By Jiyeon Kim, Fellow at KDI

※ This article is part of KDI Journal of Economic Policy, May 2021

The recession in 2020 caused by COVID-19 is unprecedented in many ways. In this paper, I explore one of the unique features of the pandemic recession: its disproportionate impact on female employment. It has been well documented that women, especially married women, have a lower cyclicality of employment than men (Albanesi, 2019). This is explained to some extent by a high share of female employees in jobs that are less sensitive to business cycles, such as service occupations (Albanesi and Sahin, 2018). Married women’s tendency to stay employed in economic downturns in response to the increased risk of spousal job loss also plays a role (Ellieroth, 2019). Consequently, we usually observe a larger drop in male employment during recessions.

During the COVID-19 recession, however, a different pattern emerged. Figure 1 describes employment losses for men and women throughout the year 2020 in comparison with the 1998 recession. With year fixed effects and seasonality controlled for, the employment-to-population ratio for married women dropped much more than that for married men in March, when the number of confirmed cases of COVID-19 spiked for the first time. The difference becomes more striking considering the lower reference employment rate for married women. The gap narrowed as female employment recovered more rapidly in the lull periods but started to widen again with the start of the third wave of infections in December of 2020. No significant gender differences were observed between single men and women at least in the first half of the year, but the gap began to broaden starting with the second wave in September of that year. This pattern is in sharp contrast to the 1998 recession, in which men experienced a greater drop in the employment-topopulation ratio than women.

The reason COVID-19 took a greater toll on female employment, unlike in previous recessions, appears to be twofold. One factor is related to the types of jobs the pandemic hit. The risk of infection and social distancing measures imposed to curb the transmission of the virus mainly disrupted jobs in the services industries. Women were more affected by this disruption because they are overrepresented in such jobs. Another important factor is the increased need for childcare at home caused by school closures. Given that it is commonly the mother who is in charge of childcare in the household, when children spend more time at home, it becomes difficult for working mothers to stay in the labor market.

This study evaluates the impact of the COVID-19 crisis on employment from monthly survey data in Korea. For the impact evaluation, counterfactual outcomes are constructed for each subdivided group based on simple assumptions. While there is more than one way to construct counterfactual outcomes, the evaluated impacts provide a reasonable reference point from which to evaluate the national-level shock. Subsequently, the employment shocks by the COVID-19 crisis are decomposed into their persistent and transitory components by utilizing the observed events of the three major outbreaks and the subsequent recovery periods as the source of identification.

In this paper, I examine both possibilities. I first document the heterogeneous impact of COVID-19 across different jobs along with the share of men and women employed. I find that jobs with a high share of female employees are most affected by the pandemic. To ascertain if this is the main reason women fared much worse than men, I explore gender differences in the outflows from employment using the individual-level data. I find that the transition from employment to non-employment (E to NE) for married women rose by an additional two percentage points from its pre-pandemic level of 1.9 percentage points compared to married men in the first wave of the pandemic. Controlling for job characteristics such as occupations, industries, and worker arrangements mitigates the gender differences, but a statistically significant gap of 0.9 percentage points remains. Decomposing the E to NE transition into the employment to unemployment (E to U) and the employment to non-participation (E to N) transition, I find that the gender gap in the E to N transition is more than twice as large as the gap in the E to U transition. Moreover, job characteristics explain most of the gap in the E to U transition but only half of the gap in the E to N transition.

The aforementioned results imply that a sizable gender difference unexplained by women’s concentration in service jobs exists in labor supply behavior in response to the pandemic. As likely as it seems to be associated with added childcare responsibilities at home, it is not possible to obtain direct evidence of this due to data restrictions. Instead, I use workers’ marital status and age as a proxy for having children. The largest gender gap in the E to N transition is observed among married women aged 39-44, the group most likely to have elementary school age children. Women in this group were 1.4 percentage points more likely to leave the labor force than men during the first wave of infections taking all job characteristics into account. In the other age groups, gender disparities do not exist or are mostly explained by gender differences in the job characteristics. The heterogeneity observed among parents may reflect a disproportionate increase in the childcare burden according to children’s ages during the pandemic. Older children do not need as much supervision from parents. Families of preschool children who most likely need parental care the most were provided intensive governmental support such as emergency childcare services and extra child benefits. The fact that mothers of children between these age groups were most likely to drop out of the labor force during the COVID-19 crisis suggests that increased childcare needs played a sizable role in the excess drop in female employment. Since the start of the pandemic, a large body of work has examined its economic consequences from various angles. A number of papers are concerned with gender differences in the labor market impact of the pandemic. Most of them focus on occupational distributions, emphasizing that female-dominated jobs tend to require employees to work in a close physical proximity to other people and are difficult to be conducted remotely, which makes them especially vulnerable to the COVID-19 shock. A few studies state that femaledominated jobs’ excessive exposure to COVID-19 does not explain all of the gender disparities. Cajner et al. (2020) finds that even within detailed industries, women experience larger job declines than men. Adams-Prassl et al. (2020) point out that the gender gap persists even with job characteristics controlled for. Alon et al. (2020b) stress that men and women’s different labor supply responses to school closures make an additional contribution to women’s incremental employment losses. Albanesi and Kim (2021) show that the gender gap in employment losses is larger among parents than non-parents and that differential occupation declines do not fully account for the sharp increase in non-participation among mothers. Despite growing interest in the topic, evidence from non-US countries is still scarce.

This paper aims to fill this gap by providing evidence from the Korean labor market.

The paper is structured as follows. Section 2 documents the distributional impacts of COVID-19 across job characteristics and the share of female employees. Section 3 describes the data and methodology used in the paper. Section 4 examines the individual-level data and investigates the gender-related impact of the COVID-19 recession on outflows from employment. Section 5 discusses COVID-19’s long-run implications for female employment.