Authors
Article Highlights
- Increases in unemployment are not typically a result of increases in layoffs
- As it becomes more difficult to find jobs, more laid-off workers end up nonemployed
- In cooling labor markets, laid-off workers are more likely to stay in the labor force and be counted as unemployed.
Starting in January 2023, the U.S. unemployment rate steadily crept up through July 2024, before edging down in recent months (blue line in Figure 1). At 4.1 percent, the unemployment rate currently remains 0.7 percentage points above the January 2023 value of 3.4 percent. Strikingly, this increase has occurred without a discernable rise in layoffs (gold line). Media commentaries often note that both rising unemployment rates and rising layoff rates are indicators of a cooling labor market. These rates are expected to move together because unemployment reflects a scarcity of jobs and layoffs reflect employers cutting jobs. But recently, the two rates have diverged. So, what signals are we seeing in the labor market?
In this note, we provide a fresh look at labor market data, arguing that recent developments are in line with historical patterns around labor market cooling. Our main finding is that increases in unemployment are typically not due to increases in layoffs; rather, they happen because laid-off workers are less likely to quickly find a new job, more likely to stay in the labor force, and thus more likely to join a growing pool of unemployed people hunting for work.
A new dataset provides new labor market insights
In July 2024, we introduced the QLmonthly, a new dataset, updated monthly, containing time series on quits and layoffs resulting in nonemployment. The data are computed from the Current Population Survey (CPS) as described in our 2024 paper “Quits, Layoffs, and Labor Supply.” The CPS data offer a perspective not seen in the most-often-used series on quits and layoffs, the Job Openings and Labor Turnover Survey (JOLTS). Whereas the JOLTS tracks what happens to a job, the CPS tracks what happens to people. The CPS data have two advantages we exploit in our research. First, the data series extends back further, to 1978, while the JOLTS begins in 2000.1 Second, the CPS adds information about what happens to a worker after a layoff: Do they become unemployed or exit the labor force? Notably, CPS layoffs capture layoffs to nonemployment; laid-off individuals who quickly find another job are not included.
Leveraging the nearly five decades of CPS data, Figure 2 shows that the current puzzling situation is, in fact, quite normal: The joint movements of unemployment rates and layoff rates we have seen since January are well within normal historical values. Further, although the layoff rate and unemployment rate are strongly correlated—that is, they go up and down together—increases in the unemployment rate are typically accompanied by much smaller increases in layoff rates. Indeed, the dots fall well below the 45-degree line, consistent with a more than 1-to-1 relationship. Hence, factors other than increases in layoffs usually drive up unemployment.
It may seem puzzling that layoffs in our CPS data have risen recently alongside the rise in the unemployment rate (the gold dots in Figure 2) given the little movement in layoffs in the JOLTS data and the large focus of media on these developments. The key difference between the two data sources is that our CPS-based series measure layoffs that result in a worker becoming nonemployed, whereas the JOLTS data measure all layoffs. As we discuss later, times in which these series diverge are times when jobs are harder to find, indicating a decline in labor demand.
Introducing labor supply: Fewer laid-off workers leave the labor force
As discussed in our research paper, the growing pool of unemployed workers—without a commensurate rise in layoffs—is a combination of two forces. The literature following the research of Robert Shimer in 2012, such as a recent analysis by the Minneapolis Fed, focuses on firms’ labor demand. Falling labor demand lowers the job-finding rate, which means that unemployed workers stay unemployed longer and laid-off workers are less likely to move quickly to a new job. A second, less explored channel, is labor supply. Since our data track people and not just jobs as in the JOLTS, we now can see that laid-off workers are less likely to exit the labor force (not actively search for work) as the economy cools and more likely to transition into unemployment (actively search for work). This labor supply channel provides important information about the state of the labor market.
Figure 3 shows that the share of laid-off workers who remain in the labor force increases strongly in recessions. This observation points to a fundamental change in labor force participation over the business cycle. Generally speaking, workers who aren’t able to quickly find a new job after they are laid off can either keep searching and join the ranks of the unemployed, or they can stop searching and exit the labor force. Figure 3 shows that that choice is strongly influenced by the state of the economy and labor market conditions, with more workers deciding to keep searching in recessions.
