The Wild Ride of Pandemic Employment Status and Why Race *and* Gender Matter

It was back in August 2020 that Mina Kim and I first started looking at the (kinda-secret yet publicly-available) Bureau of Labor Statistics data on Asian female employment status and noticed that the Asian female pictures were like holding a magnifying glass to the more general stories documenting the so-called “She-Cession.” It turns out that the one race-and-gender category with the largest absolute increases in unemployment for several months from July through November were Asian women. I continued to track the data releases each month (not always remembering to update my blog I now realize, but always sharing on Twitter). I saw the overall female unemployment rate finally drop back below the male one back in October, while the Asian female unemployment rate has remained higher than the Asian male one even through last Friday’s data release for February 2021.

Now, employment status is a notoriously imprecise thing to measure, and responding to a short survey question with a short, categorical answer about something as complicated as what kind of work you’re doing during the pandemic is difficult. I know, because over the summer I got surveyed by the Census Bureau’s “household pulse survey” and was often not sure which answer to pick and wanted to add “but let me explain…” I had taken a buyout/severance package offer from my last full-time job, so I was not working that job anymore but was continuing to get paid (for another month or two at the time, at least). While the “loss” of my job was of my own doing, at the same time it certainly felt like it was a pandemic-related job loss that certainly hadn’t been completely my choice, and my lack of success in finding a new full-time role really made me feel like a legitimately involuntarily (and now long-term) unemployed person.

I started to imagine how many other people struggle to respond to those household surveys, maybe get confused about the different labels for employment status, maybe get worn out by the end of the questions and start answering without careful thought. This, of course, is the challenge of collecting data through fielded surveys, rather than letting people self-report what’s on their minds based on their Google searches or their other means of asking for help and interacting and transacting with the world.

But the BLS data remain the best data we have to study what’s going on with US employment, so that’s what I’ve been sticking with (while urging agencies and research organizations to expand efforts to collect more disaggregated data to better understand the macroeconomy as well as the disparities that aggregate and average statistics hide, per one of President Biden’s Day One executive orders). Now that a full year has passed since the economic starting point of the “pandemic recession” (the last employment peak in February 2020), I’ll put aside the more erratic unemployment story and the messiness of defining “labor force participation” (which also defines the denominator of the unemployment rate) and focus on the cleaner picture of the employment-to-population ratio.

Here are some charts that go back to January 2019 to get a bit of a running start so you see what “normal times” looked like in terms of employment-to-population. I first separate by race, then by gender, then by race *and* gender –to make the point that each way of cutting that data reveals a pattern you couldn’t see in the others.

First, by race:

Note: The onset of the pandemic (as seen in April 2020 employment) collapsed differences across the races while everyone’s employment rate suffered. Asian employment (in blue) was slowest to recover but is now (Feb 2021) as high as Hispanic employment (in red), a relative improvement from pre-pandemic.

Next, by gender:

Note: while overall female employment (in red) declined more in absolute terms than male employment (in blue) at the start of the pandemic, it has also recovered faster since the fall, so the *relative* differences between male and female E/pop are essentially back to pre-pandemic levels.

And now, women by race:

Note: Asian women (in blue) had the second highest employment rates for much of 2019 (close to keeping up with Black women, in green) but have been on the lowest end among the race categories (sticking close to Hispanic women, in red) over the course of the pandemic. In the most recent data (Feb 2021), the female-by-race categories have converged except for Hispanic women.

And now, men by race:

Note: differences across the men by race in employment-to-population also collapsed at the start of the pandemic except for Black men (who suffered a similarly bad fall from the worst starting point). The disparity between Black men and other men is even higher now (Feb 2021) than pre-pandemic.

The last two charts showing E/pop for women by race vs. men by race make obvious that gender plays a different role in employment status depending on one’s race. Hispanic and Asian women are less likely to be in paid employment than other women, while Hispanic and Asian men are more likely to be in paid employment than other men. If you look at Census data on the composition of households by race, it shows that Hispanic and Asian households are more likely to be opposite-sex married couples with children. This suggests cultural explanations for the gendered roles of women vs. men in providing household work and caregiving vs. working in the labor market.

