it requires that we note the job performance gap between black and white people. Brining up such facts may be rude, but it is no less rude or racist to blame inequality on the supposed immoral behavior of white people.
Another common objection to this argument is that it is unlikely that employers know about this data and so it cannot explain their discriminatory behavior. I agree that most employers aren’t familiar with this sort of data, but this does not preclude the possibility that they simply have intuitions based on experience about who is likely to make a good employee. These intuitions may be valid even if the employers don’t know exactly what is behind these intuitions and have not seen data showing that their own experience is generalizable.
That said, these intuitions are imperfect as shown by the racial gap in job performance among co-workers. Whatever the current mix of motives is for employers, the outcome of those motives is an economy in which black people are hired more easily than they would be if they were held to the same job-performance standards as white people. For this reason, anti-white bias in hiring is better evidenced than is anti-black bias.
Differences in Loan Acceptance Rates
Another common argument given to evidence racism concerns the fact that black people have a harder time than white people getting a loan or getting a loan at a good interest rate. Data from Pew shows that black people are indeed more likely to be denied for a mortgage loan, but even among blacks the rate of denial is only 27% (Desilver et al, 2017).
Concerning interest rates, Cheng et al. (2014) analyzed data from the U.S. Survey of Consumer Finances for the years 2001, 2004, and 2006 and found that controlling for measures of consumer behavior and debt risk reduced the black-white average interest rate gap to 0.29%
More importantly, credit scores don’t predict behavior equally well across races. Consider the following from a report given to congress by the federal reserve on how well loan performance is predicted by credit scores: “Consistently, across all three credit scores and all five performance measures, blacks… show consistently higher incidences of bad performance than would be predicted by the credit scores (p.89)”.
In other words, loans to black people have a higher risk of default even after controlling for credit score.
This report also notes that this is largely just true of black and white people with poor credit scores. Among those with high credit scores, there isn’t much of a difference across race in risk. This fact implies that if firms charge higher rates for black applicants due for economically rational reasons rather than unfair racial bias, then this will largely just be true of people with poor credit scores. And in fact a study by the Chicago federal reserve found no racial bias in loan approval rates among those with a good credit score but a significant bias in favor of whites among those with a bad credit score.
Even stronger against racial bias in lending comes from Bhutta and Hizmo (2019). They analyzed a data set consisting of all FHA-insured mortgages that originated in 2014 and 2015. After controlling for lender effects, credit score, and income, they found a black-white interest gap of 0.03% and a Hispanic-white gap of .015%. Moreover, it was found that minorities had on average paid for fewer discount points than whites. Correcting for this, it was shown that people of each race face the same price schedule. Thus, there appears to be no racial bias in the willingness of banks to give people loans.
Conclusion
In conclusion, the evidence does not justify the view that anti-black racism in the modern economy explains racial economic inequality. Of course, proponents of a racism based explanation may respond to this by deferring to historic racism as a cause of modern inequality but that will be the topic of another article.
Nagai (2008), Lerner and Nagai (2002), Nagai (2008), Armor (2004), Nagai (2008), Lerner and Nagai (2002), Danielson and Sander (2014), Armor (2004), Nagai (2006), Lerner and Nagai (2002), Armor (2004), Lerner and Nagai (2001), Nagai (2011), Danielson and Sander (2014), Lerner and Nagai (2000), Lerner and Nagai (2001), Nagai (2011), Lerner and Nagai (2000), Lerner and Nagai (2002).
"Comment for the algorithm."
Article or tweet topic suggestion: the claim that NBA referees are more likely to call fouls against black players. Commonly cited paper: Price, Wolfers 2010