The Yale Labor Survey, a research project from Yale University’s Tobin Center for Economic Policy, is designed to provide policy makers and the public with real-time information about the US labor market. Its questions are modelled on those of the monthly BLS-Census Current Population Survey. The project has been designed and developed by researchers at Yale University, Stanford University, and the Boston Fed’s research department. Data are made available weekly at the websites of the Boston Fed and the Yale Labor Survey website at the Tobin Center. Background documentation for the Yale Labor Survey is available in Cowles Foundation Discussion Paper No. 2243. The Yale Labor Survey is not an official forecast of the Federal Reserve Bank of Boston or its president, the Federal Reserve System, or the Federal Open Market Committee.

The Yale Labor Survey (YLS) currently covers labor market activities from early April 2020 through the week of August 2-8. Data are not seasonally adjusted. The survey’s key findings are as follows:

• The unemployment rate for the population aged 20 and older peaked in early May and has improved substantially since then. For the week ending August 8, the YLS estimates an unemployment rate of 11.9%, down more than 3 percentage points since mid-June.
• In official government data, “employment” includes (i) persons who “work for pay” during the reference week plus (ii) persons who are absent from work because of vacation, illness, or similar reasons. The YLS measures the ratio of work-for-pay to population (WFPR) and the employment-to-population ratio (EPR). The WFPR was 53.7% for the week ending August 8, close to its level over the last month and up 1.8 percentage points since mid-June. The EPR was 56.5%, up by 1.7 percentage points since mid-June.
• The labor-force participation rate (LFPR) was estimated at 65.7% for the week ending August 8, down about 1 percentage point over the last month.
• The YLS agrees more closely with CPS estimates of employment than it does with CPS estimates of unemployment and labor force participation. Although YLS estimates of unemployment and participation are higher than the corresponding rates in the CPS, movements over time in unemployment and participation are similar across the two surveys.

The work-for-pay ratio (WFPR) measures the fraction of survey respondents who report that they worked for pay or profit during the reference week. After falling sharply when the pandemic hit, the WFPR rose significantly from late April through mid-June but has been stable since then. The last CPS estimate of the WFPR in July stood at 53.2%. The WFPR in that week for the YLS was 53.0%. During the following three weeks, the YLS estimate has risen 0.7 percentage point, to 53.7% for the week ending August 8.

The labor market picture changes little in moving from the work-for-pay concept to the employment concept, the latter being the standard measure of labor market activity reported by the Bureau of Labor Statistics (BLS). For the July CPS week, the BLS reported a 20+ employment-to-population ratio (EPR) of 57.1%, up slightly from 56.6% in June.

During the pandemic, the BLS has stated that the CPS has had difficulty distinguishing between workers who were absent from their jobs for the usual reasons (such as illness or vacation) and workers who were displaced by the pandemic. An alternative measure of the government’s EPR calculated by YLS researchers, “EPR-alt”, is designed to remove pandemic-displaced workers from CPS employment (see the background documentation for details). The EPR-alt measure is close to the YLS definition of employment, which includes only paid absences and is therefore less susceptible to classification error. Both the calculated CPS EPR-alt estimate and the YLS EPR estimate stood at about 56% during the week of July 12-18. The YLS measure has risen by about ½ percentage point since then.

The YLS has consistently overestimated the unemployment rate, although the divergence depends upon whether corrections for recent CPS classification errors discussed above are made. To be classified as unemployed, a non-employed person must either (i) have actively searched for a job in the last four weeks, or (ii) be laid off from a job with an expectation of return. The questions that gather information about job-search activities and layoff conditions differ slightly between the CPS and YLS.

Despite the differences, the basic trend in unemployment-rate estimates are similar in the CPS and YLS. Unemployment fell significantly from April through June, consistent with the significant increase in employment that both surveys measure. During the July CPS survey week, the YLS unemployment rate was 13.4%, compared to 11.7% for an adjusted CPS unemployment rate (“U3-alt”). Since that week, the YLS unemployment rate has declined by 1½ percentage points, to 11.9% for the week ending August 8.

The decline in unemployment in recent weeks results from both the modest increase in employment noted above and a modest decline in the labor force participation rate (LFPR). According to the YLS, the LFPR has declined by ½ percentage point since July 12-18. Over the entire sample period, the YLS participation rate has been higher than the corresponding CPS rate. This discrepancy is due almost entirely to the overestimate of the unemployment rate, which in turn is due largely to the YLS’s overestimate of the number of persons searching for jobs. (An increase in the number of job searchers raises both the unemployment rate and the participation rate.)