Ed Dolan's Econ Blog

Tracking Part-Time Work: Some Surprising Details and a Look Ahead

The share of part-time workers in the U.S. labor force rose to unusually high levels during the recession of 2007-2009 and has fallen only slowly during the subsequent recovery. As the following chart shows, the share of workers putting in less than 35 hours a week hit a low of 15.5 percent of the labor force in October 2007 and then rose sharply to a peak of 18.1 percent in May 2009. As of December 2013, the share of part time work was still over 17 percent, less than halfway back to the pre-recession low.

Many observers find the trend toward part-time work alarming. In her blog for The New York Times, Catherine Rampell writes,

One of the more unsettling trends in this recovery has been the rise of part-time work. We are nowhere near recovering the jobs lost in the recession, and the track record looks even worse when you consider that so many of the jobs lost were full time, whereas so many of those gained have been part time.

Elsewhere, a writer for the Minnesota Budget Project laments,

Many workers are struggling to climb out of the Great Recession. They are still looking for jobs, working part time, earning less than they did before the recession, or accepting jobs that don’t meet their abilities.

How worried should we be? To understand what is going on, we need to dig beneath the summary statistics shown in the chart and look the details, some of which I find surprising.

Trends in part-time work: some details

The numbers begin to look a little less alarming if we break down the total of part-timers into three categories used by the BLS:

  • People who normally work part time for “noneconomic reasons” such as childcare, family or personal obligations, school or training, retirement or Social Security limits on earnings, and other reasons. These are often called voluntary part-time workers.
  • Those who work part-time for “economic reasons,” popularly called involuntary part-time workers, a category that includes:
    •  People who are working part-time because of slack work or business conditions. By implication, many or most of these are people who took jobs that they expected to be full time, only to have their employers cut their hours. Presumably, they stay on the job either because short hours are better than none, or because they expect to return to full time status in the future. At least some of them presumably continue to receive full-time benefits during temporary periods of short hours.
    • People who can only find part-time work. By implication, these include people who take jobs that they know will never be full time and would leave them if they found full-time work.

In understanding these categories, it is important to realize that they measure part-time work, not part-time jobs. A person working 35 or more hours per week at two or more part-time jobs counts as a full-time worker as far as the BLS categories are concerned. The next chart shows the breakdown:

It is somewhat reassuring to see that even in the worst months of the recession, more than two-thirds of all part-time work is voluntary. Furthermore, more than two-thirds of the involuntary part-time workers hold jobs that carry some hope of a return to full-time hours if and when business conditions return to normal.

The kinds of part-time worker often featured in media reports—the laid-off autoworkers or schoolteachers who grasp at a few hours a week bagging groceries in order to buy shoes for their kids—belong to the smallest of the BLS categories. True, their numbers have doubled during the Great Recession, but “can only find part-time work” still accounts for less than 1.7 percent of the labor force. To look at it another way, there are four people who are altogether unemployed for each person who has reluctantly taken part-time work because that is all there is.

The preceding chart reveals some interesting details, but, because the absolute size of the different categories differs so greatly, it conceals others. Some surprises come to light if we replot the data to emphasize the differing dynamics of the three categories of part-time work over the course of the recession and recovery. The next chart does that by showing each category as an index with its value in June 2009, the month the recession reached its trough, set equal to 100. The trendlines help to bring out some key features:

The first surprise is that number of people voluntarily working part time fell as the economy slipped into recession. That decline continued through the first year of the recovery, and the number of voluntary part-time workers has still not returned to the prerecession level. The drop in voluntary part-time work seems to contradict two popular narratives. One is that financial stress caused by the recession has forced many people who otherwise would have retired to continue working at least part-time. The other is that secondary earners in households hit by financial stress have had to cut back on leisure activities or time devoted to household responsibilities in order to take part-time jobs they really don’t want. Yes, there are some people who find themselves in those circumstances, but evidently there are not enough of them to have caused an overall increase in the number of voluntary part-time workers.

