Who benefits from Uber surge pricing?

In the last decade, new technologies have led to a boom in dynamic pricing. I analyze the most salient example, surge pricing in ride hailing. Using data from Uber in Houston, I build an empirical model of spatial equilibrium to measure the welfare effects of surge pricing. My model is composed of demand, supply, and a matching technology, and it allows for temporal and spatial heterogeneity, as well as randomness in supply and demand. I find that, relative to a counterfactual with uniform pricing, surge pricing increases total welfare by 3.66% of gross revenue. Only riders benefit: rider surplus increases by 6.52% of gross revenue, whereas driver surplus and Uber’s short-run profits decrease by 1.63% and 1.18% of gross revenue, respectively.

That is from a new job market paper by Juan Camilo Castillo of Stanford University.  At the link his other papers are interesting too.


Of course. The central assumption any market-oriented approach such as surge pricing is that need and demand are proportional to the ability to pay.

The voting booth is a market? What do they charge to vote in Germany?

Germans keep better records. Its a thing with them.


My experience in the UK is that one is turned away at the polls if not properly registered.

This is patronizing to the low income. Almost anyone with a credit card will be able to charge $60 to it. This will cause burdens for some. But the alternative in the bad old days was just not getting a car at all. If the alternative is missing a flight and paying a change fee, $60 suddenly becomes not just cheap but actually profitable, no matter how "poor" you may be.

If you think that lower income Americans can just throw around $60 on a credit card, it sounds like you have little experience with real lower-income Americans and instead are thinking of people who are grad-school poor. The real system that would benefit a lower-income person would be no surge pricing but a long queue. It's for this reason that in places where it's available, lower-income people take public transport where they can sacrifice speed and efficiency for low prices. The unfortunate side of this is how many lower-income communities have no public transport options.

Yes, I'm sure all 70,000 workers who earn at or less than the minimum wage in Virginia have to traverse that very corridor. Indeed, I am sure that none of them could choose an alternative arrangement, nor do they live in less expensive parts of the state.

Rt 66 is a disestablished federal highway that used to run from Chicago to Los Angeles. I-66 runs through Virginia.

The people spending $12 each way are funding the HOV lanes (a benefit to all) while taking pressure off the regular lanes (also benefit to all -- including drivers who cannot afford the tolls). And, of course, there is a zero-cost way for low-income workers to use the HOV lanes -- join (or start) a car-pool.

Of course wealthier people can afford to minimize inconveniences on many dimensions (including living close to work or even having the option to work from home and not commute at all).

Each and every Uber ride is subsidized by Uber so yes there is definitely a rider surplus, surge or not. Lyft and Didi also do the same just to remain competitive with Uber. Competition is wonderful.


That explains how my Didi rides are so cheap. My daily commute is 10RMB ($1.40) for 4km. The question I have though is: why does Didi, which has a virtual monopoly on the Chinese market, need to subsidize at all? Is the Chinese government involved? Is it using Didi as a para-public transport system?

You will win tremendous cash from friends or from legal sports betting arena
if you bet for the right horse. Any excellent tipster will even help
the client by picking out the height and width of the wager to stake which could be considered a small
section of his betting bank. The company which has been an integral part of live
dealer casino gaming in Asia more than a decade already provides
live dealer solutions for Bet-Victor, My-bet, and Stan James.

I'm mildly surprised that the benefit is so small. I'm also mildly surprised that there's no mention of the externalities - do third parties gain or lose by dynamic pricing? For example, does the Holy Environment gain?Perhaps the externalities of dynamic pricing are trivial compared to the externalities of having Uber at all?

This issue is of course completely distinct from the impact of surge pricing

So to clarify, should we then expect Uber or Lyft to adopt uniform pricing at some point? Given their current profit position (as Jorus points out), aren't they strictly better off under uniform pricing according to this paper?

No- note "short run profit"

So far, Uber has only long term losses, with no strategy to reach break even, and nothing created, no assets of any sort, created with any value since the first six months of losses.

Ie, if the cash ran out to fund the losses, what assets would be sold to liquidate all the liabilities accumulated by the years of losses in Federal bankruptcy court wealth redistribution? The job of the government technocrat will be easy, its easy to divide wealth when none exists.

A previous study (linked in this blog) found that surge pricing dd not materially increase supply (surplus) because drivers set a goal for the total revenues that wish to realize and then quit for the day (even if they forego additional surge pricing). Surge pricing gets them to the goal faster. Again, ride hailing is just one more example of the fallacy of the economists' infatuation with a simple concept (supply and demand) that isn't so simple.

I think you're right. The precis of the article has lots of long words but I don't believe it. If surge pricing doesn't materially increase supply then it's just more money for Uber and the drivers.

There are a lot of drivers who do that, but only go out during the surge periods. So it does increase supply.

