There is good reason to believe that the pay gap between women and men is largely a function of different preferences (with respect to working hours, the tradeoff between wages and benefits, the work-related dangers posed by different occupations, and flexibility, among other things). But is this the end of the conversation or just the beginning? One could argue that the real problem we face is that women and men have such different preferences, as these preferences are shaped by social pressures that, for example, lead women to devote more time to raising children than men. Megan McArdle recently addressed this thesis in Bloomberg View, and now Evan Soltas, also writing in Bloomberg View, has done the same. While McArdle concludes (correctly, in my view) that there is relatively little the government can do to address the sources of the residual wage gap short of going “inside families, or the subconscious recesses of our minds.” Soltas, in contrast, believes that conservatives are choosing “to ignore lingering inequity,” and that even if they reject Democratic proposal to widen the scope of anti-discrimination lawsuits, they must “figure out an agenda to advance women in the workplace.”
Soltas contributes to the debate by observing that women are actually more underrepresented in certain industries, manufacturing and information, than they had been in the past. Intriguingly, he notes that while women represented 32 percent of the manufacturing workforce in 1990, their share had fallen to 27 percent as of last month, the lowest share since 1971. And in the information sector, the share of women in the workforce has fallen from 49 percent to 40 percent over the same interval. In Soltas’s view, this shift demonstrates that the notion of “men’s jobs” and “women’s jobs” has reasserted itself, and that the substitution of capital for labor in fields like manufacturing and information is “pushing out women more so than men.” (Soltas also briefly addresses inflexible working hours in high-end occupations, a subject we’ve discussed in this space.)
I was hoping that Soltas would offer further thoughts on why the substitution of capital for labor would push out women more than men, as it seems entirely plausible that it would have the opposite effect. The manufacturing sector placed a greater emphasis on physical strength in earlier eras than it does in its current, more heavily-automated incarnation, and you’d think that this would improve the relative position of women seeking employment in factories. So what could be going on?
Manufacturing and information (e.g., computer engineering, telecommunications, and traditional publishing) are both part of the tradable sector, and employment levels in the tradable sector have grown only modestly since 1990 while employment in the nontradable sector has increased considerably. Moreover, there has been a great deal of dislocation within the tradable sector as workers have shifted from manufacturing, which lost 6 million jobs from 2000 to 2009, to other sectors. It could be that women have chosen not to sort into manufacturing in larger numbers because they recognize that manufacturing employment is vulnerable to offshoring, not because of the reassertion of the social concept of men’s jobs and women’s jobs.
In “Is White the New Blue? The Impact of Gender Wage and Employment Differentials of Off-shoring of White-collar Jobs in the United States,” the economists Ebru Kongar and Mark Price found that after dividing white-collar service occupations between those at risk of being offshored and those that were not, low-wage women’s employment decreased in the at-risk occupations, which in turn led to an increase in the average wage of the women who remained. Kongar and Price were assessing the period from 1995 and 2005, and it is entirely possible that much has changed since then. But their findings point us in an interesting direction: it could be that women are more likely to remain in an at-risk sector if they are earning a higher wage, perhaps because it is the higher wage jobs that are less vulnerable to offshoring or automation, or because the higher wage compensates for the risk that the job might eventually vanish.
Could it be that it’s not so much sexism that accounts for the declining share of women in the manufacturing and information workforces but rather some difference in underlying risk preferences? In their 2011 paper on “The Evolving Structure of the American Economy and the Employment Challenge,” Michael Spence and Sandile Hlatshwayo divide the economy into tradable and nontradable sectors, and though they don’t explicitly address gender, it is noteworthy that of the large nontradable industries — government, health care, retail, accommodation/food service, and construction — construction is the only one where women don’t represent at least half of the workforce as of 2013 (and I’ll note that with the rise of modular construction, it’s not difficult to imagine that construction will at some point become a tradable sector); and that year, women represented 79.6 percent of the health care workforce.
Even if there is on average a female preference for stable employment in the nontradable sector over less-stable employment in the tradable sector, and I’m not at all certain that this is the case, one could argue that this risk-aversion reflects a legacy of discrimination. But it could also reflect good judgment. So while I am sure that there is lingering inequity in the labor market, it’s not clear to me that the declining female employment share in manufacturing and information are a clear instance of it.