Higher Risk Aversion and Lower Overoptimism Among Female Business Majors Explain Key Features of Gender Pay Gap
By Emanne Khan
Although women have made remarkable strides in the professional world, the gender pay gap stubbornly persists.
On average, American women earned 84 cents for every dollar men earned in 2020, a disparity that has remained relatively consistent over the past 15 years, according to the Pew Research Center. Although a substantial body of research has attributed at least some of the gender pay gap to differences in men’s and women’s educational attainment and chosen occupations, few studies have traced gender disparities in pay to the initial job search process.
A new National Bureau of Economic Research working paper by Human Capital Initiative Associate Director Patricia Cortés, Jessica Pan, Laura Pilossoph and Basit Zafar aims to document novel facts about gender differences in the job search process by analyzing survey data from recent graduates of the Boston University Questrom School of Business. Based on the survey data, Cortés and coauthors find that women tend to accept job offers earlier than their male counterparts, and that the gender pay gap narrows as the job search process progresses. The authors link these findings to gender differences in behavior, in particular higher levels of risk aversion among women and higher levels of over-optimism among men.
Cortés and coauthors also propose a job search model that incorporates gender differences in risk aversion and overconfidence to explain features of the gender pay gap. They conclude that “Intuitively, higher levels of risk aversion for women lead them to have lower reservation wages, to start searching for jobs earlier and to accept jobs earlier. On the other hand, higher levels of optimism on the part of men increase their reservation wages, lead them to accept jobs later and make the gender gap in accepted earnings smaller over time as they learn.”
Two categories of Questrom students participated in the researchers’ surveys: recent alumni and current students. The recent alumni group consisted of approximately 1,000 members of the 2013-2017 graduating classes. The survey administered to this group included questions on participants’ demographic and academic backgrounds, salary and negotiation behavior. It also included questions about the timing of job offers and the types of offers that participants ultimately rejected during their initial search.
Cortés and coauthors write that the scope of the survey questions allowed them to “construct a detailed timeline of how the job search process unfolds for each individual in our sample in the months leading up to and after graduation.”
The retrospective survey administered to the group of alumni provided the majority of the empirical evidence that the researchers gathered on men’s and women’s job search behavior. In particular, alumni were asked to rate their risk preferences through the following questions: “How would you rate your willingness to take risks regarding financial matters?” and “How would you rate your willingness to take risks in daily activities?,” with willingness measured on a scale from 1 (not willing at all) to 7 (very willing). Participants’ risk preferences were calculated as the average of their responses to those two questions.
The authors write, “Consistent with the prior literature, women in our sample report significantly lower willingness to take risks in financial or daily matters relative to men.” Furthermore, “Despite having similar GPAs as men on average, women report significantly lower perceived relative ability, consistent with the previous literature documenting that men tend to be more (over)confident than women.”
In addition to surveying recent alumni about their job search behavior, Cortés and coauthors surveyed a cohort of then-current Questrom students in the 2018 and 2019 graduating classes about their earnings expectations and intended job search behavior. The prospective nature of the current student survey allowed the researchers to analyze differences in men’s and women’s expected versus realized earnings. The group of current students was surveyed three times in total, twice before graduation and once after, in order to capture men’s and women’s evolving job search expectations.
The first novel finding the authors derived from the survey data is a systematic gender difference in the timing of job acceptance among men and women in the sample. Figure 1 below plots the proportion of men versus women who accepted a job in the months leading up to and following graduation. Illustrative of the gender difference, 60 percent of women in the sample had accepted a job by graduation, compared to 52 percent of men.
Figure 1: Job Acceptance Timing, By Gender
Source: NBER, 2021.
The second novel finding the authors documented is a decline in the gender earnings gap over the job search period, depicted in Figure 2 below. In August of the senior year, the average salary of jobs accepted by male students in the sample was 17 percent higher than the average salary of jobs accepted by female students; however, by the October following graduation, the average salary for men was only 10 percent higher than the average salary for women. The researchers write, “This implies that relative to women, men who accept jobs early tend to accept jobs that offer higher pay and over the course of the job search period, men increasingly accept jobs that offer lower pay.”
Figure 2: Cumulative Mean Accepted Salary by Months Since Graduation and Gender
Source: NBER, 2021.
Taken together, these two novel findings indicate that in the initial job search process, men tend to accept jobs later than women, and are more likely to accept lower-paying jobs the longer the search lasts.
Using empirical data collected from the surveys, Cortés and coauthors aimed to create a model to account for gender differences in risk preferences and earnings expectations. This model was then used to predict the outcomes of a scenario in which males and females searched for their first post-graduation job while still in school.
Returning to previous theories that link the gender pay gap to men’s overconfidence, overconfidence stems from a lack of complete information, so the authors used the model to study what would happen if both men and women had “perfect” information about their job prospects. Under the perfect information condition, the model indicated that the gender gap in earnings would decrease by 25 percent. The model’s prediction suggests that in the ‘real world’ in which men and women have less than complete information during the job search, men’s over-confidence in their abilities and options is a significant contributing factor to the gender pay gap.
The authors also used the model to study what would happen if students were allowed to hold onto job offers for an additional month, limiting the practice known as “exploding offers.” They posit that limiting exploding offers minimizes the role that risk preferences play in the job search, as candidates are given more time to consider an offer without having to make a swift decision.
Similar to the perfect information condition, the model predicted a significant reduction in the gender pay gap under the no exploding offers condition, at 40 percent. This prediction reinforces the role of gender differences in risk preferences, with men being less risk averse on average than women.
The study reveals new facts about the job search behavior of male and female college graduates in the entry labor market. Not only did Cortés and coauthors demonstrate that women tend to accept jobs earlier than men and that the gender pay gap decreases the longer the job search progresses, they also showed that gender differences in confidence and risk aversion contribute significantly to early career earnings gaps among the highly-skilled.
Is there hope for ever reducing the gender pay gap? Drawing on the predictions of the model, the authors suggest policies to better inform students about the expected timing and distribution of offers during the job search and to allow students to hold onto job offers for longer.
While further work remains to be done on the factors that lead men and women to apply to specific jobs, the authors showed that the gender pay gap need not be a static feature of the professional world for years to come.Read the Working Paper