Job losses among women during the recent recession exhibit a curious pattern by marital status that is revealing about the importance of labor demand and supply factors.
During the 2008-9 recession, job losses were not equitably shared; employment rates fell more for some groups than others. Not surprisingly, employment changes varied by industry, with the greatest percentage job losses in residential construction and large job losses in manufacturing.
It is also well known that job losses were greater among men than among women – the so-called mancession – largely because men had been more likely to work in the residential construction and manufacturing industries that were hit hardest.
In this way, changes in the patterns of demand help explain why employment changes were different for men than for women. People stopped buying new houses, cars and other items that were disproportionately produced by men, and many times continued buying health care and education that were disproportionately produced by women.
The chart below displays quarterly employment rates separately for married women and unmarried women who were heads of households, both under age 65. Not surprisingly, the latter are somewhat more likely to work because they have no spouse to help provide for the family.
More surprising is that employment rates fell so much more for these unmarried women who were heads of household. Employment per capita fell 4.7 percentage points among them, compared with 1.6 percentage points among married women. The job-loss gap associated with marital status turns out to be as large as the more widely recognized job loss gap associated with gender.
Neither group of women had many members working in construction, so the decline of construction cannot explain these differences. In fact, married women in 2007 were a bit more likely to be working in an industry that would decline thereafter.
An “added-worker effect” has been observed during a number of recessions: more married women worked during a recession than during an expansion because wives sometimes begin work to help replace the income lost by their unemployed husbands.
At first glance, this added-worker effect might seem to explain what is shown in the chart: perhaps married women were laid off at essentially the same rate as unmarried women, but offsetting those layoffs were the wives who started working when their husbands were laid off.
In this way, the added-worker hypothesis is a supply theory, one made possible by the changes in the composition of industry demand that created job losses for men.
But the added-worker effect is much too small to explain the sharply different job-loss rates by marital status. The employment rate among nonelderly married men fell 4 percentage points, to 83 percent from 87 percent. While that is a large decline by historical standards, it still means that roughly 95 percent of wives whose husbands were employed in 2007 had husbands who continued their employment during the recession.
Among the 5 percent of wives who were not so fortunate, roughly two-thirds of them had already been working before the recession and therefore could not react to their husband’s unemployment by starting work.
Estimates of the impact of the added-worker effect are that two or fewer wives who start working for every 100 husbands who become unemployed — and, by definition, none for every 100 husbands who continue to be employed.
The added-worker effect alone might explain why employment rates would fall a tenth of a percentage point more for unmarried women, but not the 3.1 percentage point gap that actually occurred.
Industry factors do not explain why the job market was different for married and unmarried women. Nor do age and schooling. Yet I think the relationship between marital status and job losses is revealing about other demand and supply factors, to which I will turn in next week’s post.