≡ Menu

A Note on the Pay Gap

Earlier today I was talking with my GMU Econ colleague Tom Rustici about why the best empirical studies of men’s and women’s pay find that, although women get paid almost the same amount for the same work as men get paid, there remains a small difference – roughly five percent – between men’s and women’s pay.

Here are some possible explanations for that small gap (presented here in no particular order):

(1) The small gap reflects consumer or co-worker (or both) bias against women.  If enough consumers so prefer, say, to have their automobiles repaired by men rather than women that these consumers are willing to pay a premium simply to have such repairs done by men, then the pay of female auto mechanics will be lower than will the pay of equally productive auto mechanics.  Likewise, it’s possible that if enough workers (men or even women!) prefer for no reason other than a pure preference for men to have as co-workers doing job X men rather than women, the market wage for women doing job X will be lower than the market wage for men doing job X.

To the extent that such consumer or co-worker preferences are responsible for a real pay gap, legislative efforts to close this gap will likely make matters worse.  One obvious employer response, under such circumstances, to mandated equal pay will be to hire fewer women.

(2) The small gap is a statistical mirage; that is, it’s nonexistent.  Even the best empirical studies cannot control for – cannot see, measure, and properly account for – every factor that determines a worker’s worth to his or her employer.  Tom Rustici offered a personal example of one such factor that might help to create a statistical illusion in this setting.

Tom’s family owned supermarkets in Missouri (at which Tom once worked in a managerial capacity).  His company would offer workers the option of working fixed-time or flex-time.  As Tom explained, the former allows the employer to manage the worker’s time at work while the latter allows the worker to manage his or her own time at work.  Flex-time, in other words, is more valuable to the worker than is fixed-time, but flex-time is more costly to the employer than is fixed-time.  The reason is that a worker on flex time can pretty much come to work, and leave from work, as he or she chooses.  This arrangement is good for the worker but has obvious downsides to employers who must manage to ensure that the enterprise continues to operate smoothly over the course of the work day.

Tom reports that workers who chose flex-time were paid ten percent less per hour than workers who worked fixed-time.  He reports also that flex-time was used much more by women than by men.  An empirical study of Tom’s family’s supermarkets would therefore find that women are paid ten percent less for the “same” work as are men.  Of course, this finding would be mistaken, for the work is not the same.

A very careful empirical researcher can, in principle, discover this difference in available work options and adjust for it.  (Such a very careful researcher would find, therefore, that men and women employees with the same job title and years of work experience are paid, at these supermarkets, the same for the same work.)  But the question is: How able are empirical researchers to control for such nuanced and often practically invisible-to-the-observer differences in work options?  Surely researchers cannot learn of all such factors that result in economically legitimate differences in the pay of men and women.

Given this reality, there’s no particular reason to suppose that this ‘failure to observe’ all that is relevant is necessarily biased toward falsely finding a pay gap that favors men rather than biased toward falsely finding a pay gap that favors women.  But if the world contains (as it surely does) many such small – hence, practically unobservable – differences that cause different workers to be differently productive, and if women are generally even just a bit more likely than are men to discount the importance of work-in-the-market relative to work-at-home (or to being with family or friends), then this ‘failure to observe’ all that is relevant to setting worker pay will indeed result in an empirical finding of higher pay for men.

Comments