U.S. employers added a jaw-dropping 528,000 jobs in July, far more than expected, and wages jumped higher than expected—and yet the number of people working or looking for work declined. Where is everyone?

The Department of Labor said on Friday that the size of the labor force fell by 63,000 in July even though the population grew by 177,000. The labor force participation rate—the share of people working or looking for work—fell by one tenth of a percentage point to 62.1 percent.

The mystery of the shrinking labor force and the decline of labor force participation is one of the persistent conundrums of the post-pandemic recovery. The labor market has been very strong—so strong that many think we cannot be in a recession despite two consecutive quarters of negative economic growth—but somehow this has not pulled a larger share of Americans into the labor force.

“I can’t have a client meeting without someone asking me where all the workers have gone. I have no idea, but they are apparently gone,” Credit Suisse analyst Zoltan Pozsar wrote in a note to clients on Monday.

Harvard economist and former Obama administration advisor Jason Furman said on Friday that he is “baffled” by the July decline.

I am baffled by the decline in labor force participation. Peaked at 62.4% in March & has fallen nearly continuously since–despite tons of job openings, lots of jobs added, dwindling cash balances, mostly improving COVID, recession talk scaring people into accepting jobs, etc. pic.twitter.com/M7VskiBVq5

— Jason Furman (@jasonfurman) August 5, 2022

There’s also another mystery going on here, which is almost the mirror image of what is baffling Furman. The household survey says that the number of employed people only increased by 179,000, and the employment to population ratio jumped just 0.1 percent. Where did all these workers come from in July if the participation rate did not rise and the total level of employment barely grew? Just as we do not know where the workers have gone, we do not know where they are coming from.

A Brief Tour of Recent Labor Force History

Travel restrictions, business closures, and COVID-concerns sent labor force participation crashing in the spring of 2020. We hit a low of 60 percent in April, but then rebounded quickly for the following three months. In the summer of 2020, however, the recovery stalled out. Employment grew rapidly through October but not participation.

Participation got a second rebound in the spring of 2021 when the vaccines were rolled out, and a few months later employment began rising rapidly again. Then we hit another air pocket, which may have been related to rising infection numbers and schools staying closed. The reopening of schools and a fall in infections led to a third surge that began in November 2021 and lasted through March of this year.

After the March peak, participation began to fall again. It’s notable that job growth peaked the month before that. Until this month, job growth as measured by the payroll survey was at its slowest pace in more than a year. Part of the retreat of participation may be simply that the pace of job growth decelerated.

fredgraph

Another weird thing started in March of 2020. Even though we were still adding jobs, the employment level as measured by the household survey flatlined. This is the survey that the Department of Labor uses to construct the unemployment rate, the participation rate, and other demographics about who is working. It is constructed by asking households about their employment situation. It differs in various ways from the payroll survey, which asks businesses, nonprofits, and governments about employment, hours, and wages. (You can read about the differences between the surveys here.)

Luring the Self-employed onto Payrolls

When you run the household survey alongside the payroll survey, you notice a few things. First, the household survey shows a higher level of employment than the payroll survey. That’s because it includes agricultural workers, self-employed workers whose businesses are unincorporated, people on unpaid leave from work, and unpaid family workers. The payroll survey only counts people who are on the payrolls of nonfarm establishments (it is also called the establishment survey).

fredgraph

Second, you can see that the household survey shows that employment growth came to a halt in March. Businesses were adding to payrolls, but the overall level of employment was not rising. The best explanation for this is that businesses are drawing in many workers from what the financial pundits call the sidelines; but instead of hiring people who weren’t working, they’re hiring the self-employed. Workers are being drawn onto the payrolls of outside employers from their own businesses. Also, some of the people who are dropping out of the labor force are being replaced by new entrants, keeping the employment level steady.

We can see this in the data for the self-employed in the household report. The number of self-employed people in July shrank by 279,000. On a five-year chart, we can see that self-employment crashed during the pandemic spring and peaked last August. It has been coming down rapidly since then, likely because so many workers are shifting from self-employment to payrolls.

fredgraph

That solves the secondary mystery of where all the workers we are adding to payrolls are coming from.

The decline of labor force participation lines up chronologically with the flatlining of the employment level, which suggests there may be a connection. Perhaps payroll growth on its own is not enough to encourage labor force participation to climb. It may be that the employment level as measured by the household survey is what really drives participation now. One possibility is that opportunities or rewards for self-employment are contracting, driving some workers onto payrolls and sending some out of the workforce altogether.

fredgraph

Real Wages Are Falling

Inflation may be playing a role as well. Although nominal wages have been rising, they have not kept up with inflation. So real wages are falling. That makes giving up leisure for work less compelling on the margin. Basically, employers are offering workers less money to work than they were before, which means that more people who do not absolutely need to work will choose not to. The self-employment gig may not be worth holding on to if your customers are not paying you as much in real terms, so you decide to hang up your boots, so to speak. Or perhaps you are a young person who decides to hold off on looking for paying work until the wages improve. In short, the real return on labor has fallen, so some people are choosing not to invest their time in it.

If it sounds absurd to say that people would choose not to work because prices are rising, keep in mind that unemployment is extremely low and the employment level at nearly an all-time high. People are far more likely to be related to a person with a job than any time in recent history, and those jobs seem very secure right now. This makes it easier for a small percentage of the population to decide not to work.

Keep in mind, the labor force shrank by only 63,000 in July, so we’re not talking about a mass exodus of workers.

Many good hypotheses offered in the comments. One of them: real wages have declined so would expect a shift along the labor supply curve. Should matter more for women and older workers who have higher elasticities. https://t.co/JH8lq0CoqG

— Jason Furman (@jasonfurman) August 5, 2022

The Role of Seasonal Adjustments

It’s also work keeping in mind that the labor force participation rate did not really contract in July, and payrolls did not grow by 528,000. Before the seasonal adjustments, the participation rate increased from 62.5 percent to 62.6 percent. The payroll figure contracted by 325,000. This is not nefarious. The headline figures you read are seasonally adjusted to smooth out variations in employment from month to month so as to give us a better perspective on the health of the economy. What the seasonal adjustments for July tell us is that the economy is expected to lose a lot of jobs in July, and this year it did not lose as much. On the chart on nonseasonally adjusted payroll growth below, the big dips are January, when payrolls shrink post-holidays, and the little dips are July. This year’s July payroll contraction was the second smallest since a 314,000 payroll contraction in 1966, exceeded only by last year’s tiny 41,000 contraction.

In other words, it was a very strong report for July even if you ignore the seasonal adjustment.

fredgraph

Similarly, the seasonal adjustment tells us that the labor force participation rate is expected to jump in July, which it usually does and did this year. But this year’s jump was smaller than typical, so it gets seasonally adjusted into a decline. The chart below shows the non-adjusted labor force participation rate. Each of the spikes is a July.

fredgraph

The decline in real wages coupled with very low unemployment has raised the opportunity cost of working. So the number of workers is not keeping up with population growth, sinking the participation rate and keeping the employment level flat even as payrolls are growing rapidly.