If you have been walking lately anywhere between left and right, you surely have heard the salvos about the number of new part-time jobs flying constantly over your head. One side believes that the recent limited increase in employment has been overwhelmingly concentrated in part-time jobs, surely because employers are trying to circumvent the coverage mandate in the Patient Protection and Affordable Care Act (known for short as the ACA, or sometimes as “Obamacare”). The other side believes that the recent moderate job growth has followed generally the historical division between full- and part-time work, adjusting for the economy’s current position in an arc of recovery from the extraordinarily deep economic downturn following on the financial crisis – and that the ACA has had absolutely no influence.
My own conclusion is that both extremes are trying too early to answer the wrong questions on the basis of tortured data. About the most I would feel confident in saying at this time is that employment in general, and full-time employment in particular, still suffer from the after-effects of the historically severe recession, but that current healthcare policy will not help its future recovery – which is why six years ago CED proposed fundamental changes in our nation’s health insurance system that would have been very different from the law that was enacted in 2010.
Background: Under the ACA, firms with at least 50 full-time employees, where “full-time” is defined as working at least 30 hours per week, must provide coverage to all of their full-time workers, or pay a penalty. This is the “employer mandate” much discussed in the ether. Thus, firms on the size borderline might reduce their full-time employee count below 50 by cutting current workers’ hours to below 30 and thereby avoid the coverage mandate entirely; and larger firms might avoid covering particular workers by cutting their weekly hours (or by hiring new workers at fewer than 30 hours per week). As you know, the effective date of the employer mandate has been postponed from 2014 to 2015.
In recent weeks and months, several pundits have argued from the Bureau of Labor Statistics’ data that an enormous share of employment growth in the current recovery has arisen from part-time jobs. One story claimed that there were either 35 or 110 new part-time jobs for every new full-time job created over the last six months. Another put the figure at 4.3 part-time jobs per full-time job in the first six months of this year. A third said that 96 percent of new jobs this year were part-time – so a ratio of about 24 to one. All assigned this mass movement toward part-time jobs to employer behavior under the impending mandate of the ACA.
With 144.5 million persons employed in August of 2013 (from the not-seasonally adjusted data), and only 25.4 million – 17.6 percent – of those persons working part-time, such extreme ratios of part-time job creation might seem unlikely, though not impossible. Here are some factual points to help you to decipher this dialog:
1. The data on part-time jobs come from the monthly employment survey of households, which is inherently less reliable than the survey of employers. As you probably know, the unemployment rate and related figures that are reported each month are calculated from a survey of households. The more frequently referenced total number of jobs in the economy and the numbers in each industry are calculated from a much larger survey of employers. The results drawn from the household survey have much more month-to-month variability than those from the employer survey (though the latter is far from perfect). This variability can come from technical problems in choosing and executing the sample and in people’s conscientiousness in answering the questions, but a part comes from purely random sampling variation that is inevitable when data come from only a part of a total population. (Sampling variation, of course, can be reduced by enlarging sample size, which of course entails greater cost – budget-sequester watchers, take note.)
To give a sense of scale, there is a 10-percent chance that the true number of unemployed persons differs from the result of the household survey by 300,000 or more. In comparison, the 10-percent margin of error for the number of employed persons in the employer survey is 90,000 – less than one-third as much. The greater sampling reliability of the employer survey is one reason (but not the only reason) why, over a long period of time, people have come to focus more on the new-jobs number from the employer survey, and less on the unemployment rate, as a timely economic indicator.
2. Another reason why comparisons of full- and part-time jobs and workers over short periods of time can show extreme values is the wide seasonal variation in part-time work. Temporary part-time hires are much more common in the summer and in the Thanksgiving-to-New Year holiday shopping season. At any time when such seasonal factors play, comparisons over time – sometimes even for one calendar month in one year and the same calendar month in the next – are subject to considerable error. And large percentage fluctuations in the actual numbers in the economy, with a little sampling error thrown in, can lead to enormous fluctuations in the measured data.
Given this reality, it is not terribly surprising that there has been a rash of stories about a surge in part-time work based on the employment survey for this August – a month when part-time work is much more common than in most other times of the year.
The employment data are available “seasonally adjusted.” But even the professionals at the Bureau of Labor Statistics would be quick to admit that seasonal adjustment is an imperfect art at best. (So inspired, my stock answer when a non-economist friend asks me how I am doing is to say that “seasonally adjusted, I am just fine.”) For example in this particular context, consider that over the last several years, school openings for the fall have begun to move earlier in the calendar, from September into August. Therefore, the history of August data on student part-time jobs may not offer a consistent benchmark for seasonal adjustment. Compounding such problems is the particular timing of a sample period within a given month, which can lead to spurious results. In one year, the August sample week might fall just before students generally leave the jobs and return to school; in another year, it might fall just after. Those phenomena might not only distort the sample results in one particular year; they might also corrupt the historical base on which seasonal adjustment procedures are calibrated in every year.
3. At least one potential problem cited by ACA critics is not even represented in the employment data, because the household survey asks individuals how many hours they work in total – combined over all of the jobs they may have. One criticism of the ACA is that in inducing employers to hire part-time instead of full-time workers, it has forced some individuals to cobble together two or more part-time jobs to make ends meet (but, of course, not to provide health insurance). So to the extent that individuals are so afflicted, and to the extent that their multiple part-time jobs combine to at least 35 hours per week (the employment survey’s definition of full-time work), then they will be classified as full-time workers, and will not have been included in the part-time count.
