Gender Inequality's New Face
The cyclical plight of unemployed men, and the return of gender segregation by sectorRecently, I've been thinking about this idea of the "mancession" -- the disproportionate job losses on the part of men during the 2008 recession. It is not obvious that one gender's unemployment rate should be significantly more cyclical than that of the other gender.
Yet I've taken the last few days to do significant research, analyzing this intersection of gender and economics in different ways. This post is the product of that research and analysis. Since there are a couple threads, let me give you the "too long, didn't read" summary right here, and then show you the data:
- Male unemployment rates have displayed stronger cyclical behavior than female unemployment rates since the 1980 recession, with the divergence between rates hitting at least 1 percent in every recession since.
- The 2008 recession was the worst yet for gender employment equity, with the male unemployment rate rising 2.6 percent above that of women. It has since returned to par.
- This cyclicality of the gender unemployment gap is not adequately explained by the claim that "male sectors" like construction were hit disproportionately hard by the recessions. The timing of the gap doesn't support that thesis, and furthermore, the disparity appears within sectors.
- Disparities between sectors, however, are returning, reversing the progress made in the 1970s and 80s -- that is, "male sectors" are more male, "female sectors" are more female, and there are fewer gender-equal sectors.
- This wave of gender segregation in the workplace is the product of two trends: (1) the completion of female integration into the workforce, (2) the exit of females from "male sectors."
- The combination of relative procyclicality in male unemployment and increasing concentration of men in a limited number of sectors is worrying, and suggestive of the declining rank of men in their places of work and a gender productivity gap.
As you can see, from 1960 to 1980 women were substantially more likely to be unemployed than men. Now, remembering the Bureau of Labor Statistics' definition of unemployment -- having actively searched for work in the last month -- I think we can pretty easily explain this with some mix of employer gender bias, lower social pressure for women to find jobs, lower human capital and employment experience, less education, etc. Notice that for the 15 years up until 1980, the gap is relatively constant, at 2 percent -- there's no cyclical behavior.
And then something pretty dramatic happens between 1979 and 1983. We went from the female unemployment rate being 2 percent higher than for men, to 1 percent lower, seemingly in the midst of the 1980 recession. The data makes this seem like a sharp, secular trend. Only a year later, in the double dip of the Volcker recession, we see the first cyclical swing, as the rate stabilizes in the mid-1980s at par between the male and female unemployment rates.
In every recession since then -- 1990-1991, 2001, and 2007-2009 -- the male-female unemployment rate gap has been cyclical. The swings are usually 1 percent in size. Another way to think about this is that the male unemployment rate, in recessions since 1980, tends to swing upwards 1 percent more than does the female unemployment rate.As we can see at the right end of the graph, the 2008 recession saw much stronger gender-biased unemployment rate changes, with the male unemployment rate 2.6 percent higher than the female rate in May 2009. Let's zoom in on the most recent recession, in hope that we can discern a bit more closely what is going on:The primary conclusion I can draw from this graph is that the gap appeared almost entirely during the sharpest period of real output declines -- the autumn of 2008 and winter of 2009. This strikes me as important because one of the most common explanations for the "mancession" that gets batted around was the collapse of the housing market, and with it, presumably, mass unemployment in the construction sector. The gap we see above doesn't support this story -- the men behind the "mancession" came from the full range of sectors which cut jobs in the panic, not just construction, which is predominately male. The timing, in other words, doesn't work to be explained away by factors unique to this recession. That should make us worry about next time. (If you're curious, the undoing of the cyclical swing came courtesy of the durable goods and manufacturing sectors, which have hired back a lot of men since 2010.)
Besides the timing, there's another sign in the data which should tell us that we cannot explain the "mancession" away with the different impacts of the recession on employment on "male sectors," like construction. Within any sector, male- or female-dominated, men were far more likely to be fired or laid off than were women. We can see this by looking at the women-to-all-employees ratio by sector. You see, almost without exception, a shift towards women, sometimes sharp, other times more subtle or seen in the pause of a secular trend. Below you can see what happened in housing, but if you want to look more broadly at how the ratios of other behaved, see here. I'm not denying that men happen to be more employed in cyclically volatile sectors than are women, and that certainly has an impact (see here for more on this). I'm saying it's not the whole story. The disturbing fact is that some of the disproportionate increase in male unemployment is explained by higher levels of layoffs and fires than women within any given sector.
This should raise some concerns for the economist interested in gender. It suggests that, if we were to take an average firm and line up the employees in order of most to least valuable, or least to most likely to be laid off/fired, there'd be a lot of men at the end of the line. I don't do sociology, but it's pretty obvious to me that a disparate impact on men can create some social problems -- strains on families, strains on the social model itself of the nuclear family with the male breadwinner/head-of-household, crime and violent crime in particular...you get my drift.
Now I want to make a sharp tack to another side of gender-related developments in economics. I think it's well known that in the 1970s and 1980s, American women entered the workforce in significant numbers. Both those traditionally male-dominated sectors, and those which had been favorable to women in the past (retail, health and education, etc.), saw the fraction of women out of all employees rise. In these "male sectors," which I've marked with an "M" on the graph below, female participation went from 32 percent in 1972 to 41 percent in 1989; in "female sectors," female participation in 1972 was 48 percent, and in 1995 it neared 60 percent.Another way of looking at this is asking the question: how far is the average sector from gender parity? In 1972, most sectors were significantly male-dominated, and the average difference from gender party was over 18 percent, due to the relative under-participation of women by that amount. As the 1970s and 1980s unfolded, this number shrunk consistently year after year, down to 13 percent by 1987, and this trend was driven by the industries which had been less male dominated shifting towards women, and the very male-dominated industries being less so. Both types of sectors were closer to 50-50 than they had been 15 years prior.
Then the sands began to shift. Female participation in "female sectors" flattened out, and female participation in "male sectors" entered into decline in the early 1990s. The average difference from gender parity in sectors began to creep up once more, and then accelerating as the "male sectors" began more male-dominated in the 2000s. By 2007, there was more gender segregation between sectors than in any year since 1982. During the subsequent recession, we can see the brief rise of female-to-all-employee ratios we talked about earlier, which explains an important amount about the "mancession." We can also observe, looking in the earlier graph, that this effect was more pronounced in "male" than in "female" sectors.The graph above gives us some more sobering information -- the average "male" sector is more male than the average "female" sector is female, and this is more true than at any time for which I can get a dataset. To put on my sociologist hat momentarily one more time, this really isn't a good thing at all. Most of the industries men are coming to dominate are, frankly, troubled in the United States, or have reduced the amount of hiring and growth expected for the foreseeable future -- durable and nondurable goods manufacturing, mining and logging, information, financial activities, brick-and-mortar retail their "domination" occurs because women are leaving the fields en masse, perhaps sensing that there's no future ahead for these fields. I want to leave you with one last thought about how severe the changes in employment have been by sector and gender. We're increasingly entering into a new and untested economic model. It's one in which men produce mainly goods, and women produce mainly services.