Skip to main content

Economists Revisit 2015 Measure of Alternative Work Arrangements

Economists Lawrence Katz and Alan Krueger released a working paper that examines the measurement of trends in work arrangements over recent years, including a revisit of their 2015 replication of the Contingent Worker Supplement (CWS). The CWS, conducted by the Bureau of Labor Statistics (BLS), measures contingent and alternative work arrangements, and is often thought of as the government’s measure of the “gig economy.” After being conducted in 2005, the CWS lacked funding for more than a decade. Without rigorous data on the topic for such a long period of time, Katz and Krueger replicated the CWS in 2015 as closely as they could, and estimated 15.8 percent of the workforce was engaged in alternative arrangements for their main job, up from 10.8 percent in 2005. The BLS was able to administer the CWS in 2017, and in 2018 released their finding that 10.5 percent of workers were in alternative arrangements—not the large jump estimated in 2015.

In this week’s paper, Katz and Krueger examine this discrepancy. They attribute it to three factors. First, the 2015 survey relied on a less representative sample than the BLS CWS, and oversampled multiple job holders. Second, a weaker labor market in 2015 meant that more people were likely driven to alternative arrangements due to a lack of traditional jobs. Third, the 2015 survey asked only about respondents’ work, while the BLS CWS asks respondents to report as a proxy for their entire household, and those answering as a proxy are less likely to report alternative work. Adjusting for these factors, the authors conclude that there had been a more modest increase in alternative arrangements—around 1 or 2 percentage points—than the 2015 survey initially reported.

Taking these revised findings in conjunction with other surveys and tax data, the authors conclude that recent growth in independent and alternative work is primarily among those relying on it for supplementary income. By considering multiple types of data, they come to a deeper understanding of trends than any one source can offer. In addition, the paper illustrates how revisiting studies as we learn more on a dynamic and complex topic promises to improve both our knowledge of the workforce and researchers’ methods moving forward.

Read next