The second surprise, at least to me, lies in the differing dynamics of the two types of involuntary part-time work. The downslope of the recession, from late 2007 to mid-2009, sent both kinds of involuntary part-time work soaring. However, once the recovery began, their paths diverged. Part-time work due to slack business conditions began to fall, but the share of workers who said part-time work was the only kind they could find continued to rise, and at best has only flattened out, even as the recovery has begun to gather steam.

I can see that one of the first effects of the recovery would be for employers to give their current workers more hours, bringing them back to full-time status. However, I would have thought that with a short lag, after their existing workers were back to full hours, they would have started hiring into new full-time positions. As more full-time jobs opened up, fewer workers would have to take part-time jobs that were not what they really wanted. Instead, it looks like either the lag is longer than I would have thought, or else employers are taking their existing workers back to full time but limiting new hires to part time.

How long will these trends continue? Let’s look at two hypotheses that could bear on the near-term future of part-time work.

Part-time work and effective marginal tax rates

One hypothesis links trends in part-time work to effective marginal tax rates. In this version of the story, trends in part-time work are driven by the preferences of workers and the incentives they face, with employers offering more or fewer part-time jobs to suit the needs of the labor force.

Casey B. Mulligan, a professor at the University of Chicago, explains the reasoning in a recent blog post. He begins by pointing out that choosing full- or part-time work is a matter of balancing money income against the value of what a people can do on their own time, whether that is unpaid work like childcare, volunteer work outside the home, or pure leisure. Critically, money income is subject to taxes but own-time is not. It follows that any increase in the effective marginal tax rate on money income will make part-time work relatively more attractive.

Mulligan then presents a little chart that shows the changing effective marginal tax rate. Between 2007 and 2011, he says, effective tax rates increased for many households, not because of changes in tax laws, but because of other policy changes. One was an expansion of the food stamp program (SNAP). SNAP makes part-time work attractive to people who would hit the program’s earnings ceiling and lose benefits if they worked full time. At the same time, rules for mortgage modifications  also changed.  Some homeowners found themselves in a position where earning more would decrease the amount by which federal rules would allow reductions in their mortgage balances.

In 2011 and 2012, another change came along, this time in the form of the payroll tax holiday, which, Mulligan says, largely offset the preceding changes in SNAP and mortgage rules. When the payroll tax reduction expired at the end of 2012, the effective marginal tax went rate up again.

Now, at the beginning of 2014, another large implicit tax rate is starting to hit many families, this time in the form of the insurance subsidies available under the Affordable Care Act (ACA). For many families, including some with incomes of as much as $100,000 per year, earning more will mean a reduction in subsidies. As another recent New York Times article pointed out, sometimes even a few more dollars earned can cost a family thousands of dollars in added insurance premiums. Those changes, again, should tip the balance toward part-time work.

As I understand it, a change in tax rates would affect mainly the voluntary component of part-time work. Higher implicit tax rates on incomes don’t cause employers to force workers into part-time jobs they don’t want. Instead, Mulligan’s hypothesis is that they tilt workers preferences toward part-time jobs, and employers react to the shift by offering more part-time work.

The trouble with this hypothesis is that it doesn’t seem to fit the data very well, at least not superficially. We can see this if we extract the voluntary component of part-time work from our previous chart, still indexed to June 2009=100, and modify the axes for easy comparison with Mulligan’s own chart. Here is the result:

According to Mulligan’s hypothesis, voluntary part-time work should have risen during 2007-2010, when effective marginal tax rates were rising. Instead, it fell during that period. The payroll tax holiday should have made part-time work less attractive. No such effect is apparent. It should have risen again in 2013 after the payroll tax holiday expired. Instead, it rose early in the year but fell back again.