You can go back to my August blog with Mina to see more about what we believe makes Asian women “different” from other women. On top of being raised by our immigrant mothers to do everything well (to pursue lots of education and succeed in our careers but be ever-attentive mothers), Asian women also tend to be more cautious about health and safety (and our behaviors/practices during something like a public health pandemic–just think about who were the only people you would occasionally see wearing face masks pre-pandemic?) and more “picky” (discriminating) about the quality of our children’s experiences and education. This explanation is supported by this recent Washington Post story about Asian families being slowest to send their kids back to in-person school:

As school buildings start to reopen, Asian and Asian American families are choosing to keep their children learning from home at disproportionately high rates. They say they are worried about elderly parents in cramped, multigenerational households, distrustful of promised safety measures and afraid their children will face racist harassment at school. On the flip side, some are pleased with online learning and see no reason to risk the health of their family.

By Moriah Balingit, Hannah Natanson and Yutao Chen in the Washington Post, March 4, 2021.

If Asian Americans are such a small fraction of the US population that the government statistical agencies don’t even collect enough data on them to get a statistically-reliable sample (for statistical weighting and seasonal adjustment purposes), then why should researchers and policymakers care about them? Well, Asians are the single fastest growing racial category in the US today. Asian women are the closest to closing the gender pay gap with White men among all women. Asians work disproportionately in the sectors of our economy that were growing the fastest pre-pandemic–sectors like leisure/hospitality, healthcare, and computer science/data analytics. Asian Americans, in other words, have been on the leading edge of movement in our economy–the most influential population in terms of the dynamics of the US macroeconomy. Study what’s going on with Asian women and Asian men at the most granular level you can, and you will better understand how our entire macroeconomy is doing and how to improve public policies and business practices to truly get our economy to close the potential “output gap” –which honestly we do not yet know how to measure because we don’t even yet “see” actual productivity, let alone our potential.

More Pictures on the Pandemic “She-cession” by Race

Here are a bunch of charts that illustrate why the Pandemic Recession is also sometimes referred to as the “She-cession.” These are based on BLS monthly data through August (released last Friday, Sept. 4). I’m also re-posting the bar charts I put up on Twitter and here last Friday, so the latest figures are in one place. (See original post on “Seeing Asian Women…” here and the table with all the unemployment data through August by race-gender categories here.)

First, compare male minus female unemployment during the Great Recession to during the Pandemic Recession (thus far):

During the Great Recession (Dec. 2007-2009/10 recovery period), male unemployment relative to female unemployment grew dramatically–and in general/overall male unemployment stayed higher than female even years after.
During the Pandemic Recession (since Feb. 2020), female unemployment increased dramatically and surpassed male unemployment in all racial categories, and only among Black workers did male unemployment again exceed female unemployment this summer.

The differences in unemployment across racial categories have been stark. It’s not just a recession with disproportionate impact on women, but one with disproportionate impact on people of color. Zooming in a bit more recently, here’s what the unemployment rates by race since January 2019 look like:

Unemployment rates by race categories, men and women combined. Note that Asian unemployment started the very lowest in February 2020 and is the second highest in August 2020.

Going back to looking at the employment outcomes among women by race over the past year and a half (since January 2019), here are charts showing the unemployment rates, the employment-to-population ratios, and the labor force participation rates:

Female unemployment rates, by race, January 2019 through August 2020. Note that like for race categories overall, Asian females had the lowest unemployment rate in February 2020, but in August 2020 have the second highest–as Hispanic female unemployment has shown the best recovery since its peak in April.

Female employment-to-population ratios by race, January 2019 through August 2020. Hispanic women had the lowest e/pop at worst part of the Pandemic Recession and have traded places with Asian women a couple times since. Black women had had the highest e/pop going into the pandemic but in August, white women had surpassed black women.
Female labor force participation rates by race. Asian female participation was relatively low pre-pandemic, but is second highest this summer–showing that many Asian women consider themselves still “in the labor force” even if currently unemployed.

