Excerpted From: Ricard Gil and Justin Marion, Why Did Firms Practice Segregation? Evidence from Movie Theaters During Jim Crow, 65 Journal of Law & Economics 635 (November 2022) (29 Footnotes/References) (Full Document)


GilMarionMany cities and states were slow to dismantle the institutions of segregation. Prior to the Civil Rights Act of 1964, segregation in public accommodations significantly affected African Americans' access to public services and private businesses. While southern cities frequently mandated the separation of races--often in specific settings such as hospitals, restaurants, and public transportation--segregation was practiced to a significant degree by businesses even in the absence of any legal proscription. Indeed, businesses often excluded black patrons despite city ordinances banning segregation, which were regularly ignored and unenforced.

In the absence of mandated segregation, a firm's decision to exclude minority customers reflects the racial preferences of its stakeholders, including owners, workers, and customers. Our goal is to understand how the racial biases of customers and firms influenced the practice of racial exclusion. We study movie theaters in the early 1950s, when explicit segregation was still common in southern states. The effect of segregation on firms' profitability can be used to infer the relative importance of the racial preferences of the firm's stakeholders. Racial exclusion can be profitable when responding to the biases of white customers. Conversely, business owners may be willing to sacrifice profits to satisfy their own prejudices.

In the main part of the paper, we use unique data on the weekly box office revenues from a nationwide sample of movie theaters to estimate the effects of the 1953 desegregation of Washington, DC, businesses on the revenues earned by the city's theaters. We then supplement these results by examining films with black actors cast in prominent roles and how their box office performance, as measured by revenues and run lengths, depended on the racial bias of the city in which the theater was located. As we discuss below, the results from these two empirical exercises together allow us to separately test for the influence of customers' and firms' discrimination.

The desegregation of Washington, DC, businesses occurred rapidly in the summer of 1953. Until then, the movie theaters in Washington barred attendance by African Americans despite long-forgotten 19th-century laws outlawing segregation in public accommodations in the city. A US Supreme Court ruling in June 1953, which applied only in the District of Columbia, subsequently required that those laws be enforced, which led to a rapid desegregation of the city's businesses.

Using a difference-in-difference design, we find that revenues of theaters in Washington fell by 11 percent after desegregation relative to theaters in other cities showing similar movies, and the timing of the revenue response matches the date of the Supreme Court's ruling. While the opening of movie theaters to the African American market could conceptually influence optimal theater pricing, there is no strong evidence of a price response. We conclude that ticket sales to white customers fell after desegregation, at least in the short run. Finally, we find suggestive evidence that desegregation altered the composition of movies selected to be screened in favor of those more popular with African American audiences. These results are consistent with the hypothesis that customers' discrimination played a role in perpetuating segregation.

The postintegration decline in profits strongly suggests discrimination by customers, but from this result we cannot say whether firms were also prejudiced and whether that may have contributed to racial segregation. We address this question in the second part of the paper. Using a theoretical model of screening decisions by movie theaters, we show how the revenue earned by films with black actors, in conjunction with their run lengths, can be used to test for firms' prejudice. The intuition of the test is based on the fact that a movie's run length is decided by the theater (or its agreement with the production studio) and not by the customer. Conditional on revenues earned by a movie through week t of its run, the probability of continuation (hereafter, continuation probability) in week t + 1 of a movie with a black cast member depends on discrimination by firms but not customers. The greater is the racial bias of the firm, the lower will be the conditional continuation probability of the black-cast movie. In other words, by ending the run of still-profitable movies, a racially biased theater owner makes a fiscal sacrifice to satisfy his or her racial bias.

We compile data on all movies with black actors that were produced by major studios and released during the years covered by our theater revenue data. To measure racial bias, we use the index we constructed in Gil and Marion, which was formed from respondents' views on race and segregation in public opinion polls from the late 1940s and early 1950s. We find that a movie with black actors screened in an area with greater racial bias earned less revenue-- around 11 percent less--compared with what it would have earned in an area with less racial bias. The run length contributes to this difference, as there is a .12 difference between racially biased and unbiased areas in the number of weeks that black-actor movies were screened. However, we are unable to reject a null hypothesis that firms are unbiased. Conditional on the revenue earned in the first week, the difference in the continuation probability of black-cast and white-only movies was not influenced (statistically) by the racial bias in the city. Similarly, conditional on run length, the revenue difference between the two types of movies does not depend on racial bias. Together, our results point toward customers' discrimination as a key determinant of theater policy during this era. We fail to find evidence of racial discrimination by firms, and to the extent that firms' owners or their workers are biased, the effects of these preferences appear to be secondary.