4. Not all part-time work is involuntary. Some people choose part-time work to match their other needs. The employment survey asks people whether they are working part-time “for economic reasons” – generally because their employers have cut back their hours, or because they cannot find the full-time jobs that they would prefer. The critiques of the ACA’s supposed effects that I have seen have not acknowledged this distinction or made use of the available data.
So to review the bidding, the current debate is over a change in the number of part-time workers – not part-time jobs – between this past January or February – months just after part-time holiday workers typically already have returned to their former routines – and August – when the summertime part-time work season may be in full swing. And this debate relies on data from a survey that is comparatively more useful in discerning broad trends over a period of months than it is in pinpointing estimates for particular months, like August versus January. In sum, there is a decent chance that we are way over-reading the few tea leaves left in the bottom of the cup.
But that criticism having been leveled, let’s look at the numbers much more broadly and consider what our situation might really be.
The number of hours of work truly is an important economic indicator. We know that because employers adjust worker hours to try to smooth out macroeconomic fluctuations. Employers who don’t want to lay off valued workers in a downturn will cut back hours instead – first by reducing overtime, and perhaps later with short furloughs – to avoid the possibly irreversible step of putting people on layoff (from which status they may try to find other jobs). In so doing, employers may consign some of their workers to part-time status, as measured by the household survey.
Later, when the economy improves, employers who fear making potentially costly commitments too soon may first hire workers on a part-time basis to limit their downside risk. (They may also increase overtime to postpone commitments to new hires.)
The historical data on part-time workers bear out this reasoning. A useful paper from two researchers at the Federal Reserve Bank of San Francisco, which corrects for some survey changes made by the Bureau of Labor Statistics back in the early 1990s, finds that part-time workers as a share of total employment increased from just over 18 percent in the late 1970s to more than 20 percent after the two recessions at the subsequent turn of the decade (see following chart). With some wiggles, it gradually fell to about 17 percent in the boom economy of the late 1990s. It was barely higher than that in the mid-2000s, but then it zoomed to almost 20 percent in the financial crisis – before the enactment of the ACA. So the health-reform bill cannot be blamed for that sharp increase in part-time work. As a footnote, most if not all of that increase in part-time work came for “economic reasons,” in the designation of the employment survey. But even at the worst of the post-crisis numbers, there were still about two voluntary part-time workers for every one involuntary part-time worker.
Source: Valletta and Bengali, “What’s Behind the Increase in Part-Time Work?”
The really bad news is that the part-time share of employment has remained high since the financial crisis, and the latest figures are barely below the post-crisis peak – in contrast to a relatively sharp decline from the elevated level after the early-1980s recession. Thus, it is well within the realm of possibility that even though the prospect of a health-insurance mandate on full-time workers is not what drove the prevalence of part-time employment up, it still is preventing it from going back down. That is a much more subtle phenomenon, and not subject to the kind of coarse analysis that we have seen highlighted in the popular press.
And it is possible that employers are responding now to the prospect of the application of the employer mandate in the future, even though the effective date has been postponed by a year. But that would make the interpretation of the data even more complex – in fact, it would suggest fairly strongly that it would be wise to wait for more data before drawing any firm conclusions.
So to be fair, it would be just as wrong to assert that the ACA has had no effect on employment patterns as it would be to claim that the ACA has shifted job creation to a ratio of 110 to 1 part-time to full-time. The best one can say is that the ACA has as yet had no discernible adverse effect on the mix of full-time and part-time jobs. And with the effective date of the employer mandate still 15 months away, such reticence would seem to be only common sense.
Still, questioning the possible ultimate employment effects of the ACA clearly is valid. And with all acknowledgment of pride of authorship, CED should recall the case we made against the employer insurance model. Employers clearly care about the health of their employees, out of genuine human concern as well as self-interest. But CED made a compelling case that the employer-insurance model is not the best safeguard for employee health – either physical or financial.
There are several reasons. One is that it is very hard institutionally for any individual employer to accommodate the diverse preferences of a number of employees. A second is that relatively smaller employers, with limited economies of scale, surely find it much harder to provide quality, affordable care. And this concern is not the same as saying that smaller employers cannot provide the same stability or security from business failure, or even the same amenities such as recreational opportunities, which we take as part of the territory and a question of employee choice and preference among prospective employers of different sizes. We as a society care enough about health care that in extremis we provide it at public expense in emergency rooms or through Medicaid, and we now subsidize coverage for relatively low-income workers. If we are willing to devote substantial public resources to health care, we cannot ignore that our system imposes systematic disadvantages on people who happen to work for smaller rather than larger firms.
But on today’s issue, it is at least unfortunate that the rules of the game now favor significantly an employer that is willing to cap an employee’s hours at 29.5 per week – to the potential detriment of that employee’s health, and the health of his or her spouse and dependents. Such an unintended consequence of an effort to promote health has to indicate that this entire legal structure is far from ideal. We can do much better, if we are willing to sever the artificial connection between employment and health care. That would free employers to concentrate on their own lines of business, to increase productivity and incomes and create jobs. And it also would eliminate the multiple moral hazards that soon might incent employers to pay comparative-token penalties in lieu of providing coverage, or to cut their employees’ hours solely for the purpose of off-loading the obligation to provide health care.
We cannot yet see a demonstrable statistical connection between our new health-insurance system and the number of part-time jobs. But we can see a reason why such a connection could grow over time. And we could avoid that risk, and achieve other substantial advantages including quality, efficiency and economy, with some bold changes to that system. The only things holding us back are inertia and short-sightedness. CED continues to believe that the nation should pursue health policy based on market competition to achieve quality, affordable care.