Unemployment Rates and Nonfarm Employment

Starting in March, the labor market has been rocked by the responses to the Covid-19 pandemic, from the initial public health responses involving government-imposed lockdowns, followed by staged business reopenings, and overlaid with individual behavioral changes in response to the health implications and risks of the virus. In April, unemployment rates reached levels unheard of since the Great Depression. Since then, unemployment rates have declined at the national level and, until last month, in Texas. However, after months of decreases, the unemployment rate in Texas and across all of the metropolitan statistical areas (MSAs) in Texas increased rather substantially between August and September.

In the dashboard below, the unemployment rate, the civilian labor force, and the number of workers employed and unemployed are depicted for the United States, Texas, and a set of Texas metro areas. The top panel shows the seasonally adjusted values for each series and the bottom panel shows the unadjusted series. As the name implies, the seasonally adjusted series account for seasonal changes in employment, and the adjustment helps uncover broader changes in the overall business cycle. We’ll mainly focus on the seasonally adjusted series but will point to the unadjusted series in the case of College Station-Bryan.
 
Between August and September, the national unemployment rate fell from 8.4% to 7.9%. Texas was one of 13 states in which the unemployment rate increased between August and September, rising from 6.8% to 8.3%. As noted in a companion blog post, the unemployment rate rose between August and September in all Texas metropolitan areas. Houston had the highest increase of the five MSAs depicted at 1.7 percentage points. Employment and the labor force declined in each of the Texas MSAs depicted, both seasonally adjusted and unadjusted, except for College Station-Bryan.

In College Station-Bryan, the unadjusted labor force and employment series both rose, but the seasonally adjusted series declined. Note that in the unadjusted series for College Station-Bryan there are distinct “yearly cycles” with higher employment and labor force counts in the fall and the spring of each year that coincide with Texas A&M University’s semesters and the influx of students and visitors. For example, unadjusted employment grew on average by 6,852 workers between August and September in 2018 and 2019. However, this year unadjusted employment grew by only 631 workers. Thus the typical seasonal pattern did not occur. Moreover, since the seasonally adjusted series is built on the expectation of higher growth in September, when that growth fails to materialize the seasonally adjusted labor force can decline even as the actual employment level is increasing.

The series depicted in the dashboard are derived from a household survey – the Current Population Survey (CPS) - that samples about 60,000 households monthly. But there is another survey that samples firms, the Current Employment Statistics (CES) survey, that samples about 160,000 nonfarm firms and government entities. The sample counts the number of workers on a payroll but does not include self-employed workers. The advantage of the CES survey is that compositional changes in nonfarm employment by industry can be tracked from month to month. The larger sample representing more workers makes it less variable than the estimates from the CPS.

The figure below graphs the seasonally adjusted national level of nonfarm employment by month for 2019 and 2020 (indexed to March 2006). Nonfarm employment fell 14.5% from February to April of this year. Between April and September, it rose 8.8%. To regain its pre-COVID level, nonfarm employment nationally will have to rise another 7.6%.

11-10-20.PNG
The next figure depicts seasonally adjusted nonfarm employment in the US, Texas, and the same Texas MSAs. All rates are indexed to their respective values in February of this year. Nonfarm employment levels in these Texas MSAs and the state are closer to their pre-pandemic levels than national nonfarm employment. College Station-Bryan’s employment in September was within 2% of its February level. Houston’s September employment was 93% of its pre-pandemic level. Employment in September for Texas was 94% of its February level; nationally it was 93%.  

10-30-2020-figure-2-(1).PNG
So, while the September unemployment rate in Texas was higher than the national rate, the level of nonfarm employment in Texas was closer to its pre-pandemic level than was the case nationally. This dichotomy results from the fact that the Texas labor force was larger in August and September of this year than in February. In September, employment also fell faster than the decline in the labor force.

Dashboard Sources
 
Bureau of Labor Statistics; Current Population Survey (CPS), National Labor Force Statistics; https://www.bls.gov/data/
Bureau of Labor Statistics; Local Area Unemployment Statistics (LAUS), State and Metropolitan Area Labor Force Statistics; https://www.bls.gov/data/
Bureau of Labor Statistics; Local Area Unemployment Statistics (LAUS), Smoothed Seasonally Adjusted Metropolitan Area Estimates; https://www.bls.gov/lau/metrossa.htm

Posted: October 30, 2020 by Dennis W. Jansen, Carlos I. Navarro, Andrew J. Rettenmaier