I would like to acknowledge and thank Mr. Steve Hipple, Economist at the Bureau of Labor Statistics, for his generous assistance in assembling the data used in this post from the public-use micro data files of the Current Population Survey. This post would not have been possible without his help.
There has been much discussion in recent months about workers resigning from their jobs at record high levels. This has often been attributed to workers reassessing their lives and deciding their jobs are simply not worth it. A new name has even been coined for this: the “Great Resignation”.
But while resignations have indeed been high, two quite distinct matters have often been confounded. One is workers resigning from a position in order to move to a new, more attractive and usually higher-paying, position with a different employer. The other is workers resigning from a position with no intention to take a new job, but rather to leave the labor force and do something else. The former reflects a reshuffling in the economy, with workers moving to positions where they will likely be better paid and more productive. This should raise the overall productivity of the economy. The latter (those leaving the labor force) would reduce the overall capacity of the economy, if significant. But as we will see below, while quits from jobs in order to move to a new job is, indeed, at record high levels, the number quitting in order to drop out of the labor force is at this point quite modest, and likely also to prove temporary. While the Covid pandemic led to a major shock in the labor market, previous trends in labor market participation rates are reasserting themselves.
This post will look at the data on each of these two issues – both important but also both quite different. It will start with the figures on turnover in the labor market, and present these figures in the context of the net number of new jobs being created. Quits are high, but hiring is also at record highs. Workers are quitting their jobs largely to switch to more attractive jobs.
While far more modest, some workers have, however, left the labor market. The second part of this post will look at the reasons given by those not in the labor force for why they are not, and how this has changed from before the pandemic hit. This is based on original data assembled from the public use micro data files of the Current Population Survey (CPS). While publicly accessible by scholars and researchers, these figures are not presented in the regular monthly reports of the Bureau of Labor Statistics on the CPS. This data will hopefully serve to better inform the discussion on what has been termed by some as the “Great Resignation”.
We will see that the changes in the number of US adults deciding whether or not to participate in the labor force are now modest compared to pre-pandemic trends, and are mostly accounted for by older workers deciding to retire earlier than what would have been expected, on average, under previous patterns. But to the extent some worker decides to retire now, a year or two earlier than when they had earlier planned, there will then be one less worker retiring a year or two from now. That is, there will not be a long-term impact, and one should expect to see a return to previous trends. And so far, that is precisely what we have been seeing.
B. Quits, Job Openings, and Net New Jobs
The number of workers quitting their jobs each month has indeed risen – and to the highest levels of at least two decades (the data do not go back further). But the number of workers being hired each month to fill open positions has also increased – to even higher levels. And despite the record pace of hiring, the number of open jobs employers are seeking to fill has grown to especially high levels.
The figures are shown in this chart:
The data come from the Job Openings and Labor Turnover Surveys (JOLTS) of the Bureau of Labor Statistics (BLS), a monthly survey of employers (although with reports that lag one month compared to the more closely watched monthly BLS report titled “The Employment Situation”, with its figures on such estimates as the unemployment rate and on the net number of new jobs in the economy). The JOLTS surveys are relatively recent, with data going back only to December 2000, in contrast to the CPS, which goes back to 1948. The chart here is shown in terms of the absolute number of workers or jobs in each group.
[Side Note: One might sometimes see a chart similar to this but shown in terms of rates: Hires and Quits shown as a percentage share of the number employed, and Job Openings as a percentage share of the number employed plus the number of job openings. However, for the relatively short period here (21 years) the patterns in the two presentations look very much the same,]
The number of “Hires” are the number of workers added to the payroll in the given month, according to this survey of employers. Employers are also asked how many workers left the payroll (“Total Separations”) and whether they were workers who left voluntarily (“Quits”), were laid off or discharged involuntarily (“Layoffs & Discharges”), or left for some other reason (“Other Separations”). The BLS includes in the Other Separations category those who left to go into retirement, or due to a new disability, or due to deaths. Hence quits are only one reason for workers leaving their jobs, although its share has been growing: Layoffs & Discharges have been falling, while the number in the “Other Separations” category has been flat and relatively low. (These latter two categories were not included in the chart to reduce the clutter.)