Analogous to the discrimination faced by black workers discussed in Becker (1957), the prejudice experienced by black moviegoers is determined by the discrimination of the marginal theater, and racial exclusion could remain the policy at many theaters without affecting the consumption opportunities of black audiences. Indeed, at the time there were many theaters specifically targeting African American customers (Gil and Marion 2018). However, customers' discrimination creates profit rewards for racial exclusion that can survive entry, and movie theaters (along with many other public accommodations) are characterized by increasing returns. Entry of firms specifically serving a minority market would be disadvantaged by operating at a smaller scale, and African American theaters of the era did in fact have less capacity, smaller screens, and fewer amenities. Therefore, black welfare can suffer as a result of racial exclusion even with the entry of unbiased firms.

Segregation of public accommodations has received little attention in the economics literature. This may be due in part to the practical complication that segregation laws were inconsistent and piecemeal, and their enforcement was uneven. Furthermore, segregation often resulted from informal local practices rather than formal laws. Cook et al. use data from volumes of The Negro Motorist Green Book, an annual directory published from 1938 to 1966 listing businesses serving African American customers, to document several facts about the geographic patterns of response to discrimination in public accommodations. Wright provides a history of desegregation in public accommodations preceding the passage of the Civil Rights Act of 1964, noting that the fear of alienating white customers motivated segregating firms. While not specifically attempting to identify the effect of segregation on profits, Wright shows that retail sales grew in southern areas during the 1960s at a rate meeting or exceeding that in other regions. Since that period coincided with widespread desegregation, Wright argues that desegregation was a positive force for businesses, in contrast to our findings. One way to reconcile the two results is to consider that racial attitudes improved during that time, and the effect of desegregation on demand by white customers may have been declining. An alternative interpretation is that the Civil Rights Act positively affected the economy, including southern labor markets, which could instead be responsible for the increase in retail sales in the South.

Our study also fits with a recent literature studying historical institutions of racial bias and residential segregation. Troesken and Walsh examine how residential segregation ordinances arose in the early 1900s, finding that where whites more easily organized to enforce informal residential segregation norms, segregation laws were less likely to be implemented. Boustan (2010) examines the role that postwar black migration played in the suburbanization and resulting residential segregation of northern cities. Cook, Logan, and Parman, using the detailed measures of residential segregation developed by Logan and Parman, find that segregation increased racial violence in the form of lynchings of African Americans. This suggests a causal channel running from segregation to racial preferences and discrimination. Importantly, historical lynchings have lingering effects and are related to modern rates of racial violence.

Our results also relate closely to the established literature on discrimination against customers and workers. An important antecedent to our study is Heckman and Payner, which examines the impact of federal antidiscrimination legislation on the employment outcomes of black workers. In a clear analogue to our setting, firms' owners had a profit motive to hire black workers yet also felt pressure from customers, white workers, and other stakeholders to exclude black employees. In a paper considering the modern film industry, Kuppuswamy and Younkin (2020) find that films in 2011-15 with more diverse casts are associated with higher box office revenue. As in our study, Leonard, Levine, and Giuliano (2010) use sales to uncover discrimination by customers, finding that a mismatch between the demographics of the employees at a retail outlet and the residents in the surrounding neighborhood has a small negative impact on sales. Similarly, Holzer and Ihlanfeldt (1998) find evidence of customers discriminating in US retail. Bar and Zussman examine a similar question, showing that Jewish customers in Israel prefer to be served by Jewish rather than Arab workers, which in turn influences the hiring decisions of employers. Waldfogel and Vaaler  consider how firms are willing to forgo profits to appease the ethnic biases of customers: airlines omit Israel from online route maps if they serve customers from countries with stronger anti-Semitic views.

The paper is organized as follows. Section 2 contains the model. In Section 3, we provide a background description of relevant institutional details, and in Section 4 we describe the data. Section 5 shows the empirical results related to the impact of desegregation on firms' profits, while in Section 6 we describe the results related to the box office performance of films with black actors. Section 7 concludes.

[. . .]

In this paper, we study how the profits of firms responded to racial segregation in public accommodations during Jim Crow, with the goal of understanding firms' motivations for voluntary racial exclusion practices. Using a plausibly exogenous court ruling that desegregated Washington, DC, movie theaters in 1953, we find that box office revenues fell in Washington compared with other cities in the United States. The results suggest that theaters may have excluded customers of color to cater to the racial biases of their white customers. We provide further evidence pointing toward customers' discrimination, and away from firms' racial bias, by examining the screening decisions made by theaters regarding movies with black actors in significant roles. These movies earned less revenue and were screened for fewer weeks in cities with greater racial bias. Using a theoretical model of the screening decision of theaters, we propose two related tests for firms' discrimination, and we are unable to reject a null hypothesis that firms are unbiased. These findings contribute to understanding the persistence of de facto segregation despite the profit incentive to expand a theater's market by serving customers of all races.

Explicit segregation is no longer legal, but the issues we examine in this paper remain relevant. There are direct parallels with the modern film industry, which continues to grapple with the level of representation of actors and directors of color. In a different context, recent court cases considered the question of whether business owners are permitted to refuse service to gay customers on religious grounds, which serves to highlight the continued importance of understanding firms' voluntary exclusion of customers and the forces behind the persistence of institutions favoring segregation and racial exclusion.