Closing the Gender Gap

By Teri Fritsma 
December 2009

PDF of article

In the second part of her story on the “he-cession,” writer Teri Fritsma finds that gender differences in unemployment in Minnesota have narrowed in the later stages of the recession.

In the last issue of Trends, we explored gender differences in unemployment during the current recession. Our findings suggested that men were far more likely to have lost their jobs than women and that gender differences in unemployment were more pronounced in Minnesota than nationally.

We also investigated the extent to which men’s risk of unemployment was due to where they worked. Our findings suggested that gender differences in industry employment could explain an estimated 40 percent of the gender gap in unemployment in Minnesota, while industry gender composition explained the entire gap at the national level.

This article picks up where the last one left off. Here, we investigate the following questions:

• As the recession has worn on, have the gender differences in unemployment continued?
• Do different industry employment patterns continue to affect men and women’s unemployment?
• What else, if anything, can explain the gender gap in unemployment in Minnesota?

New Trends

In the first part of this series, our analyses included data through April 2009, when the recession was in full swing. To present what happened more recently, Figure 1 shows the unemployment rates for men and women in Minnesota and the U.S. from May 2008 through August 2009 (the last month for which data were available at the time of publication). The additional months of data provide new insights into male and female unemployment trends during this recession.

First, although the Minnesota estimates are “noisier” or more volatile than the U.S. estimates, the state and national trends in unemployment have been, in general, fairly similar. The male unemployment rate initially rose much quicker than the female rate, increasing most dramatically in the early months of 2009.  Female unemployment, on the other hand, rose more gradually early in the year until the late spring and summer of 2009, when female unemployment increased fairly dramatically. From April through August, male unemployment more or less leveled off (at the national level) or declined (at the state level). Meanwhile, female unemployment rose (steadily at the national level and more abruptly at the state level).

The second noteworthy finding is that the gender gap in unemployment dramatically decreased over the last four months of the analysis, both nationally and in Minnesota, with the convergence primarily due to the recent rise in the female unemployment rate. As of August 2009, men remained at greater risk of unemployment, but given the gradual slimming of the unemployment gap, it is difficult to predict how much longer this disparity will continue.

To underscore this, Figure 2 shows what female unemployment rates over the same time period would have been if women and men were evenly distributed across industries in Minnesota, rather than the sexes being concentrated in different industries. (See Part 1 of this series in the September issue of Trends for more information on the methodology used to produce these estimates.)

From December 2008 through May 2009, Figure 2 shows that industry gender composition explained only a portion of the sex gap in unemployment in Minnesota. During this time, women’s insulation from unemployment went beyond just working in “recession-proof” industries. Beginning in June, however, that trend reversed and women abruptly became more vulnerable to unemployment, all else being equal. From June through August, women’s unemployment rates would have actually exceeded men were it not for women being concentrated in the more “recession-proof” industries.

Industry Employment and Seasonality

Recent trends suggest that the shrinking gender gap in unemployment might be explained primarily by industry employment and seasonality.  The drop in male unemployment over the summer months, for example, probably can be attributed, at least in part, to the rebounding construction and manufacturing sectors. In Minnesota, the construction industry added jobs each month from May through August, while manufacturing, which had been losing jobs earlier, held steady.

Conversely, the education and health sectors both fared relatively poorly during this period in Minnesota. Health care — one of the only industries to continue growing during this recession — remained flat during the summer months, and education lost about 15,000 jobs from April through August. An estimated 500 to 800 teachers lost their jobs at the end of the school year, which could help to explain the sharp bump in female unemployment between May and June in Minnesota. It seems clear that seasonality is at least partly responsible for the gap, but the question of whether the male/female differences in unemployment are purely seasonal remains.

Figures 3 and 4 address this question by showing male and female unemployment rates in two different industry sector groupings. It is important to note that for some industry/gender combinations, sample sizes are very small. To increase the sample sizes, we combined industries into three groups. Group 1 (shown in Figure 3) includes the most male-dominated industries: agriculture, forestry and fishing; mining; construction; manufacturing; and transportation and utilities.

Group 2 (shown in Figure 4) includes the relatively female-dominated industry education and health services. A third group, made up of the more gender-balanced industries of wholesale and retail trade; information; financial activities; professional and business services; leisure and hospitality; and other services is not presented. Unemployment generally has been more stable in these six industries over the first half of 2009, and there appears to be no distinct gender differences in unemployment throughout the period.

Figure 3 reveals pronounced gender differences in unemployment over this time period in male-dominated industries. Male unemployment was substantially higher in the first four months than in the second four months of the year. From January to April, the male unemployment rate hovered between 15 and 16 percent, but from May to August it dropped below 12 percent. 

Meanwhile, the female unemployment rate was lower in the winter months, but rose quickly and consistently from about 7.5 percent in February to about 17 percent in August. Within this group of industries, the number of unemployed men fell by nearly 34,000, while the number of unemployed women doubled, increasing by more than 17,000.

Figure 4 shows unemployment patterns in the female-dominated sector of education and health services. The first thing to note is that the unemployment rate among all workers in this sector — both men and women — is far lower than the unemployment rate in the male-dominated sectors. Unemployment stayed below 8 percent in the first half of 2009 for both men and women, followed by a substantial rise in unemployment between May and June — corresponding with the end of the school year in Minnesota. For men working in this industry, the unemployment rate more than doubled between May and June when 1,100 of them entered unemployment. The increase in women’s unemployment rate was comparable, but because of their higher concentration, resulted in an additional 16,000 unemployed women. 

