Back in 2019, in a popular article, the Economist had predicted that Uber will never be profitable. The article “Can Uber ever make money?” outlined the main challenges: “The trouble is, as competition increases, ride-hailing becomes a commodity business. Customers care little whether they ride with Uber or Lyft, as long as it gets them from a to b. That means neither firm can easily increase profits by raising fares, but may instead have to offer discounts. Likewise, the ride-hailing firms do not own their cars: their drivers do, and so have no reason to be loyal. That forces the firms to pile on incentives to stop drivers from deserting, kicking profits even further down the road.”
This is fantastic. Could you go into more detail about why a tighter labor market hasn’t overwhelmed Uber’s ability to increase its take rate? If I understand your analysis correctly, it appears that Uber has increased take rate by effectively gamifying the driver ride selection experience. This feels similar to your piece on variable pricing. Drivers value ride certainty and Uber can exploit this by offering lower pay to drivers who elect to always have the next ride booked.
It’s just surprising to me that they’ve been able to outpace competition from other industries for drivers as unemployment decreased. I had expected a tight later market to be a massive challenge for ride-sharing companies. Obviously this interacts with the supply of Uber drivers in the market. Has Uber also become better at targeting supply to use cases where it can be profitable?
If Uber has used its technology to improve how it optimizes the mind-bogglingly complex problem of balancing demand, supply, driver pay structure, and and market pricing, that would auger well for the “Silicon Valley-ization” of more old school industries over time. Do you agree or disagree? What would differentiate Uber from something like Flexport, which has struggled? Is there simply a constrain on the availability of computer science talent and a minimum scale needed to achieve these kinds of results?
My personal experience, which may or may not extrapolate, is that I select Uber/Lyft versus Taxi based on use case. I live in the Chicago suburbs. For airport trips, the suburban taxi service is both more reliable and less expensive, so I’ve switched. But for ad hoc rides around town, into/from Chicago, etc., I don’t even think of taking a taxi and start with a ride-share app.
Gad, I really enjoyed your insightful article on Uber's improving economics.
Globally, Uber reported a 30% YOY increase in drivers on their platform in 2023, and a 15% increase in engagement (ie, online hours per driver). That’s a 43% increase in supply, far outstripping (~3X) their growth in trip demand. The inevitable consequence of this driver oversupply has been: 1) enabling Uber to substantially cut driver pay per trip (11.9% in the US, as I reported in December); and 2) providing fewer trips per driver (ie lowering utilization rates). The net result has been a uniquely severe cut in Uber's US average monthly driver earnings YOY in 2023, >15% for both Mobility and Delivery and delivery drivers (as reported by Gridwise). No other gig company -- not Lyft, not DoorDash, not Grubhub -- has come close to cutting US driver earnings by double digits.
This raises the obvious question, why aren't normal labor economics working in the US gig worker market...ie why isn't Uber experiencing a drop in drivers willing to sign up for a company that is cutting pay far more than competitors? A couple of factors may explain the current phenomena. First, is the precarity of financial circumstances facing lower-income citizens, particularly the growing number of recent immigrants. In most major cities, immigrants make up the vast majority of drivers on Uber's platform, and many of the economy's most vulnerable workers may have few realistic alternatives to supplement/earn income than jumping on Uber's relatively frictionless platform. Second is Uber's market power. It's reasonable to assume that most prospective drivers would be most likely to sign on with the dominant rideshare market leader to maximize perceived income opportunities. And, as I've noted in my reporting, Uber likely has the most sophisticated, AI algorithms in the industry to minimize their labor costs by targeting lowball trip pay at the least sophisticated, newest, perhaps most income-stressed recruits to their platform. The exploitable information asymmetry between Uber and its drivers may be unprecedented in business history.
I'm looking forward to your further analysis of Uber's economics, based on a deep dive of NYC data, subject to the caveat that, as you know, NYC represents a unique market situation, with capped driver supply, far more full-time drivers, higher driver eligibility requirements, etc.
In the meantime, let the record show that despite Uber's protestations on my findings on their take rates, the company has chosen not to disclose any meaningful metrics on its US ridehail operations: e.g. recent trends in average driver pay per trip, mile or hour, utilization rates, average earnings per online hour, etc. etc. I would also love to see a distribution of their US take rates. My hypothesis is that a significant proportion of their improving US profitability has been driven by the right hand tail of such a distribution, ie trips with a 50+% take rate. This rich source of profitability would take a big hit if minimum driver pay rate legislation (as currently scheduled to start in Minneapolis on May 1st) starts gaining momentum in other cities. This may explain why Uber (and Lyft) are threatening to invoke the nuclear option in abandoning their operations in Minnesota.
