The Effects of Uber's Surge Pricing: A Case Study [pdf]

http://faculty.chicagobooth.edu/chris.nosko/research/effects_of_uber%27s_surge_pricing.pdf

by minimaxir • 11 years ago

50 47 11 years ago

faculty.chicagobooth.edu

refurb• 11 years ago

Nice analysis. It's interesting seeing the data (requests complete, etc) over the surge periods.

I also like the comment about how periods where demand is high (NYE) is also where the drivers have most incentive to not drive. I hadn't considered that aspect.

minimaxir• 11 years ago

Interesting/practical use of ggplot2 for case studies, too.

jmakeig4• 11 years ago

Great read and thorough approach. However, it's worth noting that Uber provided the data for this study. I'd imagine they had an idea as to its conclusions before they provided it.

The findings are intuitive and validate that surge pricing is an overall good thing for the market.(Yea, microeconomics!) However, the cynic in me isn't sure we'd have seen a study that didn't come to that (positive) conclusion.

How does an agreement between an academic and a data provider work to prevent this sort of bias? (In the general case. I have no evidence of that in play here.)

Update: speling

denzil_correa• 11 years ago

I am too concerned about the PR and the fact that this study was commissioned by Uber. I would view the conclusions of this study with appropriate skepticism.

http://newsroom.uber.com/2015/09/the-effects-of-ubers-surge-...

hoopd5• 11 years ago

The findings are intuitive and validate that surge pricing is an overall good thing for the market.

Good compared to what? Good compared to having a giant monolith set prices for half the market, or good compared to a real market where drivers are allowed to set their own prices?

Surge pricing is Uber taking a small step away from their static central command pricing structure and moving an inch towards market dynamism. The real question is if they believe so strongly in markets why are they setting prices in the first place?

The current regulatory structure (in theory) sets prices on behalf of the public, 1000-to-1 odds says Uber sets prices on behalf of their shareholders.

jmakeig4• 11 years ago

Maybe "efficient" would have been a better word than "good". I have no doubt that Uber is pricing to benefit their shareholders, not the public good. That's why the taxi fare regulations are there in the first place.

y0y• 11 years ago

If individual drivers set pricing, would I use the app? Not if there was a simpler alternative which, on average, led to the same overall pricing. Imagine the user experience if every driver was setting their own prices. Ugh.

steven2012• 11 years ago

Your theory fails because regular Uber prices are almost half the price of a regular cab, which according to you is set on behalf of the public.

The benefit of Uber setting the price is that it makes the marketplace more reliable. And it's set at a level where both drivers and riders are happy. If Uber moves to a more dynamic pricing model, it makes it a lot more unpredictable and more complex. Maybe this will be the next step they take, but so far, I think everyone likes the predictability of the timeliness as well as the price.

BurningFrog2• 11 years ago

The current regulatory structure (in reality) sets prices on behalf of the Taxi companies.

This is as predicted by Regulatory Capture theory, which is one of the most important concepts to know for anyone trying to understand society.

femto1132• 11 years ago

Ah, the myopic business school view of the world...

The second effect of surge pricing was to allocate rides to those that value them most. Figure 3 shows that, while the number of app openings increased dramatically, the number of actual requests didn’t increase by as much. This came from riders who opened the app, saw that surge pricing was in effect and decided to take an alternate form of transportation or chose to wait for surge pricing to end. From an economic efficiency standpoint, this was highly beneficial because those that ended up requesting a ride are those for which their outside option was worse, leading them to value Uber more in that particular moment. The gap between the red and blue line could be tentatively interpreted as a measure of this allocative efficiency.

Allow me to translate: people who couldn't afford surge pricing didn't take Uber, while those wealthy enough to be indifferent to price still did. Surge pricing is thus not as much a means to increase capacity as it is to get the huddled masses out of the way of the 1%.

gohrt• 11 years ago

Why is HN upvoting someone's poor-quality school homework assignment?

jmakeig4• 11 years ago

"Couldn't afford" is a loaded term. I can afford a helicopter ride home, but I don't value its utility enough to pay what's being asked. Isn't that how prices are supposed to work?

femto1132• 11 years ago

Only if you equate price and value, but given vast discrepancies in wealth it doesn't really make sense to do that, since an extra $20 to someone making $250K/year doesn't have the same meaning as one making $25K/year. Consider the old way of doing such things, where there was a queue to get to the taxis, but everyone paid the same price. In that scenario the best measure of value is probably how long someone is willing to wait in line. Now imagine that same queue except anyone willing to pay twice the normal cab fare could just cut in line. Would you call that "allocative efficiency"? Or perhaps you would simply call it unfair?

brc3• 11 years ago

Why are you even on this site? If you think rich people are out to get the poor, why spend time on a site whose themes are technology and wealth creation through technology? Why not go down the park and shout slogans and wave placards at the passing '1%'

You impugn morals on price activity when there are none to see. For all we know, the high price takers wer students late for an exam, a young man determined not to be late for a first date or a relative rushing to the airport to see their family arrive.

It is far better for supply to increase when needed than for supply to remain fixed and the market to stay uncleared. There is nothing myopic, wrong or class based about that, and it is childish to even look for it.

wfo2• 11 years ago

While the 'how dare you consider those less fortunate than you, you don't belong in our tech paradise where only self-serving extremist libertarian thought is allowed' line of argument is particularly revealing about your attitude it isn't particularly welcome or mature. It is childish to not notice the disparity. I don't even disagree with any of your points, but your obvious hatred for anyone who doesn't support the oligarchy is just gross, I think if you tone it down you'll find people much more willing to agree with you.

