Skip to navigationSkip to content

What is statistical discrimination?

A man wearing an end racism sign.
Reuters/Mark Thompson
End statistical racism too?
By Dan Kopf
Published Last updated This article is more than 2 years old.

Bill Spriggs hopes this is a teachable moment for economics.

In an open letter to economists published by the Minneapolis Federal Reserve in early June, the Howard University economist asked those in the field to reconsider their approach to studying racism. Economics has racism baked into it, in his view, and more economists need to open their eyes to that fact.

Spriggs, who serves on the board of the National Bureau of Economic Research, perhaps the US’s most influential organization for disseminating economic research, points out that early economics research in the US promoted the racial superiority of whites. The founders of the American Economic Association, the US’s leading economics professional group, were racists.

Spriggs, who was also appointed by Barack Obama to the US Department of Labor’s Office of Policy, believes this racist past has not been overcome, and is reflected in how economists study racism today. One example of how racism creeps into modern economic theory, according to Spriggs, is the theoretical concept of “statistical discrimination.” It is the idea that bigoted beliefs can be the result of observations about the differences between groups, rather than due to outright hostility to people of a certain background. Some researchers see the concept as a way to excuse racist attitudes.

“To black economists, ‘statistical discrimination’ is a constant micro-aggression,” wrote Spriggs in his letter. “It is a model that makes no sense.”

How economists think of racism

In economic theory, there are two main types of discrimination: “taste-based” discrimination and “statistical” discrimination. Both ideas were developed from the 1950s to 1970s as ways to understand how racism can be possible if people, like economists suppose, are acting in their best interests.

Taste-based discrimination is essentially prejudice: A white hiring manager who dislikes Black people might hire an inferior white candidate instead of a better qualified Black applicant.

In contrast, statistical discrimination is not necessarily driven by racial animus. Rather, it stems from assumptions an individual makes about others based on generalizations about their group. For example, a white hiring manager may have no issue with working with Black people, but believes they are more likely to have a criminal record than whites, and thus pose a safety risk. So, the manager decides to pass over a Black candidate in favor of a less qualified white one. But subsequently the manager learns that neither of them have a criminal record, and offers the job to the better qualified Black applicant. More information led the hiring manager to make a different decision. Statistical discrimination is based on an assumption, not a blanket dislike for people of a certain race. Outside of economics, people might call this stereotyping.

In economic theory, individual people act rationally to pursue their own well-being, so statistical discrimination is explained as an understandable, rational choice. Whether statistical discrimination is actually a useful way to think about racism is not as clear.

Missing history

One key problem with the idea of statistical racism, according to Spriggs, is that it does not have much room for history.

“[Statistical discrimination] is a belief that out of the clear blue sky, there are a set of economic actors who have convinced themselves that Black people and white people really are different, and that blackness is a salient piece of information when deciding to hire someone,” Spriggs told Quartz. In reality, he adds, this belief is the result of a long history of racism, starting with slavery. “There are 400 years of history that told you race matters. It divided who would be a slave, who could vote, who could go to school, and who can drink from this water fountain.”

“Once you would admit that groups of people had agency and created race, and that this was not economically neutral, these models don’t make sense,” said Spriggs. He thinks it’s silly that, on the one hand, the vast majority of economists admit that there is a history of racism that created modern-day society, yet when they study racism, they basically throw that out, and assume people are just making individual profit-maximizing decisions. It’s like studying climate change without taking into account humans’ role in generating it.

Timothy Taylor, an economist at Macalester College and the editor of Journal of Economic Perspectives, admits that economics is not very good at considering the myriad factors that have led to the US’s racial inequality all at once, including discrimination in the housing, education, and credit markets, among others. “We have a tendency to pick off one issue at a time. We can look very carefully at each tree, and perhaps not see the forest,” he said.

But he thinks that parsimony can be valuable. Economic analysis can help policy makers think about what to prioritize. “Should we be focusing on housing segregation? Or Black men coming out of prison? Or police reform? Or college admissions to Ivy League colleges?” said Taylor.

“We are really interested in how to drill down to some level of understanding, using theories like taste-based and statistical discrimination,” he added. “The great strength of economics is how these individual motivations come together in markets and lead to outcomes that you might or might not expect.”

