By Teri Fritsma
PDF of article
Are men bearing the brunt of the economic downturn? This is part 1 of a two-part article. Look for the second part in the next issue of Trends.
Several recent articles in the popular press have suggested that men are taking it on the chin during this economic downturn. A recent New York Times article reported that more than four out of five jobs lost since 2007 were held by men. The executive editor of The Nation is blogging about the new gender gap in unemployment, and the Los Angeles Times dubbed this a “he-cession.” 
The current economic slump isn’t unique in this regard. Men's job losses have dramatically outpaced women's in each of the last five recessions. What’s behind this phenomenon? And do Minnesota patterns parallel the national trends?
The unemployment rate for men and women shows a clear disparity and suggests that men are far more vulnerable to layoffs these days both nationally and in Minnesota. The male and female unemployment rates have both increased over the past year, but while female unemployment rose fairly modestly, male unemployment nearly doubled (see Figure 1). Sex differences in unemployment are even more pronounced in Minnesota than they are nationally. Male unemployment in Minnesota rose by 5 percent, while female unemployment edged up by less than 0.5 percent over the past year.
|All figures are not seasonally adjusted. The U.S. figures come from Table A-1 of the Bureau of Labor Statistics Employment Situation News Releases for March 2008 and March 2009. Minnesota data are from the author’s analysis of the March 2008 and March 2009 Current Population Survey. The total Minnesota CPS sample sizes for March 2008 and March 2009 were 3,478 and 3,608 respectively, including respondents not in the labor force.
Why is the risk of unemployment so much greater for men? There are probably a combination of reasons, but most analysts and commentators point to one major factor: the employment patterns of men and women. While many men are employed in the hard-hit construction and manufacturing industries, women are more likely to work in the relatively stable government, health care and education services industries. How accurate is this claim and how much can these differences account for the sex gap in unemployment?
No Hard Hats for Women?
A large body of research is devoted to studying sex differences in industry or occupational employment. What this research makes clear is that women and men generally occupy different space in the labor market. And while men frequently benefit from these differences in the form of higher average pay, they also appear to be far more exposed to layoffs during economic downturns.
Figure 2 plots each Minnesota industry based on two characteristics: its share of unemployment insurance (UI) claims and its percentage of female employees. What we see here is that two male-dominated industries — construction and manufacturing — were responsible for about half of all unemployment claims in March 2009. Meanwhile, the two most female-dominated industries in Minnesota — health care and education services — together made up less than 5 percent of all UI claims. It’s worth noting that there were many industries that didn’t fit this pattern. Three heavily male-dominated industries — agriculture, mining, and transportation and utilities — were each responsible for less than 3 percent of unemployment claims, and two industries that are almost 50 percent female (wholesale and retail trade, and information) each accounted for more than 10 percent of UI claims.
Data sources: the share of state UI claimants (vertical axis) comes from the ETA 203, Characteristics of the Insured Unemployed. These data are not seasonally adjusted. The percentage of women in industries (horizontal axis) comes from the state Quarterly Workforce Indicators, derived from state administrative records and basic demographic information from the U.S. Census Bureau.
The role of industry sex composition
Suppose you could take all people in the labor market and randomly assign them to industries so that women were just as likely as men to be employed in the hard-hit construction and manufacturing industries. Then imagine you computed a new, hypothetical unemployment rate for women. The difference between that hypothetical rate and the true rate is the share of unemployment that can be attributed to the different industry employment patterns. For example, suppose the true unemployment rate for women and men is 6.5 and 9 percent, respectively. Then suppose the hypothetical rate for women is 8.5 percent. This means that the different employment patterns can account for 2 percentage points, or about 80 percent (2/2.5) of the total gap in unemployment.
As it turns out, this type of “what if” analysis is possible with specialized statistical techniques developed by sociologists studying the labor market (see sidebar). We used this methodological approach for the following analyses to better understand the relationship between men's and women’s industry composition and the risk of unemployment. Figures 3 and 4 show the results of this analysis for the U.S. and Minnesota, respectively.
Figure 3 shows the true U.S. unemployment rate for men and women from March 2008 through April 2009, as well as the hypothetical unemployment rate for women that would result if they were as likely as men to be employed in the male-dominated sectors of the economy. The data document, again, that the true unemployment rates for men and women have both risen, with the male rate rising more dramatically than the female rate. However, the hypothetical rate for women from March through September 2008 actually exceeds the true male rate — meaning that if women and men were equally likely to be employed in the hard-hat industries like construction and manufacturing, women’s unemployment would actually have been higher than men’s by a few percentage points during those months.
