Economists tend to see discrimination as based on actions of individuals, who in turn are interacting in markets and society. However, sociologists do not feel the same compulsion as economists to build their theories on purposeful decision-making by individuals: \”Sociologists generally understand racial discrimination as differential treatment on the basis of race that may or may not result from prejudice or animus and may or may not be intentional in nature.\”   The Spring 2020 issue of the Journal of Economic Perspectives illustrates the difference with a two-paper symposium on \”Perspectives on Racial Discrimination: 

As most economists learned somewhere along the way, one can think of individual motivations for discrimination as coming in two flavors: taste-based discrimination in the oeuvre of Gary Becker (Nobel 1992) or \”statistical discrimination\” from the writings of Edmund Phelps (Nobel 2006) and Kenneth Arrow (Nobel 1972). One can dispute how economists discuss the subject of discrimination, but it would just be false to claim that it has not been a high-priority topic of top-level economists for decades.)
Taste-based discrimination is the name given to racial prejudice and animus. Statistical discrimination refers to the reality that we all make generalizations about people. Sometimes the generalizations are socially useful: Lang and Spitzer mention the generalizations that people are more likely to give up their seat on the bus or subway to a pregnant woman or an elderly person, based on the statistical generalization that they are more likely to need the seat, or that health care providers are more likely to emphasize breast-cancer screening for women than for me. However, statistical discrimination can also be harmful: say, if it is based on beliefs that those of a certain race who are applying to be hired for a job or to rent an apartment are more likely to be criminals. Moreover, when statistical discrimination is based on inaccurate statistics and exaggerated concerns, it begins to look functionally similar to taste-based discrimination. 
In addition, economists have long pointed out that the effects of discrimination may vary based on the parties involved: for example, in the context of labor market discrimination one can look separately at discrimination by employers, by co-workers, and by customers. For example, if the issue is discrimination by employers, one possible result is firms that are segregated by race, but both selling to the same consumers. If the issue is discrimination by consumers, one result may be that whites become more likely to have the \”front-facing\” jobs that deal directly with customers.  
The economic approach to discrimination, with its focus on purposeful and intentional acts by individuals, can offer some useful insights, and Lang and Spetzer give a useful overview of the research. For example, while the basic statistics show that blacks are more likely to be arrested for traffic violations, how can we know whether this is linked to prejudiced behavior by the police? One line of research has looked at traffic violations at different times of day, when there is more or less daylight. The underlying idea is that racial prejudice is more likely to manifest itself when the police can see the driver! The evidence from these studies is mixed: One study found that no effect of daylight on the racial mix of traffic stops, but another found that blacks were stopped more often at night on street with better lighting. 
Studies of \”ban-the-box\” legislation also have unexpected effects, as Lang and Spetzer point out: 

Because a higher proportion of blacks have criminal records than whites do, one might expect that preventing employers from inquiring about criminal records, at least at an early stage, would increase black employment. However, if firms cannot ask for information about criminal records, they may rely on correlates of criminal history, including being a young black man. This concern is even greater if employers tend to exaggerate the prevalence of criminal histories among black men, thus leading to inaccurate statistical discrimination. Agan and Starr (2018) investigate “ban the box” legislation in which companies are forbidden from asking job applicants about criminal background. Before such rules took effect, employers interviewed similar proportions of black and white male job applicants without criminal records. Prohibiting firms from requesting this information reduced callbacks of black men relative to otherwise similar whites. Consistent with this, Doleac and Hansen (2016) find that banning the box reduced the employment of low-skill young black men by 3.4 percentage points and low-skill young Hispanic men by 2.3 percentage points. Similarly, occupational licensing increases the share of minority workers in an occupation despite their lower pass rates on such exams (Law and Marks 2009). Prohibiting the use of credit reports in hiring reduced black employment rather than increasing it (Bartik and Nelson 2019). Taken together, these studies provide strong evidence that statistical discrimination plays an important role in hiring.

As sociologists, Small and Pager have no direct issue with this kind of work in economics: as they point out, some sociologists work in a similar vein. But their essay emphasizes that discriminatory outcomes can emerge from reasonable-sounding institutional choices and from history. 