As always, great analysis! However, wouldn't lower pay for drivers cause drivers to start to think about switching and wouldn't the market finally converge back to mean of some sort? What would be a price point at which riders would start to think of different alternatives and thus either Lyft OR some other alternative would start to gain share, which might lead to Uber having to lower the prices, leading to bad cars, leading to similar fate as taxis?
This is fantastic. Could you go into more detail about why a tighter labor market hasn’t overwhelmed Uber’s ability to increase its take rate? If I understand your analysis correctly, it appears that Uber has increased take rate by effectively gamifying the driver ride selection experience. This feels similar to your piece on variable pricing. Drivers value ride certainty and Uber can exploit this by offering lower pay to drivers who elect to always have the next ride booked.
It’s just surprising to me that they’ve been able to outpace competition from other industries for drivers as unemployment decreased. I had expected a tight later market to be a massive challenge for ride-sharing companies. Obviously this interacts with the supply of Uber drivers in the market. Has Uber also become better at targeting supply to use cases where it can be profitable?
If Uber has used its technology to improve how it optimizes the mind-bogglingly complex problem of balancing demand, supply, driver pay structure, and and market pricing, that would auger well for the “Silicon Valley-ization” of more old school industries over time. Do you agree or disagree? What would differentiate Uber from something like Flexport, which has struggled? Is there simply a constrain on the availability of computer science talent and a minimum scale needed to achieve these kinds of results?
My personal experience, which may or may not extrapolate, is that I select Uber/Lyft versus Taxi based on use case. I live in the Chicago suburbs. For airport trips, the suburban taxi service is both more reliable and less expensive, so I’ve switched. But for ad hoc rides around town, into/from Chicago, etc., I don’t even think of taking a taxi and start with a ride-share app.
Gad, I really enjoyed your insightful article on Uber's improving economics.
Globally, Uber reported a 30% YOY increase in drivers on their platform in 2023, and a 15% increase in engagement (ie, online hours per driver). That’s a 43% increase in supply, far outstripping (~3X) their growth in trip demand. The inevitable consequence of this driver oversupply has been: 1) enabling Uber to substantially cut driver pay per trip (11.9% in the US, as I reported in December); and 2) providing fewer trips per driver (ie lowering utilization rates). The net result has been a uniquely severe cut in Uber's US average monthly driver earnings YOY in 2023, >15% for both Mobility and Delivery and delivery drivers (as reported by Gridwise). No other gig company -- not Lyft, not DoorDash, not Grubhub -- has come close to cutting US driver earnings by double digits.
This raises the obvious question, why aren't normal labor economics working in the US gig worker market...ie why isn't Uber experiencing a drop in drivers willing to sign up for a company that is cutting pay far more than competitors? A couple of factors may explain the current phenomena. First, is the precarity of financial circumstances facing lower-income citizens, particularly the growing number of recent immigrants. In most major cities, immigrants make up the vast majority of drivers on Uber's platform, and many of the economy's most vulnerable workers may have few realistic alternatives to supplement/earn income than jumping on Uber's relatively frictionless platform. Second is Uber's market power. It's reasonable to assume that most prospective drivers would be most likely to sign on with the dominant rideshare market leader to maximize perceived income opportunities. And, as I've noted in my reporting, Uber likely has the most sophisticated, AI algorithms in the industry to minimize their labor costs by targeting lowball trip pay at the least sophisticated, newest, perhaps most income-stressed recruits to their platform. The exploitable information asymmetry between Uber and its drivers may be unprecedented in business history.
I'm looking forward to your further analysis of Uber's economics, based on a deep dive of NYC data, subject to the caveat that, as you know, NYC represents a unique market situation, with capped driver supply, far more full-time drivers, higher driver eligibility requirements, etc.
In the meantime, let the record show that despite Uber's protestations on my findings on their take rates, the company has chosen not to disclose any meaningful metrics on its US ridehail operations: e.g. recent trends in average driver pay per trip, mile or hour, utilization rates, average earnings per online hour, etc. etc. I would also love to see a distribution of their US take rates. My hypothesis is that a significant proportion of their improving US profitability has been driven by the right hand tail of such a distribution, ie trips with a 50+% take rate. This rich source of profitability would take a big hit if minimum driver pay rate legislation (as currently scheduled to start in Minneapolis on May 1st) starts gaining momentum in other cities. This may explain why Uber (and Lyft) are threatening to invoke the nuclear option in abandoning their operations in Minnesota.
We live in interesting times!
As always, great analysis! However, wouldn't lower pay for drivers cause drivers to start to think about switching and wouldn't the market finally converge back to mean of some sort? What would be a price point at which riders would start to think of different alternatives and thus either Lyft OR some other alternative would start to gain share, which might lead to Uber having to lower the prices, leading to bad cars, leading to similar fate as taxis?