As I mentioned in my last post, there are a whole host of considerations to take into account when looking at men’s and women’s wages to investigate any gender-based differences. Sheer earnings numbers are meaningful—after all, a difference in earnings, no matter why it exists, means a difference in what men and women are able to afford to buy. However, when stated without context, these numbers invite debate. In order to gain a more thorough understanding of the difference between men’s and women’s earnings, one reasonable starting point is to control for those things that are widely understood to determine salaries, such as education level, industry of employment, or years of experience.
Studies that carefully examine job-related skill-based differences in earnings are not uncommon and are usually called human capital-based studies. Over the last several months, I’ve come across two of particular interest to me: a study conducted for the benefit of university administration at the University of Virginia, and a research letter regarding registered nurse salaries published in the Journal of the American Medical Association. These studies share the goal of describing men’s and women’s earnings while controlling for very specific wage determinants. Each shows a significant difference between men’s and women’s earnings, even after important characteristics have been held constant. Neither definitively states the precise cause of the difference, yet each advances our understanding about the wage gap.
University of Virginia’s Faculty Salary Study Task Force Report*
In 2014, University of Virginia administration commissioned a task force to examine through quantitative analysis the salaries of tenured and tenure-track faculty at the University of Virginia. The researchers identified factors likely to reasonably influence salary—in particular, school within the university, academic field, rank (assistant, associate, or full professor), years since highest degree, and years at the University—and used a statistical model to measure the relative importance of each, alongside demographic characteristics. This allowed the researchers to examine the influence of factors such as gender or race on salary.
While researchers were careful to caveat possible drivers of compensation that they were not able to address (such as teaching evaluations, research funding, or publication history), the study demonstrates that at the University of Virginia, the average female faculty salary is less than the average male faculty salary, even after adjusting for factors related to human capital. In 2013, the year in question, female faculty members earned an average of $3600, or 2.7 percent, less than male faculty members. Interestingly, there is only a small difference between male and female professors’ salaries at the assistant (or relatively new) level—and it’s slightly in favor of the women. However, as this implies, there is a meaningfully large difference in male and female faculty salaries at ranks higher than assistant (i.e., among those professors with longer professional careers and/or tenure).
This report is important for at least two reasons. First, because all the salary observations came from within a single university, the authors could not attribute differences in salaries to variation between institutions, such as in cost of living, culture, or competitiveness. Second, and relatedly, the outcome of the study was a set of targeted, University of Virginia-specific recommendations for addressing statistically significant differences in wages. That said, and as indicated by the Provost in his comments on the study, “…this finding is not unique to the University of Virginia or to higher education.”
Salary Differences Between Male and Female Registered Nurses in the United States
In another example, the Journal of the American Medical Association published a research letter earlier this year on observed differences between male and female wages in nursing. The report details remarkably similar findings from distinct analyses of two data sets—the American Community Survey and the National Sample Survey of Registered Nurses—and is, like the University of Virginia study, meaningful because of its specificity. An analysis of nursing occupations, alone, permits the authors to dismiss the hypothesis that differences between workers’ salaries are due to occupation. In other words: Sure, men and women may pursue different occupations, but we’re just talking about nursing (a female-dominated field) here.
Like the UVa study, the model used by these researchers allowed them to determine the relative effect of a wide variety of factors. In particular, the study separates out those things that understandably influence earnings (e.g., educational attainment, clinical specialty, or job description) in order to determine the relative influence of gender. The researchers made sure that salaries were comparable by adjusting for both inflation and location (i.e. they converted all salary amounts to their value in 2013, and accounted for differences in cost of living across the country by scaling by the Consumer Price Index).
These well-documented, well-executed examinations of men’s and women’s wages have several important functions. In particular, they
- Describe meaningful differences in wages that cannot be attributed to differences in experience, education, or other measurable indicators of expertise.
- Demonstrate that these differences may be found even within a single institution (i.e., at University of Virginia) or a single occupation (i.e. nursing).
- Suggest courses of action that may be taken to ameliorate such inequality (such as examining pay structures and institutional practices).
However, they do not and cannot explain the cause of these differences, for two reasons:
First, from the statistical rigor point of view, neither analysis is able to account for the influence of every single factor that goes into wage determination. Because this kind of analysis functions as a sort of process of elimination, this leaves in play the hypothesis that something else may be a more important driver of differences. Consequently, we cannot attribute the full gap to gender, alone.
Second, these studies offer no explanation of how gender actually influences earnings. As the UVa study discusses, several possibilities exist. Perhaps women are systematically disadvantaged; perhaps women actively make decisions that slow their earnings growth; perhaps men and women negotiate with different rates of success. Without knowing the exact driver, or drivers, of these differences, it is difficult to determine the specific policy that will change the status quo.
We neither should stop investigating the cause of these differences, nor our efforts to correct the imbalances. Sometimes, inequity serves as its own justification for policy change—and, at the very least, motivates me to keep looking for data, analyzing it carefully, and searching for a deeper understanding of what’s behind the gender wage gap.
*I call attention to this report not to criticize UVa for existing imbalances but, rather, to point out the important work that the administration commissioned—and the high-quality result.