Measuring the effects of the Covid-19 Delta wave on the U.S. hourly labor market

Given the timing of the survey underlying the leisure and hospitality payroll data (noted in figure 2 by the dashed vertical lines), we may not see the full effect of the recent declines in the Homebase data until the September payroll survey results are released on October 8, 2021.2 Even so, there are some important caveats to keep in mind. Homebase data and Covid-19 cases and vaccinations In order to more formally capture the relationship between the Homebase data and the Delta wave of the virus, we decompose the variation over time in the state-level Homebase data on hourly employees and hours worked using panel regression analysis.5 Specifically, we regress the percent changes in the Homebase data since April 2021 on a daily basis on the growth in Covid-19 cases and changes in the share of fully vaccinated persons since April.6 The use of state-level, as opposed to national, data allows for us to flexibly capture regional differences in virus transmission and vaccine take-up in our panel regression with coefficients that vary by U.S. Census division.7 Also included in these regressions are state fixed effects interacted with percent changes in the number of open businesses in the Homebase data since April 2021 and a linear time trend. Together these variables help to control for sample attrition,8 as well as seasonality in the number of open businesses that may vary by state. In order to ensure that our results are nationally representative of the hourly labor market, we exploit the connection that we demonstrated in figure 2 and weight our panel regressions by each state's share of first quarter 2021 leisure and hospitality payrolls.9 Weighted in this way, our panel regressions reveal that for every Census division, hourly employees and hours worked are positively correlated with changes in the share of fully vaccinated persons since April and negatively correlated with the rate of growth in Covid-19 cases since April.