A concept that needs a new framing

For an example of how economic thinking can lead to a surprising conclusion, Taylor points to one of the most well-known recent studies looking at statistical discrimination. The paper, authored by Sonja Starr, a criminal law professor at the University of Michigan, and Rutgers University economist Amanda Agan, examined the impact of “ban-the-box” measures on employment outcomes for white and Black job seekers. “Banning the box” refers to laws that make it illegal to have a tick box on job applications that asks about criminal records. Over the last several decades, many US states have passed such laws, hoping that it might help previously incarcerated people find employment.

Starr and Agan’s research suggests banning the box has had an unintended impact. They discovered it decreases the number of callbacks that Blacks and other minorities get after submitting an application. This is confirmed by other research showing that, in the short-term, banning the box has actually decreased employment for young black men.

Starr and Agan find that statistical discrimination is the most likely answer to this outcome. Left without information on criminal backgrounds, many employers just assume all Black applicants have a high chance of having a criminal record, and thus become less likely to even interview them for a job.

Notably, Starr and Agan also found that employers made inaccurate assumptions about the criminal history of Black and white people. Though the average Black person was more likely to have a criminal record than the average white person in the places they studied, the chances that an applicant had a criminal record were only slightly higher for Blacks than for whites. The employers’ assumptions were ignorant. The information employers had received from the tick box helped them act in a less ignorant manner.

Mario Small, a sociologist at Harvard University who studies institutions and racism, believes that by focusing so much on individual choices, like those of hiring managers, economists often miss that organizations can also discriminate. For instance, companies often fill jobs through referral networks. Small points to research that suggests that if a company has mostly white employees, their referrals are likely to be overwhelmingly white, perpetuating a lack of diversity at the company. The use of referral networks does not require any one person to make a discriminatory decision, but the policy itself reenforces discrimination.

Statistical discrimination can also cover up taste-based discrimination, explained Small. Just because a manager is willing to change a hiring decision when given more information about a Black candidate does not mean that conscious or subconscious racism didn’t cause them to be skeptical in the first place.

Starr agrees that statistical discrimination can be used to excuse bigotry. Still, it’s a useful concept because it can help us better understand how discrimination works, she added.

“We need to work harder to make clear to people that it is just as bad to make negative assumptions about a group of people as it is to not like them,” she told Quartz. “There is no moral or legal distinction.” She thinks using the term stereotyping instead of statistical discrimination might make it less likely for people to differentiate them morally.

Thinking of a different world

Spriggs wishes economists would not spend so much time studying (and sometimes justifying) the minutiae of racist behavior. He thinks too much effort is put into studies looking at the marginal impacts of certain policies, such as the ban-the-box research.

He believes a focus on practicality leads economists to think too small. The field has become so narrow in its questions, it too rarely asks about anything that really matters. Economists would be better off if the discipline accepted that racism is a pervasive part of American society, he says, and focused on learning the history of how that racism is perpetuated.

For examples of smarter ways to study racism, Spriggs points to the work of Ohio State economist Travon Logan, who has looked into how Black voting rights after the US Civil War led to improved public finances and concrete gains in Black literacy that were erased when those voting rights disappeared at the end of Reconstruction. He also commends the work of Jhacova Williams, an economist at the Economic Policy Institute. She has examined how the share of streets named after Confederate leaders in a city predict differences in unemployment rates and earnings between Blacks and whites even after accounting for educational attainment, which suggests that places with a history of racism have worse racial inequality today.

Spriggs wants economists to be more imaginative about what could change. As he points out, systems do transform. The power of governments can rise and fall. Systems like slavery can be abolished.

“We’ve become so focused on marginal analysis, we’ve lost our ability to talk about systems,” said Spriggs. “We should ask how does the system work? How did we get these levels of inequality. But most economists don’t think about it that way, they assume the system. They assume they are in a racist society, and then go from there.”

What’s needed, in the minds of critics like Spriggs, are economists who are willing to look at racism and its impacts holistically in the way that the French economist Thomas Piketty has addressed inequality. It’s harder and messier work, but it can have much more impact than measuring whether hiring managers express taste-based or statistical racism.

📬 Kick off each morning with coffee and the Daily Brief (BYO coffee).

By providing your email, you agree to the Quartz Privacy Policy.