Understanding the Gender Salary Gap: A Cohort-Based Longitudinal Study
By Gail Eisenberg, lecturer in business administration, and Sam Laposata, visiting professor of economics—both at Muhlenberg College

In 1999, one of us conducted a survey of accounting, business, and economics (ABE) alumni who had graduated from a small liberal arts college ten years earlier. The survey asked alumni about their first professional jobs after graduation and about their current employment. This cross-sectional study revealed that men and women earned similar salaries in their first professional jobs after graduation. However, ten years later, men had a considerably higher median salary than women.

These findings confirm what other researchers using similar methodologies have discovered: within a single cohort, the ratio of women’s to men’s salaries deteriorates as the cohort ages. But the findings also left us with several questions. How quickly does the ratio of women’s to men’s salaries deteriorate? What underlying factors cause the deterioration? Are women’s salaries lower because women step out of the labor force to attend to family obligations and return at lower wages later in life, as many researchers assume?

To begin answering these questions, we conducted a five-year longitudinal study of the ABE graduates of the class of 2000 at the same college. By collecting data from the same people annually over five years, we observed how quickly the gender salary ratio deteriorated within the cohort and identified factors that likely contributed to its decline. Although our findings are generalizable only to a very small population, they suggest ways to understand the gender wage gap in the broader population.

Methodology

One of the authors’ marketing research classes began conducting the longitudinal study in 2001. Students enrolled in the course in subsequent years collected data in 2002, 2004, and 2005. Study data was based on a survey with forty questions in the following topical areas:

• Job status: Employment status, job title and description, annual salary, and number of hours worked per week
• Company information: Type of firm (business, government, nonprofit, or education), location of firm, and number of employees at firm
• Job environment: Amount of travel, frequency of relocation, presence and helpfulness of mentor, and gender of boss and of mentor
• Skills enhancement: Access to formal on-the-job training and graduate school attendance
• Attitudes: Attitudes regarding the importance of earning a high salary; the importance of balancing leisure, family, and work; and the perceived level of household responsibilities and chores (Attitude questions relied on a four-point Likert-type response scale.)
• Personal information: Gender, marital status, child status, race and ethnicity, and college GPA (collected from the registrar)

By studying a small homogeneous cohort, we controlled for many factors traditionally used to explain salary gaps, such as differences in age, experience, education, and racial, ethnic, or socioeconomic background. All eighty-five members of the cohort were traditionally aged college students with no prior professional work experience. All graduated from the same college with degrees in similar majors (accounting, business administration, or economics). Most are white and from middle-to-upper socioeconomic class backgrounds. This relative homogeneity allowed us to uncover other, less commonly discussed variables to explain salary differences.

Findings

In the first year of work, women’s median salary was 95 percent of men’s median salary. This ratio declined each year thereafter, falling to 62 percent in 2005.

In addition to having a lower median than that of their male counterparts, female salaries tended to cluster around the mean and have less spread (smaller standard deviation). These differences resulted in real disparities in men’s and women’s average earning power. From 2001-05, the Consumer Price Index increased about 10 percent. In this study, women’s median salary increased considerably more—about 30 percent—resulting in a 20 percent real increase. This seems like a significant gain, but it pales in comparison to the increase in men’s median salary during this time period, which was about 100 percent (corresponding to a 90 percent real increase).

Although this pattern is consistent with what other researchers have observed within cohorts, the salary ratio declined even more quickly than we expected. This may have been in part because we conducted our field work from 2001 to 2005, during the financial bubble, when salaries in the financial sector increased far above previous norms. In 2005, about 50 percent of the men in our cohort worked in finance-related jobs with very high salaries, while only 5 percent of women worked in finance-related jobs. During the years of the bubble, the differences between men’s and women’s salaries within this cohort may have been particularly exaggerated.

Significantly, the declining salary ratio within the cohort occurred even though no women had any children. Thus the main cause of the gender salary ratio decline within this cohort was not women stepping out of the labor force to care for children, as many researchers have contended. Nonetheless, women’s attitudes about work and family may have affected their careers in more nuanced ways, with women making early career and educational decisions that may negatively affect their salaries in anticipation that they will want family-friendly careers sometime in the future.

Within the cohort we studied, other measured variables may help explain the deteriorating ratio of women’s salaries to men’s:

1. Men reported working more hours per week than women.
2. Men had more formal on-the-job training than women.
3. Men relocated for work more often than women.
4. More men worked in the high-paying field of finance (50 percent of men compared to 5 percent of women in 2005).
5. Men more often had a mentor and found the mentor to be more useful than did women.
6. Men were more likely to work in locations with higher pay, such as New York City.
7. Men were more likely to work in the business sector than in government, education, or nonprofit organizations.

Implications

Economists tend to divide the causes of the gender salary gap into two broad categories: (1) the choices men and women make about their jobs and the human capital they accumulate, and (2) discrimination that occurs in the labor market.

In relation to the first category, our findings suggest that the men in our study accrued more human capital than women by working longer hours, having more on the-the-job training, relocating more often, and having more and better mentors. In addition, men made choices about jobs that led to higher salaries, including working in finance, working in or near New York City, and working for businesses rather than nonprofit organizations or the government. Thus choice and human capital accumulation did indeed seem to affect the gender salary ratio.

However, our findings may also suggest forms of discrimination. We found that women were not asked to relocate very often, had fewer formal on-the-job training opportunities, were less likely to have mentors, and didn’t find their mentors as helpful as did men. Thus women may not be offered the same opportunities as men. In addition, it was unclear to us whether women in the cohort sought jobs outside of the finance industry as a result of preference or because they perceived that the industry is inhospitable to them.

The gender wage gap is particularly troubling considering that other variables we measured are clearly inconsistent with men earning higher average salaries than women. Women had a higher average GPA than men (3.3 versus 3.1 on a 4-point scale), and more women had attended graduate school than men (twelve women versus five men).

Continued Inquiry

We recently conducted in-depth oral interviews with some participants in the cohort. Our analysis is still underway, but preliminary results suggest that eight years after college graduation, family issues are very important for both men and women. Most interviewees were either married, in a serious relationship, or considering marriage and family.

Still, notable differences exist. While male interviewees talked about working to build an economic foundation for their future families, women spoke of concerns about striking a family/work balance, sometimes even before they had children. Given these differences and the rational tendency to make decisions based on current and predicted conditions, we begin to understand why the gender salary gap set in so quickly.

Due to the small size of the specific cohort we used in this study, our ability to generalize our findings to other recent college graduates is limited. Nonetheless, our research demonstrates the value of cohort-based longitudinal studies in unraveling the complex reasons for the deteriorating gender salary ratio. A follow-up study with a large random sample of recent graduates from a variety of colleges, majors, and geographic locations might be of great value.