A Teachable Moment?

In an open letter to economists, AFL-CIO chief economist William Spriggs recently asked, "Is now a teachable moment for economists?" From his perspective, the economics profession has done a poor job of studying racial discrimination, and he expressed hope that the death of George Floyd and the protests it spawned will lead to improvement in how it approaches race. In my role at the Richmond Fed, it is important for me to think hard about how the profession has addressed race through research and what it might do in the future.

Economists have been slow to view racial discrimination, especially in the modern era, as a central driver of observed disparities. Interestingly, perhaps, this may be because economists usually assume employers have no concern for societal well-being and are only focused on maximizing profits. In early models of discrimination, notably those of Gary Becker, discriminating firms put themselves at a labor-cost disadvantage and therefore could only survive competition if a high percentage of their competitors also practiced discrimination. This led economists to suspect that other forces, including legal limits on hiring black workers, were critical for perpetuating racist outcomes, since without them, it would be easy for non-discriminatory firms to profit by hiring talented workers without regard to race. Yet even as overt barriers have disappeared, outcomes remain disparate.

Another prominent theory of racial discrimination holds more promise. This approach, first formalized by Edmund Phelps and Kenneth Arrow and advanced since then by many others, emphasizes a potentially long-lived kind of bias called "statistical discrimination." It is based on the idea that, faced with incomplete information about individuals, employers may be able to make statistically valid, but not necessarily socially rational, inferences about individuals by taking into account the average attributes of their group. Glenn Loury, for example, has developed dynamic models of statistical discrimination in which "reputation traps" create self-reinforcing cycles of poor opportunity and insufficient investment in education and training. These dynamic models suggest even more that disparities between groups can be long-standing and pernicious in the absence of government intervention. Sadly, individuals may suffer in the interim for no reason other than entrenched pessimism about them as a group. Notice that an obvious candidate is the overt institutionalized racism of the past — it "initially" limited opportunities and made such purely statistical beliefs possible to hold in the first place.

In economics, the data always matter. So whatever our theories may say, economists have produced many studies that have identified evidence of racial discrimination. Much of this research has focused on labor markets and has used statistical analysis to estimate whether race remains a statistically significant determinant of wages after taking into account various indicators of worker productivity, including education and experience.

Still, empirics can't always settle things. Because most of the data economists analyze don't come from controlled laboratory experiments, the possibility usually exists that estimated results have been distorted by mechanisms that have not been included in the analysis. This is called the "missing variables" problem.

But it might be that the economics profession has displayed a level of skepticism toward evidence of discrimination that goes well beyond what can be accounted for by methodological rigor alone. In reference to the profession's frequent use of the missing variables critique, Spriggs observed that "it looks like economists are desperate for a 'Great White Hope,' some variable that can be used to once and for all justify racial disparities."

Referring to the profession's skepticism toward evidence of racial discrimination, Arrow once said, "While one can always invent hypotheses to explain away these results, there is really no reason not to draw the obvious conclusions." Although Arrow was a giant in the field of theoretical economics, his prior beliefs about discrimination were heavily influenced by real-life history prior to the passage of the Civil Rights Act of 1964. "I can speak as a witness here," said Arrow, "it was simply well-known that most good jobs were not available to blacks." According to Arrow, "any theory of racial discrimination … has to be consistent with these patent facts."

Arrow recognized a contradiction. In his view, the market-based solutions produced by standard economic models "tend to predict that racial discrimination will be eliminated." But since, in his view, this had not been borne out by history, he counseled that "we must seek elsewhere for non-market factors influencing economic behavior." This suggests that the profession may benefit by engaging more seriously with the premise embraced by Spriggs and so many social scientists outside the economics profession — that discrimination works through slowly evolving institutions as well as through individuals.

Looking ahead, I hope recent events will energize deeper engagement on racial bias by economists — very much including the many working within the Fed — and lead to better understanding of its effects and of policies aimed at its elimination.

Kartik Athreya is executive vice president and director of research at the Federal Reserve Bank of Richmond.