Uber's explanation for why rides cost more than they used to is consistent across markets: insurance costs have risen sharply, and those costs are passed through to passengers. In California, the company tells regulators that insurance now accounts for 31% of the average rider fare. Globally, once insurance is stripped out, it says its true take is around 21% - unchanged for years.
A paper published at the ACM Conference on Fairness, Accountability and Transparency in June 2025 makes that story difficult to sustain. The research comes from Oxford University's own Department of Computer Science and was built on data from 258 UK Uber drivers covering 1.5 million trips between 2016 and 2024 - obtained through data subject access requests under Article 15 of the GDPR. The researchers had access to Uber's own backend data. That matters.
The study, led by Associate Professor Reuben Binns and DPhil student Jake Stein, found that since Uber introduced dynamic pricing in London in early 2023, its median take rate has risen from 25% to 29%, with some trips yielding a cut above 50%. Inflation-adjusted driver pay fell from over £22 per hour to just over £19 in the year following the change. Uber's surplus per driver working hour - the revenue generated for Uber while a driver is on trip, minus the driver's labour cost - rose 38%.
The core finding is structural. Uber's take rate, the Oxford paper found, is not evenly distributed across fare values. The higher the fare charged to the passenger, the higher Uber's take rate, and the less the driver earns per minute in absolute terms. Binns stated it plainly: "The higher the value of the trip, the more of a cut Uber takes. So the more the customer pays, the less the driver actually earns per minute."
That pattern is consistent with what Len Sherman, adjunct professor at Columbia Business School, found in a separate June 2026 analysis of 50,000 US trips from three veteran drivers. Sherman's regression on 100 near-identical routes found that Uber's reported "estimated commercial auto insurance and operational expenses" varied from $13.75 to $50.00 - a factor of more than 3.6 - on trips with identical risk profiles. Day of week, time of day and service type were statistically insignificant in explaining that variance. What did predict it: the rider price and the driver pay. Higher fares, higher reported insurance fees. Lower driver pay, also higher reported insurance fees.
The UK regulatory context makes this particularly pointed. Following the 2021 Supreme Court ruling that Uber drivers are workers - not independent contractors - Uber made changes to its payment model. The Oxford study found that commission charges disappeared from driver data around the same time, making take rate analysis harder. This is not peripheral: it was dynamic pricing, introduced in 2023, that restored the data fields researchers needed to calculate what Uber was actually keeping.
Consumer Watchdog reported in May 2026 that Uber's self-insurance reserves grew from $6.7 billion in its 2023 annual report to $12.5 billion in its 2025 annual report. Over the same period, approximately $4.1 billion was moved from those reserves to unrestricted cash on Uber's balance sheet. The company simultaneously pushed for legal changes in several US states that would limit its liability for accidents - changes that, if successful, would allow further releases from reserves.
As Insurance Business reported, Allianz Partners and Uber launched vehicle interruption cover for UK drivers in December 2024 - a product specifically designed to cover income loss during repairs. The launch was framed as a benefit to drivers. In the context of the Oxford research, it is also a data point about how Uber manages its insurance ecosystem: selectively providing coverage to drivers while controlling the commercial insurance line on the passenger side through its own captive structure.
The concern for UK underwriters is specific. Insurance reserves that grow faster than trip volumes, insurance fees that track margin rather than risk, and a business model that decouples passenger prices from driver pay - each of these creates challenges for anyone trying to price rideshare risk from the outside. The opacity is not incidental. As the Oxford researchers found, even with 1.5 million trips from Uber's own systems, constructing the take rate required inferring links between tables using timestamps because Uber does not provide a trip ID.
Uber's position is that insurance costs are genuinely high and rising. Its UK Supreme Court compliance changes, it has argued, increased costs. State-mandated and regulatory requirements vary by jurisdiction but are uniformly expensive. Its "true economic take," it says, once pass-throughs are deducted, is approximately 21% globally.
What that position does not address is the variance. If insurance costs explain rising take rates, those costs should vary with the risk profile of individual trips - not with the fare level. The Oxford and Columbia data, drawn from Uber's own systems in two different markets, both show the opposite pattern.
The gig economy insurance gap has been a live issue in the UK market for years. The Oxford research adds a specific dimension to it: the gap is not just about what coverage drivers have or lack, but about how the platform characterises its own insurance costs on individual transactions - and whether those characterisations reflect actuarial reality.
