This is our 15th blog post for our Job Market Paper Series blog for 2024-2025.
Saloni Chopra is a PhD candidate in the Department of Agricultural and Resource Economics at the University of California, Davis. Her research interests lie in the field of applied microeconomics with a focus on education, labor and gender economics. You can find her JMP here.
Despite global efforts, gender disparities in science, technology, engineering, and mathematics (STEM) fields persist, with women’s representation among science graduates remaining stagnant at 35% globally over the past decade (Global Education Monitoring Report, 2024). However, this aggregate figure masks significant regional variations – from as low as 15% in Niger to over 50% in Tanzania within Sub-Saharan Africa. This continued under-representation, particularly amid rapid technological advancement, threatens to exacerbate both technical skill and wage gaps, potentially limiting women’s upward economic mobility. While most policy interventions target early education (middle and high school), relatively few studies have evaluated strategies implemented at the tertiary level.
Affirmative action (AA) is a widely adopted policy tool for promoting equity in higher education, traditionally targeting students based on race, ethnicity, and socioeconomic status. Given the consistent gender disparity, a natural question arises if gender-based AA could be used to address the gender gaps in STEM fields. Would it be able to close the gap? How would it impact the enrolled cohort’s composition? More importantly, how would these changes in composition affect students’ educational outcomes in higher education?
My job market paper (co-authored with Isaac Ahimbisibwe) examines these questions by analyzing a gender-based AA policy implemented at Uganda’s largest public university. We empirically investigate how increased gender diversity, achieved through AA, affects students’ higher education outcomes in STEM fields. The policy’s net impact on university academic performance could be ambiguous for two key reasons: First, if binding, gender quotas in STEM majors would admit more women at lower admission thresholds, potentially creating academic mismatch and affecting overall cohort performance. Second, existing research on gender peer effects suggests that both male and female students can benefit from improved gender balance and increased exposure to same-sex peers, particularly for women.
Policy, Context & Data
In 2020, Makerere University Kampala (MUK), Uganda’s oldest and largest university, implemented a gender quota policy in STEM programs that capped each gender’s admission at 60%. Given that men dominated most STEM programs, this policy effectively reserved spots for women in 65% of all STEM majors offered. Two key features of Uganda’s higher education system provide important context. First, students make a critical academic decision when transitioning from middle to high school by choosing between science and arts tracks. This choice, which involves selecting a specific combination of subjects such as mathematics, physics, chemistry, biology, and economics, shapes their academic trajectory and determines which university majors they can pursue. Second, the university admission process requires students to select their intended major during the application process, and once enrolled, they cannot switch to a different program. Admission decisions are based on three primary factors: students’ performance on A-level exams (national standardized tests taken at high school completion), the capacity of each program, and student’s stated major preferences.
For our study, we assembled a comprehensive novel dataset spanning 2015-2024, linking students’ admission records, high school performance, and university academic outcomes at MUK. To assess the policy’s impact, we employ a difference-in-differences (DID) framework, comparing outcomes before and after the policy implementation. Our treatment assignment is based on pre-policy gender distributions: STEM majors with women’s representation below 40% prior to 2020 constitute the treatment group, while majors that historically maintained gender balance (40-60% women) serve as our comparison group.
The Impact of the Policy on Access and Quality of the Cohort Enrolled
Our DID analysis at the major-year level reveals that the policy effectively increased women’s representation in male-dominated STEM majors. Figure 1 presents our key findings through two panels. The left panel displays event-study estimates showing that treated majors experienced a 9 percentage-point increase in women’s admission share relative to non-treated majors following policy implementation. The right panel demonstrates that this increase in admissions translated into corresponding gains in women’s enrollment in treated majors.
A student-level DID analysis reveals that improved gender balance in treated majors is accompanied by changes in the composition of the enrolled cohort when comparing the high school academic performance of men and women. Post-policy, women enrolled in treated majors had 0.2 grades lower than their counterparts in comparison groups, while men in treated majors had scored 0.5 grades higher than those in untreated majors. These A-level exam score patterns suggest the policy achieved greater women’s representation by simultaneously relaxing admission criteria for women while effectively raising standards for men.
Despite lower high school scores for women enrolled, the average university performance of the students in treated cohorts improved
To evaluate university-level academic performance, we examine three key measures: first-year grade point average (GPA), likelihood of failing a course, and first-semester dropout rates. The reduced form results for our outcomes of interest are presented in the event study graphs in Figure 2 below. On average a student in treated cohort had higher GPA and lowered probability of failing coursework in first year of classes post policy. We found no effects for dropping out after the first semester.
Our examination of treatment effect heterogeneity across gender, funding status, and program characteristics reveals several nuanced findings. Post-policy, students in treated majors showed significant academic improvements: men’s GPAs increased by 0.26 points and women’s by 0.10 points. Both genders experienced a substantial 35% reduction in course failure rates. These positive outcomes extended across funding categories, with both scholarship recipients and self-funded students demonstrating improved academic performance.
To assess the role of treatment intensity, we divided treated majors into two groups: those with pre-policy women’s admission rates of 30-40%, and those below 30%. The analysis revealed stronger positive effects in majors that were initially more male-dominated and consequently experienced larger increases in women’s enrollment post-policy. This suggests that the magnitude of gender balance improvement may have influenced the policy’s effectiveness.
Improved gender diversity has positive impact on both top performing men and women
To investigate whether improved gender balance drives our observed results, we focus on high-achieving students who would have gained admission regardless of the policy intervention. We establish this threshold using the first-year post-policy male admission cutoffs, which were consistently higher than women’s cutoffs across all majors. By restricting our sample to students who exceeded these cutoffs in all years, we identify a subset of “always-admitted” students whose enrollment was independent of the policy. We ran student-level regressions for high-performing men and women separately. Our results show that both genders in treated majors achieved higher GPAs compared to their counterparts in the comparison group after policy implementation. Notably, the positive effect on GPA was twice as large for women as for men, suggesting that increased gender balance may particularly benefit female students’ academic performance.
Key Takeaways
Our analysis of a unique gender-based affirmative action policy at Uganda’s largest public university yields several important conclusions:
- The policy effectively increased women’s representation in STEM programs, demonstrating that targeted interventions can successfully address gender disparities in traditionally male-dominated fields.
- Contrary to concerns about academic preparedness, students in treated majors exhibited improved academic performance overall.
- This improvement appears to stem from both changes in cohort composition and enhanced classroom gender balance.
- The benefits of increased gender diversity extended beyond the directly targeted population. Even high-ability students who would have gained admission without the policy, experienced positive impacts, with women experiencing double the gains in GPA.
- Finally, our findings suggest that affirmative action policies can serve as an effective correction mechanism when admission criteria rely on potentially noisy measures of ability, such as high school test scores.
Feature image shared by the author, taken from: https://law.mak.ac.ug/pre-entry-examinations-for-admission-to-bachelor-of-laws-2023-24-academic-year/
