Gig work does not appear to be replacing ordinary jobs so much as filling the gaps between them. When Uber began operating in an American city, the local unemployment rate fell by between a fifth and a half of a percentage point, according to a study of its staggered spread across the country.
The paper, by economist Tucker Omberg of Jacksonville University, was published online in Labour Economics on May 9, 2024 (read the paper).
A natural experiment in 323 cities
In April 2013, Uber’s UberX service let drivers use their own cars for the first time. At that point residents of only 26 US cities could sign up. Within five years Uber operated in 310 of the country’s 384 metropolitan areas. That gradual rollout gave Omberg a way to compare cities where Uber had arrived with similar cities where it had not yet launched.
He combined monthly unemployment figures from the Bureau of Labor Statistics with Uber launch dates gathered from company press releases and local news, leaving 323 metro areas with reliable dates. He then ran the comparison with several modern difference-in-differences methods and a synthetic control approach, which all pointed the same way.
The basic estimate was a fall of about 0.21 percentage points in the unemployment rate, or 3.2%. Across methods the effect ranged from a fifth to a half of a point. Before Uber arrived, treated cities showed no downward drift relative to others, which supports reading the drop as a result of Uber’s entry rather than a cause of it.
The Austin test
Austin, Texas, offered a cleaner check. Uber operated there from June 2014 until May 2016, when it left over a city rule requiring drivers to submit to fingerprint background checks. It returned in June 2017 after a state law overrode the ordinance. Because the exit had nothing to do with the local job market, it works like an experiment. During the year without Uber, Austin’s unemployment rate ran 0.38 percentage points above trend.
A bridge between jobs
Omberg tested two explanations. The first is that Uber acts as an employer of last resort for people between jobs. Consistent with that, the share of Current Population Survey respondents working as taxi drivers and chauffeurs rose by about 25% after Uber arrived. More drivers also meant lower unemployment: each additional 1,000 driver-partners in a city went with a drop of 0.019 to 0.028 points.
The second is that rides help jobseekers reach work they could not get to before. Average commute times rose very modestly, by about 1.1%, and the effect was strongest in cities with tight labour markets, where Uber’s arrival came with nearly half a point less unemployment, while cities with the most slack saw no significant change. Cities with higher car ownership also saw bigger drops.
On pay, Uber’s arrival had a very small positive effect at the 25th percentile of wages and no clear effect at the median or mean. That, Omberg writes, casts doubt on the view that Uber hurts workers by destroying better-paying jobs.
A back-of-envelope calculation in the paper suggests that, if the whole drop came from people driving, roughly one in five Uber drivers would otherwise have been unemployed.
The author notes limits. Uber is measured simply as present or absent rather than by how many drivers are active, the results may not carry over to other gig platforms or even to Lyft, and an unseen factor could still have moved unemployment at the same moment Uber arrived. The Austin episode, where Uber left and came back for purely political reasons, is the hardest of those objections to explain away.
Study Details:
- Title: The labor market impacts of ridesharing on American Cities
- Authors: Tucker Omberg
- Journal: Labour Economics
- Publication Date: May 9, 2024
- DOI: 10.1016/j.labeco.2024.102560