For example, many companies, when they are hiring, encourage current workers to refer their friends and neighbors. This practice is not overtly racial. But given US patterns of residential segregation and friendship, it means that new hires will tend to reinforce the earlier racial composition of the workforce Or consider the standard practice that when doing layoffs, last hired will be first fired. If a company has only fairly recently started hiring minority groups, then the weight of layoffs will fall more heavily on these groups. As Small and Pager write: 

It is not surprising that a national study of 327 establishments that downsized between 1971 and 2002 found that downsizing reduced the diversity of the firm’s managers—female and minority managers tended to be laid off first. But what is perhaps more surprising is that those companies whose layoffs were based formally on tenure or position saw a greater decline in the diversity of their managers; net of establishment characteristics such as size, personnel structures, unionization, programs targeting minorities for management, and many others; and of industry characteristics such as racial composition of industry and state labor force, proportion of government contractors, and others (Kalev 2014). In contrast, those companies whose layoffs were based formally on individual performance evaluations did not see greater declines in managerial diversity (Kalev 2014).

In other cases, actions taken for discriminatory reasons in the past can have effects for long periods into the future. For example, blacks are much less likely to accumulate wealth through homeownership than whites, and one reason dates back to decisions made by federal agencies in the 1930s. 

However, the Home Owners Loan Corporation and Federal Housing Administration were also responsible for the spread of redlining. As part of its evaluation of whom to help, the HOLC created a formalized appraisal system, which included the characteristics of the neighborhood in which the property was located. Neighborhoods were graded from A to D, and those with the bottom two grades or rankings were deemed too risky for investment. Color-coded maps helped assess neighborhoods easily, and the riskiest (grade D) neighborhoods were marked in red. These assessments openly examined a neighborhood’s racial characteristics, as “% Negro” was one of the variables standard HOLC forms required field assessors to record (for example, Aaronson, Hartley, and Mazumder 2019, 53; Norris and Baek 2016, 43). Redlined neighborhoods invariably had a high proportion of AfricanAmericans. Similarly, an absence of African-Americans dramatically helped scores. For example, a 1940 appraisal of neighborhoods in St. Louis by the Home Owners Loan Corporation gave its highest rating, A, to Ladue, an area at the time largely undeveloped, described as “occupied by ‘capitalists and other wealthy families’” and as a place that was “not the home of ‘a single foreigner or Negro’” (Jackson 1980, 425). In fact, among the primary considerations for designating a neighborhood’s stability were, explicitly, its “protection from adverse influences,” “infiltration of inharmonious racial or nationality groups,” and presence of an “undesirable population” (as quoted in Hillier 2003, 403; Hillier 2005, 217).

More recent research looks at the long-term effects of the boundaries that were drawn at the time. 

The results are consistent with the HOLC boundaries having a causal impact on both racial segregation and lower outcomes for predominantly black neighborhoods. As the authors write, “areas graded ‘D’ become more heavily African-American than nearby C-rated areas over the 20th century, [a] . . . segregation gap [that] rises steadily from 1930 until about 1970 or 1980 before declining thereafter” (p. 3). They find a similar pattern when comparing C and B neighborhoods, even though “there were virtually no black residents in either C or B neighborhoods prior to the maps” (p. 3). Furthermore, the authors find “an economically important negative effect on homeownership, house values, rents, and vacancy rates with analogous time patterns to share AfricanAmerican, suggesting economically significant housing disinvestment in the wake of restricted credit access” (pp. 2–3).

While economists have not totally neglected the role of institutions and history in the transmission of racial discrimination, it\’s fair to say that it hasn\’t been their main emphasis, either.  My own sense is that through most of US history, the main issue of racial discrimination was explicit white prejudice. But the balance has shifted, and current differences in racial outcomes are a difficult combination of history, institutions, and social patterns. 
For example, one theme that has emerged from earlier research both by economists and sociologists is that discrimination can reduce the incentives to gain human capital. Indeed, a group that is has experienced discrimination may end up with less human capital for interrelated reasons: less access to educational resources, reduced motivation to gain human capital (because of lurking future  discrimination), reduced expectations or less support from family and peer groups,  and other reasons. Once this  dynamic has unfolded, then even employers who have zero preference for taste-based discrimination, but just hire on the basis of observable background and skills, will end up with different labor market outcomes by race. 

Published by conversableeconomist

Leave a Reply