The hypothesis that tax rates affect workers’ preferences for part-time work sounds plausible, so I am not ready to give up on it completely. Maybe a sharper definition of marginal tax rates or a deeper econometric analysis that took additional variables into account would confirm it. I hope Mulligan or others will continue this line of inquiry. Meanwhile, on to the next hypothesis.

Part-time work and the Affordable Care Act

Another much-discussed hypothesis blames the trend toward more part-time work on a different provision of the ACA, the employer mandate. According to that provision, employers with more than 50 workers will have to provide healthcare benefits to their full-time employees or face a penalty. In principle, they could avoid the penalty by shifting workers to part-time status.

Late in the game, the Obama administration delayed implementation of the employer mandate from 2014 to 2015. That delay has muddied the data. Even so, many people argue that delay or no, the ACA is already having an effect as employers proactively adjust their personnel policies. For example, a report  CNBC posted in November 2013 had this to say:

Many business leaders say they are already making personnel decisions based on the Affordable Care Act. . . Among franchised businesses, 27 percent report their company has replaced full-time workers with part-time workers and 31 percent have reduced worker hours. Among non-franchised businesses, 12 percent are replacing full-time workers with part-time workers or reducing hours. This is happening now, with more than a year before the mandate goes into effect; and undoubtedly, these numbers will rise as we approach next July’s “look back” period for tabulating workers’ hours.

Some sources, not content with generalizations, list specific businesses that have cut back or said they will cut back on full-time work. (Examples here and here.)

Other observers vigorously dispute this conclusion. For example, a study from the Center for Budget and Policy Priorities concludes “recent data provide scant evidence that health reform is causing a significant shift toward part-time work, and there’s every reason to believe that the ultimate effect will be small as a share of total employment.”

Anecdotes aside, it is hard to find any evidence of an ACA effect in the data for 2013. Presumably, the impact of the ACA would be on the involuntary component of part-time work. At first, it might show up as people working part-time because employers have cut back their hours. Over time, it would increasingly take the form of more people working part-time because that was all they could find. However, the above charts show no upturn in either of the involuntary part-time categories in 2013. In fact, both decreased slightly as a share of the labor force.

The Bottom Line

We can be certain that public figures like Senator Ted Cruz will continue to point to “millions of people who are hurting . . . who are losing their jobs, being forced into part-time work” as one of the major ills facing the U.S. economy. In reality, though, the data just don’t justify that kind of alarmism. The vast majority of part-time work is voluntary. The kind of part-time work most talked about—people who are working part time because that is the only work they can find—is a tiny part of the labor force, just  1.6 percent and holding flat or falling slightly in 2013. Still, things could change. It will be worth tracking trends in involuntary part-time employment closely over the coming year. Watch this space for new data as they come in.

9 Responses to “Tracking Part-Time Work: Some Surprising Details and a Look Ahead”

Jed_GrahamJanuary 15th, 2014 at 11:45 am

It is inappropriate to look at aggregate workforce data to gauge a trend that will primarily impact low-wage workers.
Here are 2 charts that provide a much better window into how these workers are being impacted.
One shows the drop in hours worked across a range of low-wage sectors, matching up with anecdotes about cut to work hours. The second shows how the low-wage workweek has sunk back near the record low seen at the depth of the recession, while the workweek for higher wage earners has fully recovered.

Ed Dolan EdDolanJanuary 15th, 2014 at 3:16 pm

Thanks for the interesting links. I agree with you that the more fine-grained our view of part-time work, the more we learn

Your second link raises an interesting point that I did not address in the post. It quotes AFL-CIO head Richard Trumka as (apparently) supporting Congressional initiatives to raise the definition of full-time work to 40 hours. The study I cited in the post from the Center for Budget and Policy Priorities ( ) argues that doing that would increase, rather than reduce, the incentives of employers to cut hours. What is going on? Is CPBB wrong? Is Trumka just mistaken? Or is he cynically calculating that the workers whose hours would be cut would mostly be nonunion, so that the 40-hour rule would give union employers a competitive advantage over their nonunion rivals?

