How many gig workers are there? Depends on how you ask

Companies like Uber have enabled the gig economy. (Source: Bloomberg/Getty Images)

Our understanding of the US labour market is based to a remarkable extent on a few questions asked each month as part of the Current Population Survey, which is conducted by the Census Bureau in cooperation with the Bureau of Labour Statistics. The most important goes something like this (the wording changes depending who is asked and when): LAST WEEK, did you do ANY work for either pay or profit?

If the answer is no, there’s a series of questions to suss out whether the respondent is looking for work or is not working because she’s retired, in school, disabled, or for some other reason. If it’s yes, there are follow-up questions about hours worked, self-employment, multiple jobs and so on. This is how the BLS determines the unemployment rate as well as other key indicators such as the employment-population ratio and the labour force participation rate.

Sometimes the Current Population Survey includes additional questions, such as this one:
Last week, were you working as an independent contractor, an independent consultant, or a free-lance worker? [That is, someone who obtains customers on their own to provide a product or service.]

That’s from the May 2017 Contingent and Alternative Employment Arrangements supplement, the results of which received a lot of attention from me and others. They showed the percentage of US workers who are independent contractors or freelancers to have fallen to 6.9 percent, from 7.4 percent the previous time the questions were asked in 2005. The share in “alternative work arrangements” — contractors/freelancers plus temporary help agency employees and on-call workers — had dropped to 10.1 percent from 10.7 percent.

These declines were the opposite of what most had expected, given the much-heralded rise over that same period of the “gig economy” enabled by online platforms such as Uber Inc., Taskrabbit (now owned by Ikea) and Upwork Global Inc. They also seemed to contradict the findings of a survey modeled on the contingent work supplement and conducted by two prominent labour economists in 2015, which found that 15.8 percent of US workers were in alternative work arrangements and that such work had accounted for 94 percent of net job growth since 2005.

It could be, though, that the Census Bureau survey takers were just asking the wrong question. “There are serious questions about what’s being measured” by the independent contractor question, said Katherine Abraham, an economics professor at the University of Maryland and former head of the BLS. In a paper published in February (an earlier, paywall-free version is available here), Abraham and Brad Hershbein, Susan Houseman and Beth Truesdale of the W.E. Upjohn Institute for Employment Research in Michigan show that different word choices can double the share of workers classified as independent contractors or freelancers to about 15 percent.

In one form of the authors’ survey, administered by Gallup Inc. as part of its daily Education Consumer Pulse, respondents who said they worked for an employer were asked a follow-up question to clarify whether they were “an employee on the job or work as an independent contractor, independent consultant, or freelance worker.” In another they were asked whether their employer took taxes out of their pay. Both resulted in similar increases in the percentage classified as independent contractors.

The paper is one of a flurry of recent studies finding that the CPS is probably undercounting not just freelance and independent contracting work but work overall. A sample:

*Anat Bracha of the Hebrew University Business School in Jerusalem and Mary Burke of the Federal Reserve Bank of Boston concluded from a survey of informal work participation conducted annually from 2015 through 2022 that the actual US employment-population ratio could be as much as 5.1 percentage points higher than the ratio derived from the CPS, and aggregate hours worked 1.4 percent higher in 2019 and 1.1 percent higher in 2021.

*After comparing 2012-2016 data from the CPS and the American Time Use Survey, another Census/BLS collaboration, Mary Dorinda Allard and Anne Polivka of the BLS concluded that between 0.4 percent and 3 percent of the adults counted by the CPS as not employed actually engaged in some “income-generating hobbies, crafts, food, performances, or services.”

*Abraham and Ashley Amaya, then of RTI International and now of the Pew Research Center, found that asking additional questions of respondents, especially about the employment situation of others in their household, resulted in higher rates of employment and multiple job holding.

*Lawrence Katz of Harvard University and the late Alan Krueger of Princeton University, who conducted the aforementioned 2015 survey finding 15.8 percent of US workers to be in contingent or alternative work arrangements, concluded that one of the reasons it delivered such different results from the CPS was that, while it asked the same questions, it asked only about respondents’ own situations, not those of other household members as is done in the CPS.

This research doesn’t tell us whether these undercounts have worsened over time. The problem certainly isn’t new: Asking follow-up questions and providing hypothetical vignettes resulted in higher employment rates in a 1995 Census/BLS study as well. But the widespread sense that work arrangements in the US have become more tenuous and complicated than they used to be, coupled with evidence from Internal Revenue Service data that the share of taxpayers reporting self-employment income has risen, does seem to point to at least some increase. I’m a longtime skeptic of claims that the US labor market is being taken over by “free agents” or gig workers and still think such claims are often exaggerated. But something is going on out there. “Our measurement system is designed for a world where most people have one job, and it’s a job with an employer,” Abraham told me. Nowadays, she added, “I am surprised at the number of people who have a side thing that they do.”

Can the BLS and Census Bureau do a better job of measuring this? Changing the core employment questions in the CPS is generally a no-no because it risks breaking the continuity of data series that are available on a monthly basis back to 1948. Supplements such as the one on Contingent and Alternative Employment Arrangements are less sacrosanct. It was inaugurated in 1995, conducted every two years through 2001, and since then (because of funding constraints) only in 2005, 2017 and 2023.

The questions in last year’s survey, the results of which should be out in a few months, were mostly unchanged from 2017 — although the parenthetical about an independent contractor or freelancer being “someone who obtains customers on their own” was dropped, which may result in more people who obtain customers from online platforms answering yes. By the next time around, it appears that there will be quite a lot of research pointing to the need for bigger question changes than that.

Justin Fox is a Bloomberg Opinion columnist covering business, economics and other topics involving charts. Views are personal, and do not represent the stand of this publication.

Credit: Bloomberg