And below is the headline bar chart I previously posted comparing absolute changes in unemployment rates for women by race in the Great Recession vs. during this Pandemic Recession (through August):

The Pandemic Recession is a “She-cession” and has had disproportionate impact on women of color, particularly on Asian women–who face a still higher unemployment rate in August (at 11.5%) than at their worst point in the Great Recession (8.4%).

And looking at Asians alone (both men and women), the “double whammy” of being both Asian and female in this Pandemic Recession is underscored:

Asians (especially the women) had very low increases in unemployment during the Great Recession relative to other race categories. This time is totally different.

Seeing Asian Women to Better Understand the Pandemic ‘She-cession’

By Mina Kim and Diane Lim (with assistance from Taylor D. Jones), August 2020

This pandemic recession has been like nothing we’ve ever seen in our lifetime. The U.S. has experienced the largest contraction since the Great Depression. Every week, we eagerly await each data release for any sort of good news, only to be met with confirmation after confirmation of the calamitous impacts of the coronavirus pandemic. Economists were quick to point out early in the pandemic that increases in unemployment have been more severe for women than men and more severe for people of color than for White people. But in all the analyses, Asian women have been missing. Given the goal of national representativeness in most surveys, the “invisibleness” of Asian women is understandable. We represent a very small percentage of the population. We are also probably not as responsive to surveys given how 1st and 2nd generation Asian American women are taught to blend in and not stand out. (Both of us reflect on how this “blend in” behavior was reinforced in us very early on in our childhoods, by our teachers, classmates, and even our immigrant parents.) As Asian female economists, we had personal incentive to take a good look at the data—we just wanted to find ourselves there! What we found is that not only have Asian women experienced the largest increase in unemployment during the pandemic recession, but that the case of Asian working women probably best shines a spotlight on how women’s employment depends as much or more on what’s going on at home (and the demands for our time there) than on what’s going on with the market demand for our paid work. What follows is a story of what has happened to women in the labor market during this Pandemic Recession and how this recession is different from the last, made clearer by our ability to dive deeper into details across race categories.

NOTE: For this analysis we used the BLS “One-Screen Data Search” tool here. While the monthly employment reports of the Bureau of Labor Statistics (BLS) present breakdowns of the unemployment measure by race and by race-gender categories, they only show data for the Asian category as a whole because of the valid concern that the Asian sub-sample sizes are too small to be statistically comparable with other race-gender categories. But the BLS does publish and make publicly available the data on Asian women separated from Asian men on its website. All data are monthly and unadjusted (not seasonally adjusted), for the populations age 16 and over. The “Asian” category includes only people who identify themselves as (only) Asian on the Census survey, not those who identify as two or more races; it also does not include Pacific Islanders. (A helpful BLS report which explains their categorization of people by race and ethnicity is here.)

  1. The “Pandemic Recession” is a “She-cession” (Figure 1). C. Nicole Mason, CEO of the Institute for Women’s Policy Research, first dubbed the pandemic recession the “She-cession” back in early May in this New York Times story, when it had become apparent that more women than men had become unemployed since February. At the time, the leading explanation was that women make up a disproportionately large share of the workforce in the industries that have been most adversely impacted from the “stay-at-home” orders: the leisure/hospitality sector (think restaurants, bars, sports venues, concert halls, movie theatres), other people-intensive personal services businesses (think beauty salons/barbers, gyms, yoga studios), and non-emergency medical care offices (think general/family practices, dentists, eye doctors).

Now there is recognition of the role childcare (and in-person school), or the lack thereof, plays in affecting the ability of women to work (see Alon, Doepke, Olmstead-Rumsey and Tertilt; Mongey, Pilossoph, and Weinberg; Dingel, Patterson, and Vavra; Modestino, Ladge and Lincoln; and Heggeness). Compared with a typical economic recession that impacts male employment more than female, the Alon et al. paper stresses that the “She-cession” has more severe short-term/cyclical impact on economic activity due to diminished “within-household insurance”—a reduced ability for parents to trade off/coordinate roles at home vs. work—and will also lead to a longer-term (persistent) widening of the gender wage gap even after the recession is over.