Jed_GrahamJanuary 15th, 2014 at 3:30 pm

Trumka opposes 30-hr week because he wants onus for coverage to be on employers and 30 hours gives them a loophole with regard to part-time work.But i seriously doubt he supports raising threshold to 40 hours. If anything, he'd rather lower it to 20 as in Hawaii.I have proposed eliminating the mandate below $10/hr and phasing it in between $10 & $20/hr.

Jacob LeeJanuary 16th, 2014 at 9:24 am

I would like to make two points in response to the interpretation of your second chart. Firstly, public sector employment is predominately full-time, and a substantial number of the jobs lost and not recovered have been public sector jobs, especially at the state and local level, and job losses in this sector continued to occur even after the private sector began to recover; this could partially account for the continued increase part time employment due to unavailability of full time jobs rather than slack business conditions.

Secondly, the decrease in voluntary part time employment might partially be explaibable as an artifact of changing subjective orientations toward their work: for example, if a voluntary part timers spouse lost their full time position, the part time worker may no longer consider their part time status voluntary, since a full time position would be more desireable, and the unemployed spouse is available for childcare etc

rjsigmundJanuary 16th, 2014 at 2:16 pm

ed, some time ago – maybe september – i queried you about the divergence between unadjusted non-farm payrolls and the unadjusted count of the employed from the household survey, giving you the numbers i had at that time…you reassured me that within a month or two the two would even out…well, they havent…

from July to December, the unadjusted count of the employed from the household survey fell by 690,000, from 145,113,000 in July, to 144,423,000 in December…over the same time frame, the unadjusted non-farm payrolls rose by 2,176,000, from 135,577,000 in July to 137,753,000 in December…

here are the unadjusted payroll jobs monthly (000's):
2013-07-01 135577
2013-08-01 136002
2013-09-01 136612
2013-10-01 137523
2013-11-01 137999
2013-12-01 137753
here's the raw unadjusted count of the employed from the household survey (000's):
2013-07-01 145113
2013-08-01 144509
2013-09-01 144651
2013-10-01 144144
2013-11-01 144775
2013-12-01 144423

Ed Dolan EdDolanJanuary 17th, 2014 at 6:57 am

Thanks for the data and the chart. As you can see from the chart, the payroll jobs data and the household survey employment data do not track closely on a month to month basis. There are several reasons for that. First, they are separate surveys, each subject to sampling error and revisions, but different kinds of sampling errors and revisions. Second, the household survey includes self-employed and farm workers, the payroll survey does not. Third, the employment numbers count how many people are employed whereas the payroll numbers count jobs, so that one person with two jobs is double-counted in the payroll survey but not the household survey. There are some other methodological differences, too.

The latest data show a fairly large discrepancy between the two series, but it is not outside the bounds of what we have seen in the past. For the whole period your chart reports (2000-2013) the average difference between the two series, without regard to sign, is equal to 2.81 percent of the sum of the two series. The latest observation, for December, is a little smaller than that, 2.36 percent. The average difference for all of 2013 is 2.86 percent, only slightly larger than the 13-year average.

rjsigmundJanuary 17th, 2014 at 8:25 am

i had considered that seasonal farm work was part of the explanation…but we still have a situation where payroll jobs are running far ahead of the employed from the household survey, even seasonally adjusted, which hasnt been accompanied by an increase in multiple job holders…

notice the chart brings the two unadjusted surveys into approximate alignment…if we focus on the last five monthly changes, we can see that the red payroll jobs line is well above the blue household employed for the first time in years….the last time there was such a large increase in payroll jobs that was not accompanied by a relatively equal increase in the number of employed was in 2007, just before the recession started…i'm not suggesting that this predicts a recession, but that it’s inconceivable for these two lines, which are in effect measurements of the same function of employment, to continue moving in opposite directions indefinitely…