This is the second in this year’s series of posts by PhD students on the job market.
Despite continuous efforts to promote gender equality, women’s representation in STEM jobs remains low worldwide at only 29% (Gender Gap Report 2023). The gender gap in high-paying STEM careers contributes to the overall gender wage gap (Blau & Kahn 2017). Educational institutions often implement affirmative action (AA) to address these disparities and to increase minority representation in both education and the labor force. While such policies can improve diversity and outcomes of the beneficiaries (Bagde et al 2016; Khanna 2020), they may also reinforce negative stereotypes (Coate & Loury 1993). Specifically, gender-based AA policies that lower admission standards at top STEM colleges to boost female representation may shift firms' perceptions of women’s average ability, potentially increasing statistical discrimination by gender and widening labor market gender gaps.
My job market paper focuses on studying the implications of one such policy that was introduced in India’s top engineering colleges, the Indian Institutes of Technology (IITs). The IITs are highly selective and prestigious institutions, with many CEOs of top U.S. companies, such as Google, Deloitte, and IBM, among their alumni. The policy, introduced in 2018, created additional seats that were reserved specifically for women. By design, therefore, these additional seats did not displace any students already gaining admission. Prior to 2018, IITs admitted about 8.9% girls in a cohort of 16,000 students. After the policy was implemented, the percentage of women enrolling at IITs nearly doubled to an average of 15.7% in each cohort entering between 2018 and 2021 (Gupta 2023), an increase of approximately 1,300 females per cohort on average, based on the admissions data available in IIT Annual Reports.
In a world where ability is imperfectly observed by employers and colleges act as a signal, such a policy can have ambiguous effects on women’s employment outcomes – both in the targeted (IITs), as well as non-targeted (non-IITs) colleges. On the one hand, allowing women access to “elite” colleges could close the gender gap as firms view top colleges as signaling high ability. However, the impact could also be negative if firms perceive AA policies to lower the average ability of female graduates in these colleges. There could also be an indirect effect in non-IITs, ranked slightly below IITs, as the policy could alter the ability distribution of women within these colleges as well.[1] For example, if such policies attract the smartest women towards IITs who would have otherwise enrolled in elite non-IITs, firms could update beliefs regarding the average ability of women in non-IITs as well. Due to these opposing forces, whether such AA policies can reduce gender gaps in STEM employment remains an empirical question.
Experiment: Policy exposure within correspondence study
In line with the literature, I design and implement a correspondence study with potential employers who recruit from IITs to study the impact of the AA policy on the gender gap in job callback rates. Specifically, I create resumes that randomize the identity of the applicant along three parameters: gender, year of college entry (to signal policy exposure: pre-policy years: 2016 and 2017 and post-policy year: 2018), and college (IITs or non-IITs).[2] This generates 8 comparison groups.[3] I targeted my correspondence study to firms which have hired IIT candidates in recent years and identified them based on the placement information available on IIT websites. Qualitative interviews with hiring managers at some of these firms during their recruitment drives at IITs confirm their awareness of the policy and its year of implementation. I identified jobs posted on the website of each of these firms and submitted fictitious resumes to 616 entry-level engineering job postings, completing a total of 5,236 applications. Each job posting received 8 resumes, covering all possible combinations of the three parameters.[4]
Who gains and who loses?
The overall callback rate from the study is 3.4%. For the pre-policy cohorts (before 2018), I do not find any gender differences in callback rates at both IITs and non-IITs. To examine the impact of the policy, I then implement a difference-in-differences estimation strategy, where I examine the change in the callback rates between men and women, before and after the policy. I do this exercise separately for IITs and non-IITs.
I find no change in the male-female callback gap within IITs after the policy. This indicates that the policy did not negatively affect female callbacks at these top institutions (in contrast with what some theoretical models of negative stereotypes would have predicted). However, I find the emergence of a substantial gender gap in callback rates in non-IITs after the policy. Specifically, the probability of non-IIT females receiving a callback declined by 52%, relative to that of the males, widening the male-female callback gap by 2 percentage points at non-IITs.