Figure 1.
  1. This recession is very different from the last one (Figure 2). The “Great Recession” (left panel) impacted male-dominated jobs in the economy more than the female-dominated ones. Before the start of that recession, goods-producing jobs (particularly in the durable goods/”heavy” manufacturing sectors) had already begun a longer-term loss due to both offshoring and automation. As the economy began its long, slow recovery from the recession (officially in “recovery” by June 2009), unemployment rates continued to rise, peaking for men in early 2010 at a much higher rate than for women. Overall female unemployment peaked later (in the middle of 2010) but remained below male unemployment until the middle of 2011. Many of the men who had been laid off from their manufacturing jobs during the Great Recession never came back to those jobs after the economy recovered. These were disproportionately middle-aged, White men who came out on the other side of the recession saying “well, I guess I’m retired now.” (These were also the White men disproportionately represented in the “deaths of despair” that Anne Case and Angus Deaton revealed.)

This time (right panel) the economic impact was felt suddenly due to a non-market event and has fallen disproportionately on the employment of women. After several years of very low unemployment where women and men switched ranking but stayed very close to each other, female unemployment very quickly overtook male unemployment at the onset of the pandemic and has retained its “lead” thus far.

  1. Pandemic unemployment has been dominated by the laying off and rehiring of largely part-time workers (Figures 3 and 4). The part-time share of total employed fell as unemployment peaked (in April) and rose as unemployment has declined (since May). The timing is consistent with job losses concentrated in the leisure/hospitality, personal services, and retail sectors, where workers are disproportionately young, part-time, and low-wage (no benefits). (This is also why government assistance has been critical for households to be able to maintain basic consumption and prevent a more severe “reverse multiplier” effect on the economy more broadly.)
Figure 3.

While the trends in unemployment of part-time workers impact women more than men across all race categories, Black and Asian women more often work full-time jobs than do White or Hispanic women.

Figure 4.

If pandemic unemployment were mainly driven by occupational and industry factors (layoffs from part-time jobs in the leisure/hospitality, retail sales, and personal services sectors), we might see Hispanic and White women suffering the greatest unemployment, and Asian and Black women, less so. In fact, it’s not that simple…

  1. The “She-cession” story differs across race (Figures 5 and 6). It appears to have come and gone among Black Americans but has persisted for other race categories. By June, Black male unemployment was back above Black female unemployment, while in all other categories female unemployment has exceeded male unemployment since April. In the prior recession, men overall faced more severe unemployment than women, and it was especially severe for Black men. In this recession, Black men have not had as large an increase in unemployment as many of the women, but only because they started so high. Before the pandemic, in February, the Black male unemployment rate (at 7.3%) was the highest of all the race-gender categories. As of July, the Black male unemployment rate (at 15.6%) was still the highest of all the categories.
Figure 5.
Figure 6.
  1. The “She-cession” experience has had biggest impact on Asian women (Figures 7 and 8). Among all women, Asian women had the lowest unemployment rate in February, yet their unemployment rate has been above average and in line with the unemployment rate of other women of color during the pandemic. In terms of the change in unemployment rates during this pandemic recession so far, Asian women have experienced the single worst change in unemployment through July—not just among women but across all the race-gender categories. (Hispanic women suffered the largest increase in unemployment from February to April; besides figures below, see details in the Appendix Table at the end.)
Figure 7.
Figure 8. (Note: Worst point in the Great Recession is going to differ for each race category. See the Appendix Table for the exact dates.)