Surge pricing gets them to the goal faster.

But then you have to factor the additional leisure into your calculation of welfare gains. There's a big difference between working eight hours for $100 and working four hours, taking four hours off, and earning the same $100.

If that doesn't come into the analysis then the analysis is very sloppy.

Ah, those pesky backward bending labor supply curves!

Surge pricing occurs in markets where regulated taxings have a rate cap. This means that regulated taxis will go to the places with the densest demand, forgoing some parts of an inner city and the suburbs.

During periods of Uber surge pricing, regulated taxis gravitate to areas where they can make many round trips quickly, or to airports where they can take a long trip to a downtown hotel, and from there take another trip to the airport.

I would be interested if the author looked at not just Uber as a competitor in the market, but other regulated taxis, and how they adjusted as well.

If you don't subscribe to ACM, here is a working paper version on geographic discrimination for Uber and Taskrabbit sharing models:


I used to run into this all the time in Cleveland. The bars would let out, and if your destination was far away, the drivers would make up an excuse about why they couldn't take you. It was illegal but of course not enforced.

We sometimes had to lie to get into the taxi and then tell the driver the real destination.

Makes sense. As a consumer, I am a little annoyed by surge pricing, but I am way more annoyed when something becomes unavailable due to being underpriced.

As a profit maximizing company why would Uber operate surge pricing if it causes short-run profits to decrease ? Does it somehow cause long run profits to increase ?

One would expect that as a company is better able to match supply and demand via flexible pricing it would be able to increase profitability if that was its aim.

The paper addresses this point as follows:

'I assume that, in order to maximize long run profit, Uber chooses
among pricing policies to maximize a weighted sum of short-run profit, rider surplus, and driver surplus. Uber cares about rider and driver surplus because its goal
is not to maximize short-run profit, but investors’ value in the long run. Higher
rider and driver surplus means satisfied consumers that are likely to return to the
platform, thus consolidating a client base that permits long-run growth.'

I would think, that if Uber believes this result they will adjust their algorithm.

It's not like there is one algorithm anyway. Such things evolve.

I think they follow the principle of "the best type of business is repeated business". If you catch a ride at high affluence time, it is much more likely that you'll use the service in the future. If they let you down one time, brand loyalty may disappear.

I don't think it's about increasing profitability but avoiding even larger loses.

Can you do both? There are a lot of variables in a "surge" algorithm. You can change rate of onset, height of adjustment, radius of concern, speed of return. Some combination of those might both make the customer feel "served" and also split returns with the company.

yes, Uber could benefit from higher revenue and riders may prefer paying more for shorter wait time so no reason why it could not be a win/win.

Though no guarantee that riders will gain. Some riders would have preferred to wait longer and get a lower price, and some will choose not to ride at all at the higher price. These loser may outnumber the winners.

Uber doesn't make a profit from rides, it makes a profit from convincing investors that it is a good investment. Surge pricing signals to investors that Uber is willing to innovate and is technologically advanced (it requires fancy algorithms and big data to set dynamic prices).

Well that's good news. Everyone benefits!

Depending on the value of your time, some people would benefit from a queue system over a surge pricing system. Not everyone has access to less costly alternative means of transportation like public transport.

The value of clockwork's time is clearly zero, so he can just wait until surge pricing ends.

But it wouldn't really be a queue system. For some people wanting a ride to or from the 'wrong' place, it would be a never-ever-get-a-ride system as it was in the pre-Uber days where there were places taxis just wouldn't go -- either because of perceived safety issues or because there were more profitable places to serve (profitable because there would be shorter waits between fares and no long, empty, non-paying return drives after a drop-off).

I always wondered if companies like Uber couldn't create a special 'auctioning for a ride during congestion' mode of the app where riders could directly bid on their rides.

The research fills in a Hayek triangle. And the Hayekens should not give up their Hayek triangle, it works.

I'm just impressed that complex human behaviours and responses to pricing are so well understood that economists can make predictions down to 1/100 of a percentage point in accuracy. All economists must be rich playing the stock market with precision like that.

A pure price analysis misses the point entirely. Surge pricing calls out drivers for times and events where they might not otherwise offer rides. To someone coming out of a ball game or concert in the rain or snow this is worth far more than the price markup.

How can this be anything but a test of one specific surge pricing algorithm, not a result about surge pricing in general? Any without doubting that Uber is in the business of increasing rider welfare, it is odd that the firm has chosen an algorithm that reduces its profits over uniform pricing (at what level?) that would be costless to implement.

https://www.rawstory.com/2019/08/uber-and-lyft-put-up-60-million-for-ballot-fight-to-avoid-paying-their-drivers-as-employees/ I wonder if Uber is similarly mistaken about it's self interest regarding classifying employees as "contractors."

Comments for this post are closed