Gender gaps

Studies of male-female wage differentials tend to find that differences disappear once you control for standard things job type, experience, and time out of the workforce. To the extent that residual differences remain, they seem reasonably well explained by maternity risk among women of childbearing age: so long as firms bear some of the costs of an employee taking maternity leave, wages for those at higher risk of taking such leave will be lower. After controlling for everything for which we have data, some folks then interpret the gender coefficient as an upper bound measure of discrimination.

The National Post reports on a new study of the Canadian medical field finds that female physicians provide about 14% fewer hours of direct patient care as compared to their male colleagues: an average of 30 hours per week to mens' 35 hours. The report warns that increased proportions of female physicians at lower average work hours requires the hiring of greater numbers of physicians to cover the same number of patient care hours. Of course, the study's authors are now on the defensive:

Dr. Mark Baerlocher, the study's lead author, acknowledged he is tackling a thorny issue, but stressed he does not favour curbing the number of female physicians. Instead, the study calls for greater increases in medical-school enrolment to offset the phenomenon.

"It's not meant to be a negative paper in any way," he said in an interview. "It's meant to take an objective, hard look at the work-hour differences that most people would agree are very real.... You can't simply ignore it because it's a sensitive issue."
...
Dr. Baerlocher, a radiology resident at the University of Toronto, said he agrees women should not be blamed, lamented a general reluctance in the medical profession to examine controversial issues, such as gender differences and abortion.

"There are a lot of topics that aren't adequately studied, because it's deemed a socially sensitive topic."

It seems pretty important if you're trying to plan out future staffing requirements to know things like this.

From a broader perspective, here we have another source of unobserved heterogeneity that can give rise to average salary differentials: hours worked by salaried staff. In salaried professions where you don't punch a clock, hours worked can vary a fair bit across individuals; the harder workers are more likely to be promoted or given a raise. If hours worked vary by gender, then we expect average salary differences across genders. If we don't have observations on hours worked, average differences then show up in the coefficient on gender in a wage regression. One more reason that the gender coefficient is only an upper bound measure of discrimination rather than a good measure of overall discrimination.