Caveats, Take-Aways, and “To Be Continued”s:

  • We’ve provided a one-step-more-granular analysis here, simply by including Asian females in our look at employment outcomes by race and gender. But this still doesn’t include everyone! “Asian” doesn’t include those who identify as only part-Asian (of multiple races), nor the “Pacific Islanders, American Indians, Alaska Natives, Native Hawaiians, and Other Pacific Islanders” who also tend to get left out of the “Asian” category in analyses by race—due to even smaller sample sizes than that of Asian females.
  • Yet, this race-gender intersectional analysis is still meaningful in that we do see clear differences in employment outcomes across both race and gender and all the race-gender categories. The analysis “works” because as broad as these race-gender categories are, and as heterogenous as the characteristics of the populations within each category, there are still enough characteristics that are common (or prevalent) within each category yet distinct enough from those in other categories. We suspect, for example, that the “Asian Female” category contains a sizable group of women who have relatively low educational attainment and work relatively low-skilled and low-paid personal services jobs (think nail salon), which would partially offset whatever we see that is driven by the high-skilled, higher-paid Asian women working in professional service jobs. Yet the average characteristics and outcomes within the Asian Female category still put the group as a whole in higher-skilled occupations and higher-paid jobs compared with women in the other race categories.
  • And those distinct characteristics and outcomes are not merely random differences. Race is a social construct, and that’s why it is useful in studying social and economic behaviors, influenced by social and cultural norms. Asian women look “special” in this analysis of unemployment during the Pandemic Recession because their roles in the labor market and at home and the resulting tradeoffs they face in making decisions about work vs. home are likely different enough from other women. For example, Asian women are less likely to experience the gender wage gap, probably because of their high educational attainment and occupational choices—and that’s a plus for holding onto jobs during this pandemic. (Distinct from other women, Asian women are more likely to work high-paid professional service jobs than low-paid leisure/hospitality or retail jobs.) But Asian women are also more likely to be in a traditional family structure, married with children, and married to a man who has even higher market-earning power. (We hope to do a follow-on piece on the family structure and marriage/”assortative mating” factors later.) So, as highly productive as an Asian woman might be in her job, she is likely to be the household’s “secondary earner” and also the household’s “primary caregiver.” This is why a deeper study of what’s going on with Asian women in this pandemic recession would shine a light on the struggles of all working mothers, who all face demands on their time at home with the kids. Those demands (regardless of the financial opportunity costs) too easily tip the scales in favor of staying at home and cutting back on or quitting their paid work.
  • For Asian women, the cultural factors influencing our labor market decisions are perhaps stronger than for other women. (See this paper on “Babies, Work, or Both?” by Mary Brinton and Eunsil Oh.) Many working-age Asian women are immigrants or children of immigrants. Diane’s parents immigrated to the U.S. from Korea (dad) and the Philippines (mom, of Chinese ethnicity) in the 1950s, both to pursue PhDs in Chemistry. Diane’s mom finished her PhD in 1962, just months after Diane was born, and spent nearly a decade staying at home with Diane and her younger sister Janice and was never able to make up for that lost first decade of her career. Mina’s parents immigrated from Korea in the 1970s to pursue a PhD in Physics (dad) and a Masters in Nursing (mom). Mina’s mom put her career on hold to take care of her and her two sisters when family could not help. Yet both moms constantly coached their daughters to “do it all”: to both stay ever-active in the workforce and take primary responsibility for the proper care of the kids on the home front.
  • This “first look” at Asian women’s employment outcomes has underscored the importance of considering how social and cultural expectations about the roles of women at home vs. at work influence female employment outcomes, especially in this pandemic with kids learning from home and daycare and elder care facilities closed or downsized (or viewed as undesirable/unsafe by families). It also suggests that the increased demands placed on women’s time on the home front does not just force women who earn relatively low wages (below the cost of child care) to give up their paid job, but is likely causing many higher-earning women (who can afford to pay for child care even as it has become scarcer) to cut back on or even drop out of the labor force, even when the financial and economic opportunity cost of doing so is high.
  • Further investigation is needed, going still more granular and down to more household-level data, to better understand what truly drives women’s choices and outcomes concerning work. These are questions fundamental to the study of Economics and requires research methods that have largely remained outside the Economics profession. Further research should: (i) start with interviews and focus groups; (ii) use these findings to develop surveys; and (iii) field the surveys to intentionally oversample the population subgroups that best represent and emphasize the decisions we are trying to better understand.

Appendix Table: detailed unemployment rates and changes, comparing previous to current recession