Gauging the unemployment rate run-up
admin | April 3, 2020
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
Economists have been ratcheting up estimates of how high the unemployment rate can go in the coming months. Forecasters are seemingly competing to have the highest figure, with projections running as high as 20% to 30%. While millions of jobs have unquestionably been lost, I envision a smaller surge in the jobless rate to 10.5% in April and then gliding back to around 5% by the end of 2020.
The household survey
Thinking about where the unemployment rate is headed is far trickier than one might initially imagine. The household survey questionnaire goes as follows
- Are you working?
- If yes, then you are employed. If not, go to the next question
- Did you actively look for work over the past month?
- If yes, then you are unemployed. If not, then you are not in the labor force.
That leads to three groups of people: the employed, the unemployed and those outside of the labor force. A key distinction in the current episode, however, is that if you are considered “unemployed on temporary layoff,” then you do not have to look for a job to be classified as unemployed.
Different ways to classify those whose jobs are affected
In the current situation, there are two highly uncertain margins along which firms and workers respectively will operate. At the firm level, the recent fiscal relief package offers a good framework. If I am a small business owner, I can choose to take care of my workers in one of two ways. I can participate in the small business lending program, the Paycheck Protection Program (PPP), in which case I agree to keep them on the payroll, with the federal government effectively paying them through me. Or, I can lay them off, knowing that the federal government has promised to offer four months of full pay. Many lower-wage workers, due to a quirk in the law, will get significantly more in unemployment benefits than they were earning on their jobs. In essence, the end result in either case is that the workers are being paid by the federal government, but in scenario 1, the workers may be considered employed and in scenario 2, they are unemployed.
On the worker side, the key question is what respondents tell the Bureau of Labor Statistics and how the surveyors and the agencies classify their answers. In March, the BLS made it easy on their survey staff. Anyone whose business was closed due to the coronavirus was supposed to be classified as “unemployed on temporary layoff.” Presumably, that would be the case whether they were paid or not. However, the way that the responses were actually coded was not nearly as clean as that. In round numbers, apparently between 1 million and 1.5 million workers were misclassified as employed but absent from work for “other reasons.” They could have been misclassified because the respondents said that they were employed (even though they did not actually work) or because the staffer conducting the survey put them in the wrong category. If all of those workers had been labeled unemployed on temporary layoff, then the unemployment rate in May would have been almost a full percentage point higher.
I suspect that the April reality will be even more complicated. With the PPP operating, will small business employees being kept on staff answer the survey as “employed” or “on temporary layoff?” My guess is that April will show a mix again. Moreover, it will be more problematic for the BLS to wave a hand and consider everyone not at work on “temporary layoff” as some firms will unfortunately be forced to close. By the time of the April survey, there may be some workers who can no longer consider themselves furloughed, and then, to be considered unemployed, they will have had to actively look for work. In March, the labor force contracted by over 1.6 million, which accounted for more than half of the decline in the number of employed in the household survey. I expect further shrinkage in the labor force in April, which will further limit the surge in the unemployed.
From a macroeconomic standpoint, it matters very little how people fall within each of these two margins, but it will matter crucially for how high the unemployment rate goes.
Just to provide some framework, the consensus view appears to be that the unemployment rate will spike to 12% to 15% over the next few months and then slowly recede in the second half of the year and 2021. There are several elements of the consensus view that I struggle with. The first one is that employment will continue to contract into the third quarter. Perhaps I am a hopeless optimist, but it hard for me to imagine that the economy will be operating at a lower rate in, say, August than it will in April and perhaps May, when much of the economy will be shut down.
Second, many unemployment estimates appear to assume that the programs in the fiscal relief package have virtually no impact. I look for millions of workers at small business firms to be called back. Just to work through the math, the federal government is offering $350 billion in loans to small businesses to cover their costs for the next eight weeks. Let’s assume that, as is being reported, the program is fully subscribed (there are concerns that the money could be spoken for by the end of the first day) and that a little over half of the costs covered represents payroll. That’s $200 billion. The average wage earner is paid about $50,000 annually, so eight weeks’ worth of pay would be about $8,000. The PPP program under those assumptions would cover 25 million workers for the next eight weeks. That seems very high to me, but I do not think it’s outlandish to assume that the PPP could save 5 million to 10 million jobs. As discussed above, it is a critical question how individuals in the survey choose to answer the questions and how survey staff code those answers. Based on what the BLS did in March, I am assuming that when there is a question, people will be classified as unemployed, so that the impact of the PPP and other programs may not be fully reflected in the jobless figures, but there is tremendous uncertainty. In March, for example, it appears that roughly 4 million jobs were affected, but only one-third of those ended up in the unemployment tally.
In sum, the surge in unemployment may be limited by two other factors. First, for those that do lose their jobs, I assume that a significant portion of them will be classified as falling out of the labor force rather than unemployed, as described above. Second, some fraction of workers who are furloughed will get scooped up by the firms that have seen a surge in demand (e.g. Wal-Mart, grocery stores, Domino’s Pizza, etc.). Based on news reports of hiring plans, this reallocation of workers will likely be worth several million jobs.
In any case, to get to 15% unemployment, we would need to see an increase of nearly 20 million in unemployed with no drop in the labor force. In my view, 20 million is probably an upper limit of the number of jobs impacted by closures due to quarantine rules. However, many big businesses have promised to pay their workers for the time being, even if operations are shut down, and millions of small business workers should be protected by the PPP. As an aside, even if initial jobless claims cumulate by 20 million over the next several weeks, I would argue that there is probably significant duplication in the chaotic flurry of filing for unemployment insurance (it will ultimately be better to track continuing claims, the total number of people collecting benefits) and that a substantial number of those who are initially laid off and file a claim will be called back, especially once the PPP is fully operational, or find work elsewhere.
Based on the way that the BLS chose to treat the numbers in March, I have bumped up my unemployment projections considerably, but I remain less extreme than the consensus. I am assuming that we end up losing about 11 million jobs in April, of which 10 million are classified as unemployed and 1 million are labeled as outside the labor force. In that scenario, the jobless rate spikes, but “only” to about 10.5%. I am assuming a flat performance in May and then a gradual recovery beginning in June, with the unemployment rate falling back to around 5% by late in 2020.
I would just underscore that I would not put much value on any of these sorts of projections for two reasons. First, no economist has any particular expertise in predicting the path of the coronavirus, which will ultimately determine the timing and magnitude of both the downturn and the economic rebound. Second, the depth of the hopefully brief downturn is far less important than the timing and slope of the rebound. I would rather see a spike to 30% in unemployment that is quickly reversed than a jump to 10% that is followed by a slow slog back toward normal.