In his State of the Union address, the president referenced the raw wage gap between women and men:
Today, women make up about half our workforce. But they still make 77 cents for every dollar a man earns. That is wrong, and in 2014, it’s an embarrassment. A woman deserves equal pay for equal work. She deserves to have a baby without sacrificing her job. A mother deserves a day off to care for a sick child or sick parent without running into hardship – and you know what, a father does, too.
First, when controlling for all the factors that influence pay, such as work experience, the number of hours worked per week, and so on, the raw gender pay gap almost disappears. The room in which legitimate gender discrimination can operate is a lot smaller than you’d think. The 77-cents-on-the-dollar claim has been far more thoroughly debunked than, say, doubts regarding global warming or the claim that lower tax rates increase tax revenues. …
Second, women tend to work in jobs that pay higher benefits relative to their salaries. Economists Eric Solberg and Teresa Laughlin of California State University found that factoring in benefits eliminated around nine percentage points of the raw pay gap, and concluded that “any measure of earnings that excludes fringe benefits may produce misleading results as to the existence, magnitude, consequence, and source of market discrimination.” …
Third, the profit motive works to reduce the pay gap. If Employer A pays women 77 cents on the dollar, Employer B can hire all Employer A’s female workers at 78 cents on the dollar to replace his costlier male employees. This raises Employer B’s profits, while Employer A must now pay full freight for his all-male workforce. …
And fourth, several studies have shown that as industries face increased competition through deregulation or international trade, the gender pay gap shrinks. And the pay gap is larger in monopoly markets without competition and smaller in start-ups and small businesses that must be productive in order to survive. So women need more markets, more enterprise, and more opportunity, not more regulation and litigation.
In “A Grand Gender Convergence,” the Harvard economist Claudia Goldin observes that the raw wage gap varies over the life course:
One way to see change in the earnings gender gap by age is to construct synthetic birth cohorts, as shown in Figure 1 a for college graduate men and women working full-time, full-year and in Figure 1b for college graduates with controls for hours, weeks, and further education. The data used are from the U.S. Census and the American Community Survey (ACS) for the years from 1970 to 2010.
The most obvious and important findings from these depictions are that each cohort has a higher ratio of female to male earnings than the preceding one and that the ratio is highest for younger individuals than it is for older individuals, at least up to some age. One part of the story of the preceding metaphorical chapters is that there have been large gains in the earnings of women relative to men. An important clue to what it will take to create gender equality in earnings is that something happens that decreases women’s earnings relative to those of men as they age.
Men and women begin their employment with earnings that are fairly similar, either for full-time year-round workers or for all workers with controls for hours and weeks. In the case of the latter group, relative earnings are in the 90 percent range for the most recent coh orts even without any other controls. But these ratios soon decline and in some cases plummet to below the 70 percent level.
Interestingly, in most cases the ratio increases again when individuals are in the forties (for the most recent of the cohorts to be old enough to be in that age bracket). Why it increases is beyond the scope of the present work. It would appear to be less a function of selection since in most cases the women who left would be drawn disproportionately from the lower part of the earnings distribution and those returning would presumably be the same individuals with less accumulated human capital. If anything, the function should increase and then decrease.
The main conclusion from the aggregate earnings gender gaps is that the difference in earnings by sex greatly increases during the first several decades of working life . That conclusion will be reaffirmed by the findings of studies of several highly specific occupations for which the training for both men and women is identical. The two degrees are MBA and JD. The data for these occupations, moreover, is longitudinal (or retrospective) thereby containing actual cohorts, not synthetic ones. In addition, the data contains detailed productivity-related characteristics.
Goldin finds that if our goal is to equalize earnings by gender (and she is sympathetic to that goal), it is far more important to equalize earnings within each occupation than to even out the proportions of women and men working in each occupation. So this raises the question of what is going on within occupations. She divides occupations into five sectors (Business, Health, Science, Technology, and Other), and finds that Business has the largest gender gaps in favor of men, Science and Technology have the smallest, and for the under-45 cohorts, Science and Technology actually have small gender gaps in favor of women.
So what exactly is going on in Business occupations? Goldin posits that Business occupations are particularly subject to a dynamic in which it is not just the number of hours worked that matter but rather the particular hours worked.
The employee who is around when others are as well may be rewarded more than the employee who leaves at 11am for two hour s but is hard at work for two additional hours in the evening. Even the self-employed may have nonlinear earnings because they cannot fully delegate responsibility.
And Business occupations are often occupations in which those who work twice as many hours earn more than twice as much pay:
In many workplaces employees meet with clients and accumulate knowledge about them. If an employee is unavailable and communicating the information to another employee is costl, the value of the individual to the firm will decline. Equivalently, employees often gain from interacting with each other in meetings or through random exchanges. If an employee is not around that individual will be excluded from the information conveyed during these interactions and has lower value unless the information can be fully transferred in a low cost manner.
The point is quite simple. Whenever an employee does not have a perfect substitute nonlinearities can arise. When there are perfect substitutes for particular workers and zero transactions costs, there is never a premium in earnings with respect to the number or the timing of hours. If there were perfect substitutes earnings would be linear with respect to hours. But if there are transactions costs that render workers imperfect substitutes for each other, there will be penalties fro m low hours depending on the value to the firm. A sparse framework will demonstrate these points and develop them further.
To eliminate within-occupation gender gaps in Business, we must somehow address the fact that firms don’t always operate in an environment in which every employee has a perfect substitute and transaction costs are zero. It is possible that technological advances will move us further in this direction. A combination of intelligent machines and advanced lifelogging devices might make it less costly to share knowledge within firms. But for the foreseeable future, it’s difficult to imagine that nonlinearities will vanish any time soon.
Barack Obama’s claim that the raw wage gap between women and men is an embarrassment is in part, and in a roundabout way, a claim that the existence of nonlinearities in knowledge-intensive work is an embarrassment. (This is particularly interesting because the president and his wife are Harvard Law School alums, and his wife chose to leave a large law firm, where these nonlinear dynamics are entrenched, for a general counsel role, a path that many highly-skilled women take in part to have more flexible schedules.) It also obscures the remarkable progress that has been made by women, as the wage gap between women and men working similar hours and with similar experience has largely vanished for younger cohorts.
We are thus left with the the latter half of President Obama’s statement on the gender gap, that “a woman deserves to have a baby without sacrificing her job.” It appears that once women become mothers, they find it far more difficult to work the kind of hours that many firms demand of employees reaching the highest ranks. This is a complicated issue. The fact that many women are now raising children alone makes it extremely difficult for them to share the burden of child-rearing, and it is not obvious that public policy can substitute for a second parent. Among women who co-parent, it is very common that women bear most of the burden of parenting even so. To some extent, this might reflect stereotyping and neglectful behavior on the part of men. The norms around fatherhood are changing, and it seems reasonable to suggest that they ought to change even more. But differences in time spent with children might also reflect a greater proclivity on the part of women to invest in parenting, even in the absence of external pressure. It strikes me as entirely reasonable to argue that public policy can and should be used to mitigate some of the tradeoffs involved, by making parenting more attractive. Yet it is not obvious that public policy can eliminate the tradeoffs, unless we start entering firmly illiberal territory.
In discussing the gender gap, the president has (a) oversimplified and misdiagnosed the underlying problem, (b) obscured the gains that have been made by women in the marketplace, and (c) made claims regarding what public policy can achieve that are unrealistic and potentially counterproductive.