Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

In Japanese economic review (Oxford, England)

The Labor Force Survey, a large-scale government statistics, and the causal forest algorithm are used to estimate the group average treatment effect of the COVID-19 on the employment status for each month from January to June 2020. We find that (1) because of the seasonality in employment status at monthly level, whether we use January 2020 as the base month for comparison, as done in most of the studies or whether we use the same month last year as the base comparison group makes a large difference; (2) whether we include those who are absent from work among the employed or not makes a large difference in the measure of the impact of COVID-19 and its changes; (3) if we use the employment measure which does not include those who are absent from work among the employed, 25-30% among the employed are adversely affected and that 10% of the employed experienced more than 10% decline in employment probability in April, 2020; (4) those who are the most affected by the COVID-19 are those who are unemployed or work part-time in the hotel and restaurant industry and service occupations; (5) in addition, younger and female respondents are more affected than are older and male respondents; and (6) we observe no clear differences in the impacts of COVID-19 with respect to living location, education status, and firm size among the most affected.

Fukai Taiyo, Ichimura Hidehiko, Kawata Keisuke

2021-Jul-19

COVID-19, Causal machine learning, Decomposition, Job creation