Yesterday, the Bureau of Labor Statistics estimated that nonfarm payroll employment declined by 140,000 in December. The estimate for November was increased by 135,000. The level of employment for December was therefore virtually unchanged from the November report.
For the household survey, the estimate of the total number of persons employed increased by 21,000, in December. November’s estimate was revised higher by 77,000 as a result of the BLS’s annual update of seasonal adjustment factors. The unemployment rate was unchanged at 6.7%, as was the labor force participation rate at 61.5%.
The December figures put the current estimate of total jobs lost in 2020 at 9.18 million according to the payrolls survey, and 8.77 million for the household survey. The economy has therefore gained back 12.3 million of the 22.2 million payroll jobs lost in March and April. Household data indicate that of the 25.4 million people who lost their jobs in March and April, 16.5 million people are now back at work.
Of the net 8.9 million jobs lost since February (according to the household survey), 2.3 million are temporary layoffs and 2.1 million are permanent. The BLS estimates that in total 3.0 million people are now on temporary layoff and 3.4 million have permanently lost their jobs. It does not offer an explanation about the rest of the job losers, but these could include people who are in limbo, those who have left their positions voluntarily, those that were on temporary assignment and those who were self-employed but now consider themselves to be unemployed.
Table 1 below shows the change in nonfarm payrolls by industry sector from the start of the pandemic in February to the low point of employment in April and then to the latest reading for December 2020. It also shows the change in employment from November to December.
Source: Bureau of Labor Statistics, Employment Situation reports for April, June and December 2020. Note: the February data have been revised further since the publication of the April jobs report. The official statistics now show February 2020 total nonfarm employment that is 19,000 higher than the figure shown in the table. This difference should not change the basic conclusions that can be drawn from the table about the change in employment by industry over the periods presented.
The data show that most of the jobs lost have been in the services sector, especially in leisure and hospitality, education and health services and professional business services. Job losses in leisure and hospitality represent 40% of the 9.8 million net job losses since February.
Losses of government jobs have been steady since February and now account for 13.7% of the total net jobs lost since the start of the pandemic.
The net loss of 140,000 payroll jobs in December consisted of 498,000 of losses in leisure and hospitality, offset partially by gains in professional services, retail trade and goods producing industries. The December losses reflect the impact of the second wave of the pandemic.
The reported slowdown in the recovery in employment has been accompanied by year over year increases in hourly wages that have been above pre-pandemic levels as shown in the chart below:
Source: Bureau of Labor Statistics Monthly Employment Situation Reports.
The BLS has reported that this elevated percent change in average hourly wages is due to a change in the mix of jobs, as job losses have affected lower wage workers disproportionately, and not to general broad-based increases in wage rates. Accordingly, if the recovery in employment broadens to include those lower wage workers who are now unemployed, the change in average hourly wages should be more moderate in future months. However, this may be offset in part by scheduled increases in the minimum wage in many states.
Employment Projections. In the latest Summary of Economic Projections of the Federal Open Market Committee, released after its December 15-16, 2020 meeting, the median forecasts anticipate a decline in the unemployment rate to 5.0% in 2021 (an improvement from its September forecast of 5.5%) and then further declines to 4.2% in 2022 and 3.7% in 2023. The central tendency of these forecasts (which excludes the three highest and three lowest projections) is generally plus or minus 30-40 basis points around the median.
In order to get a more complete picture of the employment outlook, the forecast should include at least two other assumptions: the rate of growth of the civilian noninstitutional population and the labor force participation rate. I provide my estimates for these in Table 2 below:
Source: Bureau of Labor Statistics Employment Situation report for December 2020, FOMC Summary of Economic Projections, December 16, 2020 and Lark Research Projections. Notes: the assumptions for this projection model are highlighted in yellow. The assumptions for the unemployment rate are the median forecasts given in the FOMC’s Summary of Economic Projections.
My projections assume that the civilian noninstitutional population continues to grow at a slow 0.4% annual rate for the forecast period through 2023. They also assume that the labor force participation rate increases by 0.5% each year to 63.0% in 2023. At that level, the participation rate would still be below the 63.3% level recorded in December 2019.
Under this model, employment increases by 4.65 million in 2021 (or by 387,500 per month), 3.19 million in 2022 (265,000 per month) and 2.73 million in 2023 (228,000 per month).
The FOMC’s estimates appear to be more optimistic than the consensus of economists, as reflected in the Philadelphia Federal Reserve’s Survey of Professional Forecasters (SPF). The latest SPF report, which was released on Nov. 16, 2020, anticipates a decline in the average unemployment rate to 6.3% in 2021, 5.2% in 2022 and 4.6% in 2023. To make the SPF forecast comparable to the FOMC’s (which I believe are for end of period and not the yearly average), I estimate the period end forecast of unemployment rates in the SPF to be 5.8% for 2021 (vs. 5.0% for the FOMC), 4.6% for 2022 (vs. 4.2%) and 4.6% for 2023 (vs. 3.7%)
Plugging those unemployment rate estimates for the SPF into my model and assuming a slightly slower rate of improvement in the labor force participation rate (than for the FOMC), I estimate that the total number of persons employed would increase by 2.85 million in 2021 and 3.57 million in 2022.
The 20Q4 SPF predicts monthly gains in payroll employment of 321,600 in 2021, which equates to 3.86 million for the full year. Payroll employment is different from household employment, which is the basis of my projections, but the difference between my estimate and the SPF’s could also be explained by differences in the assumed labor force participation rate (or other assumptions). Nevertheless, the SPF does appear to forecast a slower rate of improvement in the unemployment rate and employment growth than the FOMC.
Neither the FOMC nor the SPF forecasts indicate their underlying assumptions about possible changes in the composition of the labor force over time. Based upon the differences in their assumptions, I tend to think that the FOMC’s forecast assumes that post-COVID and after a full economic recovery, employment will look pretty much the same structurally as it did pre-COVID.
On the other hand, some economists believe that there will be structural shifts in the economy, such as an increasing number of people working from home and a reduction in business travel, among others, that will lead to permanent changes in the composition and level of the workforce. If my interpolation of the SPF’s projected unemployment rate is correct, the Philly Fed forecasters appear to be projecting that the unemployment rate will bottom out above the pre-COVID level.
January 9, 2021
Stephen P. Percoco
16 W. Elizabeth Avenue, Suite 4
Linden, New Jersey 07036