Hires and the various categories of separations are all flows, measured by the BLS over the course of a full month (and then seasonally adjusted, which among other effects will compensate for the different number of days in different months). The “Job Openings” estimate, in contrast, is a stock, reporting the number of open job positions the employer is actively seeking to fill as of the last business day of each month. Its scale on the chart therefore should not be taken as directly comparable to the number of Hires or Quits on the chart, which are flows over the course of a one-month period. While they happen to be similar in number, one could have reported the number of Hires or Quits over, say, a two-month period (in which case they would have been about twice as much). One needs to remember that stocks and flows are different.
As the chart shows, open jobs that employers are seeking to fill (“Job Openings”) have grown sharply over the last year. While the monthly rate of hires has also grown – to record levels – the hiring could not keep up. And with more workers being hired and actively recruited to fill the open job positions, it should not be at all surprising that the number of workers quitting their old jobs to take a new job – a job that is more attractive to them that probably also pays more – has also been increasing. Thus there are resignations, but not to leave the labor force. Rather, workers are resigning to switch to a new, more attractive, job.
Such “churn” in the labor market is a good thing. Not only are workers moving to what is for them a more attractive (and likely higher-paying) job, but the productivity of the economy as a whole will also go up as a result. Employers are able to pay more to attract the workers to these jobs because the workers hired into those jobs will likely be producing more than they had in their old jobs.
How do we know that the quits were largely in order to move to a new job? It is clear from the magnitudes. The number of quits in the JOLTS data from March 2020 through February 2022 totaled over 85 million over the two-year period. And this does not even include those quitting in order to retire (they are included in the “other” category in JOLTS). Yet as will be discussed in the next section below, the labor force in February 2022 totaled only about 2.7 million less as of February 2022 than what would have been the case had the pre-pandemic shares of participation in the labor force continued. And close to three-quarters of that 2.7 million reduction was due to workers entering into retirement at somewhat greater rates than was the pattern before. This is nowhere close to the 85 million quits over the period.
One can also compare the monthly averages for the labor turnover figures with the net figures for new jobs:
The chart shows the average monthly figures for 2021, all from the BLS (either from JOLTS or the CPS). January figures are excluded as the BLS changes each January the population controls it receives from the Census Bureau for its CPS figures, without revising earlier estimates. This can lead to an abrupt one-month change in January, making it not comparable to the changes found in other months.
The first three columns show the average monthly growth in 2021 in the adult population (117,000), in the labor force (192,000), and in the number of net new jobs (547,000). Over the long term, the labor force cannot grow faster than the adult population, but it did in 2021 as the labor force participation rate rose in 2021 following the turmoil of 2020. And the net number of new jobs could grow faster in 2021 than the increase in the labor force as the number of unemployed fell rapidly in this first year of the Biden administration. But the economy is now at full employment, and unemployment will not be able to fall much further. Thus over the longer-term one cannot expect the net number of new jobs to grow faster than the increase in the labor force, and one cannot expect the labor force to grow faster than the adult population (and indeed normally by substantially less, as not all adults choose to be part of the labor force).
In contrast to the figures seen in the first three columns, the average monthly number of workers hired is far higher. So is the number of separations, and it is the relatively small difference between the number of workers hired into positions and those separated from them for whatever reason that equals the number of net new jobs in the economy. The separations in 2021 mostly came from quits (70% of the total), with smaller numbers from layoffs or discharges and from the “other” category (where, as noted before, the BLS includes those choosing to quit due to retirement).
All this is consistent with a very strong labor market. Workers are indeed resigning, but this is largely due to the opportunity to move to a more attractive, better paying, open job. As we will discuss in the next section, relatively few are resigning to leave the labor force altogether.