Notably, only workers who actively search for work and do not exit the labor force are counted as unemployed, and so through increased job search, the same number of layoffs increases unemployment during a cooling labor market. To quantify this point, Figure 4 makes use of our CPS data to assess the contribution of workers’ increasing participation to the rise in unemployment in prior recessions. The first row reports the actual increase in unemployment (in percentage points) in those episodes. The second row reports how much unemployment would have risen if labor force exit rates had not changed. The third row reports the contribution to the rise in unemployment due to changing labor supply in terms of higher participation of laid-off workers.
According to our data, changes in participation had a meaningful contribution to rising unemployment in prior recessions. On average, Figure 4 shows that the participation margin accounts for about 30 percent of the increase in unemployment.2 In some recessions, this margin explains as much as half of the increase in unemployment.
| 1980 & 1981–82 | 1990–91 | 2001 | 2007–09 | 2020 | |
|---|---|---|---|---|---|
| Actual increase in unemployment (ppts) | 4.9 | 1.6 | 1.6 | 4.8 | 11.3 |
| Increase in unemployment at constant labor force participation (ppts) | 3.0–3.5 | 1.1–1.2 | 0.9–1.0 | 3.4–3.6 | 4.5–5.1 |
| Contribution of higher participation | 29%–38% | 29%–31% | 36%–42% | 25%–29% | 55%–60% |
The recent increase in labor force participation is also consistent with historical episodes of labor market cooling and workers’ perception of a riskier labor market. Our data indicate that 64 percent of laid-off prime age (ages 25-54) workers remained in the labor force in January 2023 through September 2024, up from 59 percent in 2021–2022. Being able to now see how participation is moving deepens our understanding of the current situation. Given that the share of prime age adults in the labor force is the highest it has been since prior to 2000, one might suspect that participation during this cooling episode could behave differently. So far, we have seen this is not the case.
Concluding thoughts
The recent rise in the unemployment rate without a discernable increase in layoffs in the JOLTS data may appear a bit confusing. Using a newly developed dataset that tracks workers rather than jobs, we argue that layoffs to unemployment have actually increased over the past year or so, in line with prior labor market cooling episodes. This pattern reflects developments in terms of slowing labor demand (the job-finding rate drops when the labor market cools) and increasing labor supply (laid-off workers are more likely to remain in the labor market and enter unemployment as the labor market cools).
Against the backdrop of solid U.S. economic growth, such labor market cooling has been relatively smooth to date and consistent with a return to normal business conditions. Nonetheless, our data suggest workers appear more reluctant to leave the labor force if they are laid off, indicating that they perceive higher employment risks than before. Going forward, we advise paying less attention to changes in JOLTS layoffs and look for further changes in the rates at which laid-off workers find new jobs and at which they decide to participate in the labor market. Historically, large movements in these rates have signaled sharp changes in labor market conditions and are encapsulated in our CPS layoff series.
Endnotes
1 Researchers have extended the job openings component of JOLTS (Regis Barnichon, “Building a composite Help-Wanted Index,” Economics Letters 109.3 [2010]: 175–178; and Nicolas Petrosky-Nadeau and Lu Zhang, “Unemployment crises,” Journal of Monetary Economics 117 [2021]: 335–353). Researchers have used data on establishment changes in unemployment to approximate quits and layoffs in the past, but these are not direct measurements (Steven J. Davis, R. Jason Faberman, and John Haltiwanger, “Labor market flows in the cross section and over time,” Journal of Monetary Economics 59.1 [2012]: 1–18).
2 The lower bound assumes that marginal workers have the job-finding rate of the unemployed. The upper bound assumes that they have the job-finding rate of nonparticipants. For details, see Kathrin Ellieroth and Amanda Michaud, “Quits, Layoffs, and Labor Supply,” Institute Working Paper 94, Federal Reserve Bank of Minneapolis, October 2024, https://doi.org/10.21034/iwp.94.
