Would Uber Be Safer With Different Prices?

Put yourself in the shoes of an Uber or Lyft driver, and think about the costs of being a driver. There are two main ones: your time, and your car maintenance costs. It would make sense to pay you on that basis — to pay you the cost per mile driven, say $0.40 for the typical UberX/Lyft car, plus an hourly wage high enough to get enough decent drivers to meet passenger demand, say $12 an hour.

Yet the actual price structure is very different. The rate per mile for LA UberX is $0.90 per mile, while the hourly rate while a passenger is in the vehicle, after Uber’s commission, is $7.20. With a high mileage rate and a low time rate, driver income is strongly tied to how fast they’re driving on trips. At a 15 mph average trip speed, they’re earning $12.00 per hour of trip time (after an assumed $0.40 per mile vehicle operations cost), at 30 mph it’s $16.80, at 60 mph it’s $26.40. (UberX has no base fare, but a minimum fare of $4.65).

Below, I’ve plotted what driver pay looks like per hour of trip time, based on the speed of travel. I assumed that the average speed of trips is 20 MPH, so I chose a rate structure that has the same pay as the current one, at 20 MPH - $0.27 a minute, $0.55 a mile.

Odds of a person on foot dying in a car crash, based on age and vehicle speed

And all those figures are before you account for down time, where the driver is active on the platform but does not have a passenger in the car - I couldn’t find good data on the share of downtime - which includes both time between trips, and time driving to pick up the passenger before the price meter has started.

Shifting away from this high mile cost, low minute cost rate structure, to the opposite - a lower mile cost, higher minute cost rate structure, would have a number of benefits:

  • More Stability, Less Stress: Driver incomes are highly sensitive to trip speed and vehicle congestion, making driving more stressful and incomes unstable. Not desirable for someone piloting a 3000 pound vehicle around packed city streets, with precarious finances. 
    With time a bigger factor, drivers have more control over how much money they make — as long as they’re getting passengers at least. If you upped the per minute rate and reduced the per mile rate, driver income would be more stable, and depend less on how fast they drive. Reductions in stress and income instability might reduce driver turnover, reducing the need for constant costly recruitment of new drivers.
  • Safer Speeds: With less of a penalty for driving at slower speeds, drivers have less reason to speed. Even small reductions in vehicle speeds can greatly reduce the likelihood of car crashes being fatal.
  • Insurance Savings: If paying drivers more at lower speeds reduces the frequency or severity of crashes, Uber drivers would make fewer insurance claims, letting Uber negotiate lower insurance premiums (or pay out less if they self-insure).
  • Better Service: Drivers are currently hostile to customer demands that entail stopping mid-trip - understandably so, because driver pay when the car isn’t moving is minimal. With drivers paid more for their time, they’ll be more accommodating of customer demands, as they don’t suffer much of a loss in pay if the customer wants to stop somewhere mid-trip, like at a drive through. This means better service.
  • Better handicap access: Under the current system, Uber charges the same rates for standard UberX and the ‘Access’ service for the disabled. Disabled passengers often take longer to enter and exit the vehicle, and so drivers are forced to sacrifice considerable income to service these passengers, because the per minute rate is low while the per mile rate is high. Having a higher minute rate would increase driver pay for this service. It would also admittedly increase prices for the disabled - whether their rides should be subsidized by other passengers, taxpayers or not at all, is a matter of debate, but it’s neither fair nor effective to expect drivers to take the hit, which is how Uber works right now.
  • Better Geographic Coverage: Driver coverage might improve, because driver income from different passenger types would be more uniform — drivers won’t have as much incentive to lurk in areas where requests for long distance trips predominate.
  • Bike and Bus Friendly: Higher speed trips are currently subsidizing lower speed trips, i.e. making lower speed trips cheaper — and lower speed trips tend to be shorter ones, because you’re not using the highway to make them. 
    Lower speed trips are more easily replaced by walking, biking, carpooling (which Uber and Lyft offer) and transit. Conversely, longer trips would become cheaper, and these trips are harder for people to replace with transit et al. A greater share of trips being longer means less down time for drivers, which enables some combination of higher income for drivers, lower prices for customers, or higher profits for Uber/Lyft.
  • Proto-Congestion Pricing: trips are under-priced precisely when its least desirable — during times of peak congestion where traffic moves slowly. Paying more by the minute and less by the mile would be a proto-congestion surcharge, and could have a perceptible impact in areas with high concentrations of cabs when congestion is high.
  • Honest Prices: riders currently get overcharged when the in-app navigation picks a slightly faster but much further route (say 1 minute faster but 5 miles farther and thus $4.82 more — Google Maps and Waze give me routes like this all the time). This is because the navigation app is designed to minimize time without regard for mileage, but the actual price to the passenger is based mostly on mileage. I consider this quasi-fraud, but it only happens to the extent that the shortest distance route and shortest duration route differ.

There could be some brilliant reason behind the current rate structure that I’m missing. But it seems like a relic of the taxi era, where drivers could defraud you by driving slowly on trips, so they made the per minute rate low and the per mile rate high. With Uber and Lyft, and their passenger ratings and trip route information, the risk of such driver fraud seems much smaller, and easier to catch. The main reason I can see is that people taking slower (typically shorter or peak congestion) trips are so valuable as customers that they’re worth subsidizing with fares from other passengers — but that doesn’t seem very plausible.

A modest per mile premium over car costs might make sense, to the tune of $0.10–15 per mile (compared to the current $0.60 premium), to encourage drivers to complete trips promptly. But beyond that, I don’t see why the per mile rate should be so high and the per minute rate so low. For all the hoopla around Uber and Lyft, surge pricing aside, their rate structures are still rather pedestrian.

You can change the rate structure so that the average income to drivers doesn’t change. For instance, if you assume the average trip speed for UberX is currently 20 mph, rates of $0.33 a minute, $0.55 a mile for UberX in Los Angeles would yield the same income average, with the benefits listed above.

I asked a prominent driver-blogger what he thought, and he was supportive.

NB: I did find this interesting article: http://fusion.net/story/140655/the-way-uber-fares-are-calculated-is-a-mess/ — seems like the relationship between per mile and per minute rate is all over the map, literally, across different cities and countries. Moscow UberX has a rate structure like what I propose, $0.17 a mile, $0.11 a minute. Would be interesting to see how it influences rides. Uber could do randomized trials in local markets to see what the impact would be.

Update: Uber recently updated its pricing in many cities. Detroit now has a rate structure like what I propose, $0.30 a minute, $0.30 a mile — though that mileage rate may be lower than the average cost to drivers on a per mile basis. Previously, it was $0.15 a minute, $0.75 a mile. Either Uber thinks Detroit is uniquely suited for such a rate structure, or Uber is testing this rate structure for a wider rollout if it goes well — there’s no telling which.