Citation
Collins, W. (2001). The labor market impact of state-level anti-discrimination laws. National Bureau of Economic Research working paper no. 8310. Cambridge, MA: NBER.
Highlights
- This study’s objective was to assess the impact of state-level fair employment legislation, enacted in the 1940s through the 1960s, on labor market outcomes including income, unemployment, and labor force participation, by gender and race.
- The study used a nonexperimental design, drawing on data from the 1940, 1950, and 1960 decennial U.S. Censuses for 19 non-southern states.
- The study found that state-level fair employment laws adopted in the 1940s had larger positive effects on the relative income of African American workers than those adopted in the 1950s. In contrast, laws adopted in the 1950s led to larger increases in African American male unemployment but larger increases in African American women’s labor force participation. Only states that implemented the laws in the 1940s experienced a narrowed income gap between white and African American workers.
- The quality of causal evidence presented in this report is low because the author did not demonstrate that the states that adopted fair employment legislation when they did were comparable to those that did not adopt or adopted at a later time. This means we are not confident that the estimated effects are attributable to state-level fair employment legislation. Other factors are likely to have contributed.
Intervention Examined
State-level Fair Employment Legislation
Features of the Intervention
The study assessed the labor market impacts of state-level fair employment legislation adopted by 19 non-southern states. Fair labor laws prohibited employment discrimination on the basis of race, color, creed, and national origin. Before the 1940s, many states had embedded in their constitutions a principle that banned religious discrimination for public officials. Although these principles expanded over time to cover other kinds of government employees and forms of discrimination, it was not until after 1940 that fair employment laws expanding these principles to cover private employment were introduced. The details of these laws varied by state, but they generally included enforcement methods that gave state agencies the power to receive and investigate discrimination complaints, eliminate unlawful discrimination, and issue cease-and-desist orders to noncompliant firms. The first state to adopt fair employment laws was New York in 1945. By 1964, 22 non-southern states had also adopted them.
Features of the Study
The study used a nonexperimental design. The author estimated three difference-in-difference-in-differences regression models for each outcome and for men and women separately. Each model examined the impact of state-level fair employment laws adopted in a given decade by comparing changes in native-born African American and white labor market outcomes from the beginning of that decade with the beginning of the next, across non-southern states that did and did not adopt laws in that decade. For instance, the first model examined the impact of laws adopted by four states in the 1940s (Connecticut, Massachusetts, New Jersey, and New York) by comparing changes in their outcomes between the 1940 and 1950 U.S. Censuses with changes in three states adopting laws after 1950 (Michigan, Ohio, and Pennsylvania). The second model examined the impact of laws adopted in the 1950s and the third examined the impact of laws adopted from 1940 to 1955.
Each model controlled for individual-level characteristics, including age, education, marital status, and an indicator for metropolitan residence. Outcomes included annual income, unemployment, labor force participation, measures of occupational status, and industry of employment. The data sample included people ages 20 to 59 in the 1940, 1950, and 1960 U.S. Censuses, and was further restricted for the income analysis to full-time workers who were not in school or self-employed.
Findings
- The study found that state-level fair employment laws adopted in the 1940s had a positive, statistically significant impact on the relative income of African American women, increasing income by 12 percent for African American females relative to white females.
- In the 1950s, fair employment laws led to a statistically significant 4 percent increase in African American male unemployment but an 8 percent increase in African American female labor force participation.
- Fair employment laws had statistically significant positive impacts on occupational status for African American men throughout the study period, and larger positive impacts on African American women. As an illustration, African American women moved into operatives and craftsmen occupations as well as manufacturing, while African American men moved into manufacturing jobs.
Considerations for Interpreting the Findings
States chose whether and when to introduce fair employment legislation. This choice might have depended on historical trends in African American and white labor market outcomes in the state, as well as prevailing attitudes regarding discrimination. Thus, the estimated impacts could reflect both the effect of relative African American labor market outcomes or discrimination on the timing of adopting fair employment legislation, as well as the effect of the legislation on relative African American labor market outcomes. The author did not account for possible differences in pre-intervention trends between the two sets of states.
In addition, the author estimated multiple related impacts on African American labor market outcomes. Performing multiple statistical tests on related outcomes makes it more likely that some impacts will be found to be statistically significant purely by chance and not because they reflect program effectiveness. The author did not perform statistical adjustments to account for the multiple tests, so the number of statistically significant findings is likely to be overstated.
Causal Evidence Rating
The quality of causal evidence presented in this report is low because the author did not demonstrate that the states that adopted fair employment legislation when they did were comparable to those that did not adopt or adopted at a later time. This means we are not confident that the estimated effects are attributable to state-level fair employment legislation. Other factors are likely to have contributed.