(Updated 4:55 P.M., Revised April 18, 2020)
The March report on state-level employment and unemployment showed early, initial impacts of the COVID-19 pandemic. Although the survey reference periods for the report predate many business and school closures that occurred later in the month, today’s report clearly indicates the COVID-19 pandemic had substantial impacts on on labor markets in March.
The survey reference week for both the household and establishment surveys was March 8th – 14th, the week before initial claims for unemployment insurance showed sharp increases during the last two weeks of the month. Nevertheless, the Arkansas unemployment rate was estimated to be 4.8%, up from 3.5% in February. The national unemployment rate rose from 3.5% to 4.4%.
The Bureau of Labor statistics reported several problems that were encountered in the process of collecting survey data in March. For the household survey (on which the unemployment statistics are based) the BLS noted that in-person interviews were suspended, and that survey response rates were down 10 percentage points from recent months. There were also problems with the classification those affected by COVID-19 related closures: Some workers who should have been classified as “unemployed on temporary layoff” were instead categorized as “employed but absent from work.”
In their news release for Arkansas, the Division of Workforce Services, reported that “The increase in the number of unemployed Arkansans is, in large part, a reflection of the number of Unemployment Insurance claims filed during the week of March 12th. The small gain in employment is based largely on a monthly survey of Arkansas households, which was significantly impacted by the outbreak.” Correspondence seeking clarification indicated that both the survey data and the Unemployment Insurance Claims data are typically used to estimate the monthly unemployment levels.
Evidently, data on UI Claims were used from later weeks in the month. The number of initial claims filed during the week ending March 14th was 1,382. For the week ending March 21st, new claims were 9,275. (The really big increase didn’t come until a week later.) The increase in the number of unemployed in today’s report was 18,526.
Whatever procedure was used to estimate the numbers of employed and unemployed for March, the calculations lead to an odd result: The labor force expanded by 19,032. This includes the increase in the number unemployed and an increase of 513 in the number employed (see figure, below). An observation that the labor force expanded one-for-one with the number of newly unemployed really makes no economic sense. Literally, it suggests that thousands of unemployed people, who weren’t in the labor force in February, were dumped into the Arkansas economy in March. Because the unemployment rate is the ratio of the number of unemployed to the labor force, this odd co-movement leads us to question how the March unemployment rate should be interpreted, or how it can be put into context. Perhaps the best conclusion we can draw is this: unemployment increased significantly during March.
Update (4:55 PM): After consultation with the Arkansas Department of Workforce Services, we appear to have an explanation for the anomaly described above; namely, that the number of unemployed and employed (the labor force) jumped by an inordinate magnitude.
As we described earlier, some workers who should have been classified as “unemployed on temporary layoff” were instead categorized as “employed but absent from work.” As a result of this (and perhaps other) classification issues, workers were counted as employed in the survey, but were also eligible to apply for unemployment insurance. Consequently, the Arkansas data appear to include a number of workers double-counted as both employed an unemployed.
This explains the anomaly, but does not clarify the interpretation of this month’s unemployment rate. If anything, it suggest that the March unemployment rate is understated: many of those counted as “employed” in the labor force should be dropped from the labor force total (because they’re already counted as “unemployed”), which would raise the calculated unemployment rate.
(A perusal of the data for all 50 states suggest that the unemployment rate calculations for as many as 7 -10 states may have been affected by this anomaly.)
Nonfarm payroll employment declined by 7,700 jobs from February to March (seasonally adjusted). The largest declines were in Leisure & Hospitality Services and Education and Health Services (fully attributable to a decline in Health Care & Social Assistance). Other Services and Manufacturing also showed notable declines. Employment in several sectors increased, including Construction, Financial Services and Professional & Business Services.
After the March decline, Arkansas employment is down 3,300 from a year earlier.
The BLS reported problems with the establishment survey as well. Regional call centers that are usually used for phone surveys were closed, so electronic data collection methods were used instead. The collection rate was reported to be about 9 percentage points lower than average.
Despite the data collection problems in both the household and establishment surveys, the BLS concluded in both cases that the agency “was still able to obtain estimates that met our standards for accuracy and reliability.”