Using a theoretical framework, I illustrate that the policy enables IITs to draw high-ability females within the supernumerary seats, who would have otherwise gone to the non-IITs. Qualitative interviews with hiring managers reinforce this perception. Evidence from experimental literature has similarly shown that adding gender-based AA into a tournament can induce more high-ability women to enter the competition (Niederle, Segal, and Vesterlund 2013). Consequently, the average ability of IIT females only falls marginally and still remains high because of which their callback rates are unaffected. In contrast, the average ability of women who now remain in non-IITs after the policy is lower than before and negatively impacts their callbacks.
The aggregate callback rate for female students (IIT and non-IIT combined) does not change significantly after the policy. This is because the decline in callbacks of non-IIT females is compensated by the increased number of callbacks to IIT females. Firms are substituting the hiring of non-IIT women they would have recruited in the absence of the policy with a larger number of IIT women, leaving the overall gender callback gap unchanged. I rule out alternative explanations, such as potential differences in how firms value experience between men and women, or between IIT and non-IIT women, as I find no significant differences in callbacks between the two pre-policy cohorts.
Effect on employment
A key limitation in most of the literature on correspondence studies is that the evidence relies on callbacks, which is just the first stage in the hiring process. Overcoming this limitation is challenging due to the lack of data on firm-level hiring outcomes. I exploit the fact that most individuals in my context are extremely active on a large professional networking platform – LinkedIn. I create a novel dataset on employment outcomes by scraping LinkedIn profiles of 6,980 engineering graduates from six Indian colleges (3 IITs and 3 Non-IITs), covering 45% of the total students entering these six colleges between 2016 and 2019. I find that in the post-policy cohorts, women from non-IIT colleges (compared to their male peers) are 14% less likely to be employed at top elite firms (i.e., firms which have recruited IIT students in the past five years), whereas there is no significant change in this likelihood for IIT females.[5]
What do we learn?
The findings on the IITs suggest that AA at top colleges may not harm the beneficiaries and impact discrimination at these institutions. However, the distributional consequences and spillovers of these policies are often overlooked. While these policies make elite colleges more accessible to the marginalized group, students at non-targeted lower-ranked institutions may face negative spillovers and higher labor market search costs. If graduating from lower-ranked STEM colleges is perceived negatively by employers, the policy may harm the non-beneficiaries of the protected group, undermining positive effects on direct beneficiaries at top-colleges, if any.
That said, the long-run effects of the policy may be different as the pool of engineering candidates respond to the policy. This paper is limited in its ability to assess impacts on wages and productivity, which are crucial for understanding overall welfare. Graduating from elite institutions could have positive signaling and network effects on job promotions, migration, and marriage market outcomes, for instance, that have yet to be explored.
Ritika Gupta is a PhD Candidate at the Department of Economics, University of Virginia. Her research interests are in Development, Labor, and Gender Economics.
[1] I rank colleges using the NIRF ranking data published by the Ministry of Education, Government of India.
[2] Specific IITs used in the CVs were IIT Delhi, IIT Kanpur, and IIT Indore, and the non-IITs used were BITS Pilani, IIIT Delhi, IIIT Hyderabad, NSUT, SRM, and VIT.
[3] These comparison groups are: IIT Female Pre, IIT Male Pre, Non-IIT Female Pre, Non-IIT Male Pre, IIT Female Post, IIT Male Post, Non-IIT Female Post, Non-IIT Male Post.
[4] The study was conducted in two waves and the 2016 pre-policy year was added in the second wave. In wave 2, I sent both pre-policy year resumes and therefore total resumes to each job range between 8-12. My analysis controls for wave fixed effects and total applications sent to each job.
[5] There were a total of 1,800 unique companies where the individuals I scraped on LinkedIn were employed in as their first job, out of which 330 companies are the elite firms which hire from IITs.
Join the Conversation