The lead academic behind an MIT CEEPR research paper which casts doubt on the sustainability of driving for Uber or Lyft has accepted Uber’s criticism of how driver income data was gathered and said he will revise the research using more generous income calculations.
“Uber’s Chief Economist Jonathan Hall wrote a thoughtful response expressing his concern with one aspect of our paper,” wrote lead author Stephen Zoepf in a response statement yesterday to the company’s critique of the research. “Hall’s specific criticism is valid; in retrospect the survey questions could and should have been worded more clearly.”
Last week Uber CEO Dara Khosrowshahi appeared to be channeling the spirit of former chief exec and Uber founder, Travis Kalanick, when he tweeted out a clumsy insult — writing that “MIT = Mathematically Incompetent Theories” — in his initial response to the research.
MIT = Mathematically Incompetent Theories (at least as it pertains to ride-sharing). @techreview report differs markedly from other academic studies and @TheRideshareGuy recent survey. Our analysis: https://t.co/S2aAqCuDR0
— dara khosrowshahi (@dkhos) March 3, 2018
It was an unfortunate shift away from an otherwise contrite and even humble tone the new Uber CEO has sought to strike since taking the driver’s seat last year (and inheriting a trunk stuffed full of legacy baggage and operational skeletons — from security scandals and legal challenges to culture problems, federal investigations and regulatory sanction, to name a few of Uber’s problems).
Perhaps Khosrowshahi realized the error of letting the mask slip as his angry tweet was quickly followed by a tweet of thanks to MIT for “listening and revisiting”.
The MIT research paper compared the earnings and costs of 1,100+ drivers on the two ride-sharing platforms, with the original version of the study concluding that the median profit was just $3.37 per hour — with 74% of the drivers earning less than the minimum wage in the state where they operate.
The researchers also found that a median driver generates $0.59 per mile of driving but incurs costs of $0.30 per mile; and almost a third (30 per cent) of drivers were found to incur expenses exceeding their revenue or to be losing money for every mile they drive.
It’s the driver earnings figures that Uber took issue with, rather than the costs. Lyft also told us its initial impression of the study was it “shows some questionable assumptions”.
Harry Campbell, founder of The Ride Share Guy blog, which hosted the driver survey where MIT gathered data for its study, also queried MIT’s driver earnings calculations. Though his business is full time blogging about the ride-sharing industry which inevitably connects him to its interests — if not literally as directly as Uber and Lyft.
“The expense side of things seems right on to me but $7/hr in gross earnings seems too low based off my experience,” Campbell told us of MIT’s original study. “I think there was some ambiguity around the question about driver’s monthly income where as I directly asked drivers this year how much they made per hour and the average was $16.93 per hour before expenses.”
While Zoepf has said a “thorough” revision of the research will take “a few weeks” he has published an initial assessment of the net hourly ride-hailing driver profit distribution using two new methods — both of which bump up driver median profits (to $8.55 per hour; and $10p/h, respectively). Though only one puts a majority of drivers over the 2016 minimum wage in their state.
Here’s how he describes the two methods:
Using data in this survey there are two ways to calculate hourly revenue and net profit numbers. Method 1: use monthly revenue numbers when available (Question 14) and calculate hourly numbers using the working schedule reported in Question 11. Method 2: Use reported $/hour revenue numbers when available (Questions 19 & 22). In both cases the profit from driving is higher than we initially reported.
Using Method 1, and following Hall’s advice not to adjust income, Median profit rises to $8.55/hour from the $3.37 initially reported. For 54% of drivers, profit per hour is less than 2016 minimum wage in their state. 8% of drivers lose money. Using Method 2, median profit rises to $10/hour. For 41% of drivers, profit per hour is less than 2016 minimum wage in their state. 4% of drivers lose money. I’m happy to review these numbers with Hall and his team to address any remaining questions with the analysis as I release the next draft of the paper.
Zoepf goes on to call for Uber to provide better access to data to enable robust and independent analysis of the economics of the ride-sharing model, pointing out there is “no public ride-hailing data and a paucity of independent studies outside Uber’s own analyses”.
“In the spirit of collaboration, I ask the following from Uber, in keeping with the original objectives of this paper,” he continues. “(1) Help make an open, honest and public assessment of the range of ride-hailing driver profit after the cost of acquiring, operating and maintaining a vehicle. (2) Transparently present the difference between actual and tax-reportable vehicle expenses used in the business.
“In support of these goals I am happy to share existing cost data from this working paper with Uber or Lyft, or to incorporate full and accurate revenue data from Uber in this analysis should they decide to share such data.”
You can read his full response statement here.
One telling facet of the US ride-sharing market is that for most drivers Uber or Lyft is not their full time job. According to Campbell’s data, “about 65 per cent of ride-share drivers are part-timers — “so definitely a majority of drivers are doing this on the side or have other jobs”, he says.
That undermines the simplistic notion that drivers would simply not be driving for ride-sharing platforms if it was not economical for them to do so.
The reality is that most ride-sharing drivers are juggling multiple jobs — and are therefore having to make their own complex calculations to try to determine their individual costs and earnings. And having to do so without the benefit of public data to help them.
Add to that, some drivers may also be facing a bill payment falling due, such as a lease on a car, which means they feel they have to drive to earn income in the short term to cover an imminent cost.
For drivers who own the vehicle they are using for ride-sharing, over the longer term vehicle depreciation is also be being racked up in the background — saddling them with the burden of negative equity on their cars. A cost that Uber and Lyft’s businesses offload onto drivers — hence the criticism that gig economy business models are inherently exploitative.
Safe to say, the complex considerations and calculations that surround ride-hailing mean external and robust investigation of its economics is sorely needed. As is access to high quality data to enable vigorous research.
What’s really not called for are insults to academic efforts that seek, in good faith, to quantify the pros and cons of ride-sharing.
Comment