Association for Public Policy Analysis & Management

A long history of racial segregation in American schools has resulted in a number of court-ordered desegregation policies. However, most of these policies have focused on student sorting rather than teacher employment. I use a combination of econometric methods – difference-in-differences, synthetic control, and propensity score matching – to determine the causal impact of a contemporary court-ordered affirmative action hiring policy that gives extensive preference to black applicants. There are several challenges to measuring the magnitude of these impacts. First, historical data on black teacher employment share are not publicly available. To measure employment of African American teachers, I use confidential data collected biennially by the Equal Employment Opportunity Commission (EEOC). The second evaluation challenge regards selection – estimated effects should reflect the actual impact of the hiring policy rather than the impacts of non-policy related factors. If the court-ordered hiring policy was simply implemented at a time coincident with changing social factors that would have increased the probability that a minority fills a job vacancy regardless of any policy intervention, we would incorrectly attribute outcomes to the implementation of the hiring policy. To address this selection problem I rely on multiple econometric approaches – difference-in-differences (DD), synthetic control, and propensity score matching.

I find that the court-ordered hiring policy had a statistically significant impact on the black teacher employment share in the school district with estimates ranging from a 3.7 to 5.6 percent increase, depending on the comparison group. This change seems to be driven by significant increases in black teacher hiring at the secondary level. The court-ordered hiring policy increased the black elementary teacher employment share by 4.2 percent in the post-policy period; however, it increased the black secondary teacher employment share by 9.3 percent. Only the black secondary teacher employment share estimate reaches statistical significance. The court-ordered hiring policy decreased the student-teacher representation gap by 0.1 to 4.0 percent, depending on the comparison group. The propensity score matching estimate (0.1 percent) is not statistically significant; however, the DD and synthetic control estimates (4.0 and 2.3 percent, respectively) reach conventional levels of statistical significance. The court-ordered hiring policy had no significant effect on student achievement as measured by a number of state standardized examinations. The threat of litigation does little to change the black teacher employment share, black teacher hiring share, or student-teacher representation gap. These findings would suggest that contemporary court-ordered affirmative action hiring policies in a school district could have a substantial impact on workforce composition but little impact on student achievement.