In my last post, I began a series focused on the gender wage gap, and discussed why demonstrating its existence is not always a simple proposition. Most importantly, I argued, it is not enough to simply look at the difference between men’s and women’s median wages. Not only do these estimates tell us less than we might like (recall those overlapping margins of error), but their analysis also often ends up oversimplifying the conversation by providing only a comparison between all working men to all working women—not, ultimately, the comparison we really want to make.
This is because human capital—the abilities people bring with them into the workplace—and wage structure—the amount we compensate various jobs relative to others—affect how much we can reasonably expect people to be paid, as do some outside factors, such as when and where a worker is hired. In other words, there are plenty of reasons other than gender that could explain why two individuals working full time earn different amounts of money. In order to investigate the extent to which a worker’s gender influences his or her wages, we need to make sure we’re comparing workers who are similar in as many other ways as possible.
As we try to establish whether men systematically out-earn women and, if so, whether this is a direct result of gender alone, we need to first come up with some reasonable explanation for how wages are determined—setting aside whether we think that this is how they should be determined—and then establish whether gender has anything to do with it. We do that by first identifying a set of criteria—including gender—that might explain how a person comes to earn a particular wage, and then assessing the relative importance of each item on this list. We can also try to measure how good a job we have done at capturing all of the components that go into wage setting.
Measuring human capital
Most of us walk around with a pretty good idea of the curve we’re going to be graded against as we apply for jobs, raises or promotions. For example, credentials, such as degrees or certifications, matter for some positions, and year of experience can be important in determining wages. These qualifications—educational attainment and years of experience—are measurements that seek to quantify workers’ abilities. We assume that higher educational attainment and more years of experience result in better skills, which are more highly valued (and therefore, more highly compensated) than, you know, worse skills. We should start our evaluation by comparing workers who demonstrate the same abilities, to the extent that we can identify, quantify, and measure them.
And, though this is a topic for another post, we certainly want to account for differences in demographics, such as age and race. Other worker-specific traits we might want to consider include whether a worker is part of a trade union, as membership in a collective bargaining unit will influence his or her wages, as well.
Accounting for wage structure
Better skills, that essential piece of human capital, can only be understood in context. How can we compare the abilities of middle school teachers to those of accountants, occupations which require similar levels of education? What an employer looks for in someone leading a classroom of 12-year-olds may not be what an employer would prefer in someone charged with keeping a company’s finances in order. Moreover, industries have relatively different statuses—due, for example, to their perceived value to society or the demand for skilled labor—and workers across these industries may be rewarded on noticeably different pay scales. For example, expected starting salaries for registered nurses working in a hospital may look quite different from those for software engineers working at a start-up, even though, again, individuals filling these roles may have attained similar levels of education.
Ideally, as we search for reasons that people are paid different amounts, we’d only be comparing individuals that have the same job descriptions. This isn’t always possible, but we can compare those who work in the same industry. This gets dicey when an industry sector employs both administrative assistants and CEOs so, to the extent that it’s possible, we should also try to compare workers with similar titles or ranks. Geography will also influence compensation, as local cost of living bears on the wages that employers are willing to offer (and that employees are willing to accept).
Where does all this comparison get us?
By limiting our comparison to workers who look similar on as many of these dimensions as we are able to quantify, except for gender, we get closer to identifying whether gender plays a role in determining wages. But it is actually almost impossible to compare two workers who have identical qualifications and situations. So, once we identify account for all things that could possibly be different between workers and that could reasonably influence how much these workers get paid—such as age, race, education, occupation, years in current job, labor union membership, geography, full/part-time work and, of course gender—we look for data sets that reflect as many of these factors as possible, and analyze this data in such a way that it accounts for the differences.
In other words, as long as we can identify and collect sufficient information on workers’ differences, we can actually compare workers who are not similar by artificially holding factors, other than gender, constant. And it’s this analysis that starts move us along in our investigation. In my next post, I’m going to take a look at some studies that do a good job with this variety of analysis—and introduce some things other than wage that we might consider when assessing men’s and women’s experiences in the workplace.
But, before you go, here’s important point to keep in mind: Even once we conduct the analysis I’ve discussed here, we are not home free. First, there are almost certainly factors influencing worker’s wages that cannot be measured easily, if at all—factors such as the strength of an individual’s professional network. In addition, some things that matter in establishing a wage may be related to gender, to start. This makes it difficult, or even facetious, to attempt to hold certain things constant between men and women. It’s possible, for instance, that men and women’s ages have disproportionate effect on their wages, or that wage-negotiation willingness, skills, or outcomes can be related to gender. For some discussion on these more complicated factors, don’t worry: three posts on the gender wage gap were never going to be enough, anyway.