The Impact of Eliminating Affirmative Action on Minority and Female Employment: A Natural Experiment Approach Using State-Level Affirmative Action Laws and EEO-4 Data

Ending affirmative action programs significantly decreased diversity in the workplace, disproportionately hurting Asian women, Black women, and Hispanic men.


Affirmative action was introduced into the American workplace in 1961 via an Executive Order signed by President John F. Kennedy. While it was initially created to ensure that federal employment and employee treatment did not vary due to characteristics such as race and national origin, it was expanded in 1965 to include gender as well. The original law applied to federal jobs, but 38 U.S. states eventually passed state-level affirmative action laws in an attempt to address discrimination and avoid litigation under federal equal employment laws. Arguments against affirmative action, saying such programs no longer benefit minorities and women, have gained traction in recent years. Since 1996, several states have begun to reverse state-level affirmative action laws, including California, Washington, Michigan, Nebraska, Arizona, New Hampshire, and Oklahoma. As of 2015, similar legislation is under discussion in other states.

This study uniquely uses these law reversals as a natural experiment to examine the impact of affirmative action bans on the workforce participation of minorities and women in state and local public institutions. Using the states with bans as the experimental group, the author compared them to states that still have affirmative action programs in order to determine whether ending such programs has an effect on minorities’ and women’s employment.


Ending affirmative action led to a significant decrease in workplace diversity, compared to the states that kept affirmative action programs in place. There were sharp declines in Asian female, Black female and Hispanic male representation in states that banned affirmative action.

  • At baseline (measured as a pooled sample of states who had or have affirmative action programs) full-time public state-level employees were: 53% white men, 31% white women, 5% Black men, 4% Black women, 4% Hispanic men, 2% Hispanic women, 0.6% Asian or Pacific Islander men, 0.5% Asian or Pacific Islander women, 0.4% Native American or Alaskan Native men, and 0.2% Native American or Alaskan Native women.
  • Once affirmative action was repealed in a state (four states in the years ranging from 1996 to 2008 within the study period of 1990 to 2009), minorities working in state or local government decreased relative to the control group of states that kept affirmative action in place: Hispanic men’s participation decreased by 7%, Black women’s decreased by 4%, and Asian women’s decreased by 37%, (Although this figure for Asian women might seem disproportionately large, it is due to the fact that there were very few Asian women in the workforce in the first place, so any change would produce a large effect.)
  • For Black women, this percent decline grew larger over time (up until five years after the ban). For Hispanic men, the decline continued steady through the third year. For Asian women, the decline was limited to the first year following the ban.
  • White men, however, experienced a 4.7% increase in employment following the ban (limited to the subsample of states that ever experienced a ban).

In sum, there was a significant loss of workplace diversity once affirmative action programs were repealed, and this loss specifically affected Asian women, black women and Hispanic men. These findings contradict those who argue that affirmative action policies should be eliminated because affirmative action has run its course and no longer benefits minorities and women.


The repeal of affirmative action laws in four states (California, Michigan, Nebraska and Washington) presented the opportunity for a natural experiment. The bans were an exogenous (or external) shock that allowed the author to study the before and after share of minorities and women in the workforce of state and local government agencies where the bans went into effect, compared to the before and after shares of minority employment in places where affirmative action remained in place.

The author used agency-level employment data from the U.S. Equal Employment Opportunity Commission (titled the EEO-4 State and Local Government Information Reports) from 1990 to 2009. The EEO-4 reports provided public employee counts of men and women by salary bracket in five categories: white, Black, Hispanic, Asian or Pacific Islander, and American Indian or Alaskan Native. The four states with the ban served as the experimental group, while the states that maintained affirmative action served as the control group. The author used difference-in-differences regressions to conduct the analysis. In total, the analysis sample includes 5,977 agencies.

Related GAP Studies