C. The Extent to Which the Labor Force Fell, and the Factors Behind It
As of February 2022, there were 592,000 fewer US residents in the labor force (in the seasonally adjusted figures) than there were in February 2020, just before the lockdowns due to Covid began. This is not much: Just 0.4% of the labor force. But it is not a fair comparison. The adult population grew over those two years, and thus one would expect that in normal circumstances, the labor force would also have grown. The question is by how much. For this one needs to construct some counter-factual scenario of what the labor force would have been (in normal circumstances) and compare that to what it in fact was (given the consequences of Covid) to see how much of a change there was. Is there evidence here for a “Great Resignation”, of people leaving the labor force in high numbers?
A simple and reasonable counterfactual would be to assume the labor force (in a breakdown by individual groups based on gender and age) would have grown in the absence of the crisis at the same rate as their population. Population growth is determined by long-term demographics. That is, in this scenario it is assumed that the rates at which those in the individual demographic groups chose to be part of the labor force (the labor force participation rate) would have remained the same as what they were in February 2020. Similarly, the rates of those choosing not to be part of the labor force would be the same as in February 2020 (it will simply be one minus the labor force participation rates), and similarly for the reasons given for not participating in the labor force (e.g. retirement, home or family care, full-time students, disability, and so on). One can then compare changes in the labor force and in the numbers not in the labor force (by the reasons given for this), under a scenario where the participation rates in February 2022 were the same as they were in February 2020, to what they actually were in February 2022.
The households surveyed in the monthly CPS are asked, when they respond that they are not employed and have not been actively seeking a job, the major reason for why they are not in the labor force. However, the BLS monthly report on the findings of that month’s CPS survey does not report these reasons. The monthly report is already pretty long. However, one can obtain these results from the CPS public-use micro data files on the CPS. The results reported here come from those files (and were assembled by Mr. Steve Hipple of the BLS for this post).
The basic results for the whole population, and for men and all women separately, are summarized in this chart:
Had the participation rates remained the same as in February 2020, there would have been an extra almost 2.7 million workers in the labor force in February 2022. The labor force would have been 1.6% higher than what it was. While significant, I would not see this as qualifying as a “Great Resignation”.
[Technical note: The calculations for those in the labor force and those not in the labor force (by reason) were worked out first for the most basic groups examined: men and women, each in three different age groups of ages 16 to 24, 25 to 54, and 55 and above, for a total of six groups. The aggregations for all men or all women, for both men and women in each age group, and for everyone together, were then calculated by summing over the relevant groups.]
Almost three-quarters of the 2.7 million reduction (2.0 million, or 73% of the total) reflected a higher share of adults choosing to retire. This is consistent with the story that with the disruption in the last two years, coupled also with significant income supplements being provided to most households through the various Covid relief measures passed by Congress during the administrations of both Trump and Biden, a significant number of workers decided to retire earlier than they had previously planned. It might be a year or two earlier, or possibly longer. The implications of this are important, as it implies that the changes in the labor force will be temporary rather than permanent. One more person retiring now, earlier than they had previously planned, means there will be one less person retiring at whatever that future date was to have been.
The second most important reason for leaving the labor force was to take care of home or family, with this accounting for 582,000 workers – 22% of the total reduction in the labor force in the scenario being examined. This is also understandable in the context of the Covid crisis. Many workers had to leave the labor force during the midst of the crisis to take care of school-age children when the schools were closed, but almost all schools are now once again open (albeit with some occasional disruption due to Covid outbreaks). There might also have been a need to take care of family members who became sick during the crisis with Covid itself, and that might still have been a factor in February 2022 (as the Omicron wave subsided). To the extent this has been Covid driven, these effects should also prove to be temporary as the Covid crisis recedes.
There are, in addition, a list of other possible reasons given in the CPS survey for not participating in the labor force (such as full-time studies as a student, disability, illness, and a catch-all “other” category). In the aggregate the difference these made in the scenario being examined was small: only 138,000 – or only 5% of the total reduction in the labor force in this scenario.