Thus far, we have learned three key points. First, the gap in male/female unemployment levels shrank substantially from May through August.  Second, male-dominated industries continue to post the highest levels of unemployment. Finally, part of the decline can be explained by seasonality (for example, women’s higher unemployment in the education and health care sectors, and men’s declining unemployment in construction, manufacturing and similar industries). But another portion of it does not appear to be seasonal: specifically, women’s increasing risk of unemployment in the male-dominated industry sectors.

Other Explanations

In education and health care, there are no clear gender differences in unemployment rates, although the impact on the level of female unemployment was significantly greater. Meanwhile, gender differences are quite pronounced in the male-dominated industry group. This raises the question: Why are women facing such an increased risk of layoffs in the male-dominated industry sectors? 

Occupation Mix

One possibility involves the specific jobs that women perform in this sector. If women working in the manufacturing or construction industries are overrepresented in clerical or administrative occupations, while men are more concentrated in blue-collar occupations, men and women might face different layoff patterns even within an industry. 

To investigate this possibility, Figure 5 shows the male and female unemployment rate over time in industry Group 1, separated by male, female and sex-mixed occupations. (See information below Figure 5 to learn more about these groupings.) It is important to recognize that the sample sizes in month/gender/occupation classifications can become quite small and cannot support sweeping conclusions or generalizations about trends in this sector.  That said, the patterns in Figure 5 are suggestive and do help us rule out certain possibilities.

For the purposes of this analysis:
“Male“ occupations were at least 61 percent male:
• Management	• Architecture and Engineering	• Computer and  Mathematical Science
• Protective Service	• Farming, Fishing and Forestry	• Installation, Maintenance  and Repair
• Production	• Construction and  Extraction	• Transportation and  Material Moving
“Female” occupations were at least 61 percent female:
• Education and Training	• Community and Social Service	• Food Preparation and  Serving-Related
• Health Care Practitioner	• Health Care Support	• Office and Administrative Support
• Personal Care and Service
“Mixed” occupations included the remainder:
• Legal	• Business and Financial Operations	• Life/Physical/Social Science
• Arts/Design	• Sales and Related Occupations	• Building and Grounds  Cleaning and Maintenance
Source: Author’s analysis of Current Population Survey data. Data are not seasonally adjusted.

Specifically, it appears unlikely that people’s occupations alone can explain their risk of unemployment in this sector. If it could, we would expect to see similar unemployment patterns for men and women in the different occupational groups. For example, we might expect to see increasing unemployment for both men and women working in female-dominated occupations, or decreasing unemployment for men and women in male-dominated occupations. Women’s unemployment did in fact increase in female-dominated occupations in this sector, but men’s did not. Even more telling are the unemployment patterns in male-dominated occupations in this sector. For men, unemployment in these occupations steadily decreased from nearly 17 percent in February to around 12 percent in August. For women, it increased, from about 9.6 percent in February to about 18.3 percent in August.

Figure 6 tells the same story in a different way. Again we see the male and female unemployment rates in Industry Group 1, as well as the hypothetical unemployment rate for women if they had the same occupational employment patterns as men. We learn several things. First, if women were employed in male-dominated occupations to the same extent as men, their unemployment rate would have been higher throughout the entire time period. Yet, beginning in May when women’s actual unemployment began to outpace men’s, a hypothetical shifting of women into male-dominated occupations would not have insulated them from layoffs. Put another way, having the same occupational employment patterns would not have brought the male and female risk of unemployment together — at least, for this time period.

It is worth pointing out here that people working in the same occupations in the same industry are likely to earn similar wages and have similar levels of education. The preceding findings suggest, then, that neither education nor pay differences between men and women are likely to explain sex differences in unemployment in this sector.

Part-Time Employment

Another possible explanation for the different patterns of unemployment is part-time employment. Have women (who are more likely to work part-time, even in this sector) experienced different layoff patterns because of this? One could imagine a scenario, for example, where employers disproportionately opted to lay off their more expensive full-time staff early in the recession in order to cut costs. Then, perhaps changing strategies, they began laying off part-time workers as the recession wore on in the summer months. If women are disproportionately part-time, they would likely be disproportionately affected by layoffs over time — first insulated, then at greater risk—if this scenario were true. Do the data provide support?

In a word, no. Figure 7 shows the actual male and female unemployment rates, compared with the hypothetical female rate that would result if men and women were evenly distributed in part-time and full-time jobs in this sector. The actual and hypothetical female unemployment rates are essentially the same. From January through June, the hypothetical female rate is just slightly lower than the actual rate, suggesting that, if anything, women’s propensity to work part-time might have put them at a slightly greater risk for unemployment during this period. But the differences are so slight that it appears part-time employment simply cannot be considered a major explanation for the differences.

Where Does That Leave Us?

To summarize, this analysis has found a divergence in unemployment trends for men and women during 2009. Through April, male unemployment far exceeded female unemployment. But since then, female unemployment has risen while male unemployment has declined slightly. In fact, if men and women were evenly distributed in industries, male and female unemployment would have converged in August 2009.

The gap that existed, and that has since closed, was driven by the timing of female unemployment in male-dominated industries on one hand and by the recent (and later) rise in unemployment in female-dominated industries on the other. In female-dominated industries, the increase in female unemployment can be largely explained by seasonality (seasonal layoffs in education) and industry trends, such as employment growth in health care flattening out in recent months.

In male-dominated industries, however, neither different occupational employment (which is closely related to both pay and education levels) nor differences in part-time employment can account for the recent rise in female unemployment and drop in male unemployment. 

Top