Beginning in December 2008, and continuing through April 2009, the true male and hypothetical female unemployment rates merge — meaning that men and women’s different employment patterns currently account for the entire gap in unemployment. That is, if we could rearrange people so that women and men were employed in the same industries, their unemployment rates would be almost identical.
Figure 4 shows comparable data at the state level. Again, we see that while the male rate has jumped to 10.4 percent since September 2008, the female unemployment rate has barely moved. Furthermore, the hypothetical female rate is consistently about 1 to 2 percent higher than the true female rate. In April 2009, the different employment patterns of men and women in Minnesota accounted for 2.2 percentage points, or just over 40 percent, of the total gap in unemployment. That is, if we rearranged men and women in the labor market so that women’s concentration in construction and manufacturing was the same as men’s, the male/female unemployment gap would be 40 percent smaller in Minnesota.
Conclusion: Stay Tuned
The analysis above suggests that there are some substantial differences between Minnesota and U.S. patterns. The first difference is the female unemployment rate: While the national female unemployment rate has risen (albeit modestly, compared to men’s) the Minnesota female unemployment rate has stayed more or less constant over the last 12 months.
Secondly, the industry employment concentration accounts for essentially the entire unemployment gap at the national level, but it explains only 40 percent of the unemployment gap at the state level.
What can account for these differences? What is unique about Minnesota? Time and space limitations preclude us from answering these questions in this article, but stay tuned for Part 2 in the December issue of Trends, in which we delve deeper into this topic and uncover some explanations for these differences.
Who Has It Better?
What’s clear from this study is that Minnesota men are indeed taking the lion’s share of job losses in the recession, primarily because men and women are in different industry concentrations.
What the analysis can’t answer, however, is the more subjective question of which sex has it “better” or “easier” these days. It might be tempting to assume from the patterns of job loss that men alone are affected by this recession. Before jumping to that conclusion, however, consider the following:
First, when any member of a household is laid off, the whole household is impacted — and the impact may be greater when the salary lost is the man’s. Working women still earn less than men. According to the U.S. Bureau of Labor Statistics, men out-earned women in every age, race, major occupational group and state in 2007. (In Minnesota, the average woman’s earnings were 76.9 percent of the average man’s.) This means that making ends meet on one woman’s salary is likely to prove more challenging for a household or family accustomed to relying on both a male and female salary.
Second, non-working men and women tend to spend their time in different ways. According to data from the 2003-2006 American Time Use Survey, married fathers who were not employed spent an average of 2.32 hours on housework and 1.26 hours on child care each day. Married mothers who were not employed spent nearly double the time on these same activities: 3.64 hours on housework and 2.48 hours on child care. And full-time working moms? They spent almost as much time as non-employed dads on household and child care activities (2.05 and 1.22 hours, respectively).
One can always find exceptions to these broad trends — and it’s likely that men’s and women’s salaries and time-use will continue to edge toward parity. Still, it’s worth keeping in mind that the women who remain in the labor market during this recession are, on average, working for less money and putting in nearly the same hours at home as their unemployed male counterparts.
The analysis in this article uses a statistical technique known as purging models, described by Clifford Clogg and Scott Eliason in their 1988 article in American Sociological Review entitled “A Flexible Procedure for Adjusting Rates and Proportions, Including Statistical Models for Group Comparisons” and used by Clogg, Eliason and Kevin Leicht in their book “Analyzing the Labor Force: Concepts, Measures and Trends.”
Certain types of purging models (specifically, “marginal composition-group” models) permit the analyst to statistically create a counterfactual scenario in which two independent variables are uncorrelated with each other — in this case, those variables are a person’s sex and the industry where he or she works. This mimics random assignment in experimental design.
The models presented here were estimated in the Categorical Data Analysis Software (CDAS), a specialized statistical package developed by Eliason. Full methodological details are available from the author at email@example.com .
Rampell, Catherine. “As Layoffs Surge, Women May Pass Men in Job Force.” New York Times, Feb. 6, 2009.
Reed, Betsy. “Unemployment: A New Boys’ Club?” April 6, 2009.
Burns, Greg. “Recession Hits Male Workers More.” L.A. Times, April 3, 2009.
Goodman, William, Antczak, Stephen, and Freeman, Laura. “Women and Jobs in Recessions: 1969-1992.” Monthly Labor Review, July 1993.