In terms of the gender breakdown, more women than men left the labor force in the given scenario (1.7 million women vs. 1.0 million men) even though the share of the labor force made up of women (47% in 2022) is less than the share made up of men (53%). The shares of this due to more entering retirement or for taking care of home or family are broadly similar between men and women, which is perhaps surprising. Indeed, the share reporting that they are not in the labor force due to home or family care was somewhat higher for men (25.2% of their total) than for women (19.4%), but it is not clear whether such differences should be considered significant. The underlying data comes from surveys, there will be statistical noise, and these figures are all based on changes between what the February 2022 levels were and what they would have been in a scenario where we assume the February 2020 participation patterns had remained.
The figures broken down by age group were:
The largest single cause leading to lower participation in the labor force (in the scenario where prior patterns would have remained) was an increase in the share of retirees among those aged 55 and above. This accounted for 1.5 million workers, which was 3.9% of adults in this age group. Surprisingly (at least to me) was that there was essentially no difference in this age group of those who were not in the labor force due to home or family care.
Among prime-age workers (ages 25 to 54) there were roughly similar shares among those no longer in the labor force who gave as their reason retirement or for home or family care. The total number no longer in the labor force (relative to the scenario being examined) was also relatively small for this 25 to 54 age group, at just 0.9% of the population in the age group. The share no longer in the labor force in the group aged 55 and above was substantially higher, at 3.1% of the population of that age group. This is as one would expect when the primary factor behind those leaving the labor force was early retirement.
The share of the youngest age group (ages 16 to 24) no longer in the labor force fell by 2.6%, but primarily here for reasons lumped into the “all other” category. The largest single factor here was full-time studies, but this accounted for just 144,000 of the 414,000 (about 35%) in this “all other” category. One should also note that while there is a small number in the “retired” category (19,000), this is probably just a reflection of the fact this is a survey. Respondents do not always fully understand the nature of the questions or may have been in some unusual circumstance that does not fit in well with any of the listed possible responses.
Graphically, how much of a difference has it made? Not much. In terms of the labor force participation rates, one has for men and for women, as well as overall:
And by age group, as well as overall:
The “X” on each category shows where the labor force participation rates would be had the February 2020 rates (for the underlying groups of men or women by each age group) continued to hold. There was certainly a large shock to the system at the start of the pandemic, with the lockdowns that suddenly became necessary in March 2020. There was then a partial bounceback, followed by a leveling off but with a continued but slow recovery to the earlier patterns of participation rates. While still not fully back to what they were, the difference is now relatively modest.
This return to previous patterns in the participation rates is likely also to continue. With the single most important factor (almost three-quarters of the total) being people retiring earlier than what they had planned (or to be more precise, earlier than in the observed pattern in prior years, before the pandemic), the labor force numbers should be expected to return to their previous path in a few years. As noted before, if some worker retires a year or two earlier than they had earlier planned, then there will be one less retirement in a year or two (as that worker is already retired). This is consistent with the observed slow return to previous labor force participation rates.
The number of workers quitting their jobs has been high. But the quits are not a reflection of workers dropping out of the labor force. Rather, quits have been high as workers quit one job to move to another job – more attractive and likely better paying. Hires have also been exceptionally high. And despite the high rate of hiring, employers could not keep up and the number of open jobs they have been seeking to fill has grown. While some workers have left the labor force during the disruptions of the Covid pandemic, about three-quarters of this (as of February 2022) stemmed from a somewhat higher share of workers choosing to retire. But unless there has been a permanent change in retirement patterns (and there is no indication that there has been), decisions during the pandemic to retire earlier than previously planned will be self-correcting.
The high level of quits reflects, rather, an extremely strong labor market. Indeed, the number of net new jobs created in 2021, the first year of the Biden administration, came to 6.7 million – the highest in any one year in US history. (To be fair one should also note that the fall in the number of jobs in the US in 2020, the last year of the Trump administration, was also the highest in US history. Thus the Biden record was made possible by the low starting point.) With this strong labor market, workers have more of an opportunity to move to jobs that can make better use of their talents. And they have taken advantage of this opportunity, which will be a boost both to the workers and to productivity in the economy as a whole.