Hey, Uber, release your data

During the horrific hostage crisis in Sydney yesterday, ride-sharing company Uber got a lot of criticism for enabling “surge pricing”. In response, the company replaced surge pricing with free rides out of the CBD. Were either of these actions helpful?

The Uber algorithm automatically introduces higher prices when demand is high in order to engage the market mechanism – ration demand to those most willing to pay, and increase supply by encouraging more drivers onto the road. If the supply response is sufficient, prices should fall to normal before too much time passes.

In normal times, surge pricing makes a lot of sense. It should result in a better experience for everyone when demand is high for some innocuous reason, like a big football game. With higher prices and hence more drivers available, passengers desperate to get to the game can get a ride, and drivers can enjoy some benefit from the higher value they’re adding by being available right then rather than at some other time.

Emergency situations are different. There are elements to these events that mean it may be optimal for altruism to drive individuals’ decisions, rather than profit. Unfortunately, those two decision drivers can’t work together – they are necessarily in opposition to each other. Famous studies have shown that introducing money to a decision that would otherwise be made altruistically tends to “switch off” people’s altruism. If the optimal decision is the altruistic one, offering financial incentives moves the decision further away from, not closer to, that optimum.

So which situation prevailed in Sydney yesterday?

First, let’s look at profit. In an emergency, people really need to get home (or at least away from the scene of the emergency). They’ll be willing to pay whatever they can afford for a ride. But higher prices aren’t likely to ration demand as much as usual – everyone in the CBD wants to get out right now, they’re not going to have a coffee and catch a ride later when the prices are down. Some could switch to public transport or carpool, but more than usual will stick with the expensive Uber ride.

That means a bigger supply response than usual is needed. But here’s another way that emergencies are different: the cost to the driver is higher. Normally, surge pricing tells a driver to stop watching TV at home and get into the CBD. Yesterday, they were being asked to trade their home and couch for a CBD where they could be shot. If profit is their motive, they’re going to need a much bigger financial incentive to get behind the wheel. There may not even be any incentive big enough to overcome the danger they see. This would explain why yesterday’s Uber pricing was reported as higher than any seen in Sydney before – the price signal had to offset a much larger cost than ever before.

This is where altruism comes in. Faced with an unknown risk of death, a driver who chooses to go into the city and offer rides is probably at least partly motivated by altruism. It’s desirable that people should help one another in emergency situations: in Wellington we’ve seen it often, after earthquakes or when the trains are off, bus drivers not collecting fares, strangers carpooling together. Could Uber’s surge pricing be switching off this altruistic instinct in people?

Fortunately, it’s possible to test which type of decision this is for drivers, using Uber’s own data. The conclusion from this test comes with a good solid policy recommendation.

If the number of drivers working the CBD was higher before surge pricing kicked in than after, and did not decline significantly after surge pricing was cancelled, that means offering rides in an emergency is an altruistic decision. The chance of financial benefit has encouraged drivers to think of their own self-interest to the point where they would rather not enter the CBD at any cost. To keep drivers on the road, it would be preferable in emergencies to keep prices at normal levels or reduce them to zero, depending how sensitive the drivers’ decisions are to different pricing levels.

If, on the other hand, there was no fall in driver numbers when surge pricing kicked in, but there was a decline after Uber cancelled the surge pricing, that means offering rides in an emergency is a profit-driven decision. Drivers won’t get on the road without a big enough incentive, and surge pricing really is the best way to ensure everyone gets home safely.

Wired described yesterday’s surge pricing as a public relations disaster for Uber. If Uber could prove that surge pricing helped by getting more drivers on the road, that would help mitigate the disaster. If that’s not what the data shows – if it looks like Uber drivers behaved altruistically – that information could be used to design a better pricing algorithm to maximise rides in emergency situations (and prove that the company really does care).

Either way, Uber, release your data!