Does Customer Prejudice Help Drive the Employment Gap Between White and Black Americans?

may18_15_938927176 — Marie LaFauci/Getty Images

The employment rate gap between black Americans and white Americans has remained persistently high over the past 30 years. Looking at 2014 data from the Bureau of Labor Statistics, white Americans were 10% more likely to be employed than African-Americans were. A great deal of research — using methodologies such as resume studies and other models — has found that employer discrimination and prejudice plays a large role in creating this gap. But in a recent paper, I show that discrimination from consumers can also help explain the lower employment rate among African-Americans.

The idea of customer discrimination influencing the labor market stems from the economic theory of taste-based discrimination, or when some individuals prefer not to interact with people from certain demographics. In the case of consumer discrimination, some customers may prefer not to interact with minority workers. As a result, some employers may reject minority applicants to satisfy their customers’ preferences. So even if employers are themselves unprejudiced, their hiring behavior may be driven by the fear of losing customers and diminishing profits.

Consumer discrimination in the labor market has not been extensively studied, but other research has borne this out. Using data on employers in four large metropolitan areas, Atlanta, Boston, Detroit, and Los Angeles, one paper from 1998 found that the racial composition of customers at an establishment affects who gets hired. The larger the fraction of minority customers, the higher is the probability that workers from that minority group will be hired. They found that this result is stronger in jobs that involve direct contact with customers, especially sales and service jobs. And a 2010 study of more than 800 retail stores and their surrounding communities found that sales slightly fell in white communities when black employment shares in those stores rose.

Whereas both of these studies looked at consumer race, I tried to determine whether prejudiced consumers in a given area affected the share of African-Americans in consumer-facing service jobs in that area and hence reduced their employment rate overall. This involved measuring the level of racial prejudice in different areas and comparing it to the employment rate of African Americans in those areas.

To measure the amount of racial prejudice across the U.S., I used data from the General Social Survey, which has been tracking trends in attitudes, behaviors, and attributes in the U.S. since 1972. This representative dataset has questions about issues strongly related to racial prejudice. To measure employment outcomes, I used U.S. Census data from 2000.

From the General Social Survey, I selected two waves of data to build a proxy for prejudice between 1996 and 2004. I used responses to the question, “Do you think there should be laws against marriages between African-Americans and whites?” to compute the share of prejudiced individuals for each state as the percentage of white respondents aged 18 and older who answered yes. (In the theoretical model, we supposed that racial prejudice comes from whites only.)

During my study period, 5,050 individuals answered this question, and around 15% of white respondents said there should be laws against marriages between African-Americans and whites. This share of prejudiced individuals was my proxy for racial discrimination in each state. The share of racial prejudice was most severe in the southeastern portion of the country and least severe in New England and in the West. Of course, this one survey question is not a complete measure of racial prejudice, and it may actually be underestimating racial prejudice if people are cautious in expressing their views. But compared to the other GSS questions, it is particularly suitable for revealing the true prejudice individuals may have in interacting with African-Americans.

From the U.S. Census data, I looked at the occupations of both white Americans and black Americans in 2000 to see whether black Americans were less likely to occupy customer-facing jobs than whites in high-prejudice areas. I restricted my sample to men to avoid confounders related to family arrangements, residential choices, and female labor market outcomes. I also focused on men who have at most a high school diploma since they are the group most exposed to unemployment risk. My sample size included 1,123,500 low-skilled men, of whom 488,290 were employed.

The Census details individuals’ occupations but does not indicate whether the worker is in more contact with customers. So I also used the Occupation Information Network (O*NET) database, which has a measure of how important “dealing directly with the public” is for a given occupation. The jobs that rank highly on this involve serving customers in restaurants and stores and receiving clients or guests (e.g., bartender, parking lot attendants, retail sales clerk, ticket and reservation agents, etc.).

To measure the effect of customer discrimination on employment outcomes, you have to disentangle it from two other explanations for lower black employment rates in customer-facing jobs: employers discriminating against African Americans and African-Americans themselves not seeking those jobs.

First, I reasoned that if an increase in the share of prejudice reduces the employment rate differential (employment rate of African-Americans minus the employment rate of white Americans), then racial discrimination exists. Second, if an increase in the share of prejudice reduces the probability of working in a customer-facing job, then consumer discrimination exists. I then computed the “residual employment gap” in each state — that is the gap in employment between blacks and whites that is not explained by individual characteristics (like education, age, marital status), racial-specific job preferences, or location. I also controlled for how many African-Americans lived in each state. Next, I analyzed whether or not the share of racial prejudice explained this residual employment gap — and whether or not the share of racial prejudice explained the probability of African Americans working in a customer-facing job.

On average, my estimates showed that in 2000, white American men were 8 percentage points more likely to be employed than black men. A black man had a lower probability of being employed even after controlling for characteristics like education, age, and location.

Overall, I found that black men were just as likely as white men to be employed in customer-facing jobs. But when you look at states with more prejudice, black men were less likely to occupy customer-facing jobs. In fact, they were less likely to be employed at all. Specifically, an increase in the intensity of discrimination by one standard deviation reduced the raw employment rate of African-Americans by 9%–20% and the rate of working in a customer-facing job by 21%–30%.

These results suggest that there is indirect customer discrimination in customer-facing jobs in the U.S. labor market. Both in the theoretical model and in the empirical data, the racial-specific preferences of job seekers are taken into account, so we can rule out the explanation that this result is due to African-Americans being less likely to apply for these jobs. Therefore, it seems that employers in the service sector make hiring decisions based on fears of customer discrimination.

These discriminatory practices have major implications for the employment of discriminated groups. Over the past thirty years, these service-sector jobs have replaced the manufacturing jobs that used to hire men with less education. In excluding this community from job opportunities in this growing sector, consumer discrimination particularly hurts black men economically. As a result, it reduces the overall employment rate of African-Americans.

My work suggests several avenues for future research, but one would be to explore how employers perceive local racial prejudice and how it actually influences their hiring decisions. Empirical findings based on economic theory should always be challenged with sociological resources using surveys or interviews.

We need to know more about the obstacles black males face in the labor market. Understanding the mechanisms driving discrimination is also necessary to start designing efficient ways to fight racial inequalities.