GregBuchholz• 11 years ago

So who else believes the term "Surge Pricing" is horrible? What if instead of charging more for busy times, they charged less for slack times, and call it "off-peak" pricing, with bigger discounts for times when there are fewer riders. The rate table stays exactly the same, it is just a matter of the optics of how things are presented. Now the 99% get deeply discounted luxury black car service when the 1% aren't using it. I suppose some of the "rich" won't like the association of using a discounted service, but I'm sure no one is going to get extremely angry, like some people get with the current nomenclature. Seems like someone dropped the ball in the marketing department because their libertarian mindset didn't take human psychology into account. Thoughts?

dnautics6• 11 years ago

Lyft tried this and called it "happy hour". The optics of this on the driver side were atrocious, driver morale went through, the floor, and quality of service as well.

shabda• 11 years ago

This is also going to make a base pricing look uncompetitive with the regular taxis, and they will use that in advertising.

hiou• 11 years ago

Because transportation is a vital element of our economy. The same problem with price gouging. Very short sighted market lock ups turn into much bigger problems. I guess we just need to learn this lesson once again.

Edit: I should add that that the biggest problem with gouging in a market is that the sellers actually over price.

dnautics6• 11 years ago

Have you ever actually used one of these services? If you can't afford it, the best strategy is to wait. Having driven for both of them - I can say that among my clients have included: A person with felony manslaughter on his record who had to get to his work on time (getting fired is horrible you are a felon because getting a new job is a non-trivial exercise). An enlisted (very junior) military family of 6 (only 4 were in the car) who needed to get some late-night shopping done at a distant walmart. The father was on deployment and only had a high school degree. Because of the high cost of taxis and the inconvenience or slowness of public transit, these were NOT options for them. Not to mention the scores of homosexual clients who were vehemently opposed to taking taxis because of repeated taxi drivers who had harrased them (this doesn't happen as often in a Lyft/Uber because of the rating system).

So, you can rail against the surge pricing, or think of it this way: Because it's something that happens largely during luxury times (e.g. bar closing) that taxes those who are wealthy and have disposable income, subsidizing affordable, safe, and convenient travel for the less privileged during other times.

andygates• 11 years ago

"luxury times" ? Is fair-priced travel a luxury now?

ufmace2• 11 years ago

Translated again: People who value it the most highly can get it when they need it, while people who don't need it as much can wait or find alternatives. When demand is highest, they do the simplest and most effective thing to draw in more drivers who are free agents - offer them more money.

In effect, the earlier riders who accept the surge and book anyways pay more to draw more drivers in, and in return get their rides first. Riders who can't afford or don't want to pay for the surge prices can just wait for more drivers to come and the earlier demand to be satisfied. It's effectively the same as a crowded cab stand, but instead of the arbitrary order of who got there first, it's ordered by who's willing/able to pay the most, and thus who getting that ride fast is the most important to. A lower-income person who needs to get somewhere on time can sacrifice a little more and get it, instead of having no options.

FWIW, the actual 1% most likely have actual private drivers and entourages. And are important enough that almost anything they're going to will wait for them.

drinkzima• 11 years ago

The fact of the matter is that surge unambiguously causes drops in demand. What is less clear (and more interesting) is understanding the casual, supply-side effects of surge, which isn't really tackled in this paper (one mention of the word 'casual').

Definitely some nice charts in here though, would be a fun data set to play with.

michaelt• 11 years ago

I wonder why the graphs of driver supply and rider demand don't show the surge multiplier against time?

Given that footnote 6 indicates the price increase varied between 20% and 80%, it would be interesting to see how long it takes for a price increase to cause increased driver numbers.

brc3• 11 years ago

I think looking at the supply elasticity would help to determine at what rate prices have to be escalated in order to generate supply. I would also imagine that the elasticity of demand is dependent upon time of day, day of week and even weather.

Mismatch between the demand elasticity and supply elasticity would be a fascinating area to explore here - when prices increase quickly enough to get driver active, perhaps they increase too quickly to match passenger demand. Ideally you could have a set of curves that predicted best market clearing on a multiple variable set and have that automatically take place.

There really isn't that many markets with variable pricing linked to a physical service or good, so I think there is a lot that can be learnt here.

dnautics6• 11 years ago

because it's not a monolithic value. 1) it's geographically varied. 2) Uber calculates the surge, but if it can't push that value out to the passenger-side app before the ride is arranged, it must honor the surge value it presented to the passenger. So at any given time, due to network latencies, and the like, there are going to be multiple values for the 'surge'.

dnautics6• 11 years ago

I do think that the claim that surge pricing works to increase drivers in the area - while apparently borne through the data, might not have as solid a mechanistic grounding as the writers presume. I would encourage the authors of this paper to actually go out and drive and test the presumption of agency in the matter through experience. Drivers (in their social networks) always say "don't chase surge". It's usually good advice.

gavazzy• 11 years ago

I think this will create a damped feedback loop. Over time, the delay caused by "don't chase surge" will mean that the prices remain higher longer. Drivers will realize that the surge doesn't instantly vanish. As a result, surges will mostly get the drivers who were about to end a shift, or were going to begin a shift in an hour or so, or are in the most need of $$$.

This behavior means that surge pricing will result in a smaller increase in drivers, but a one that more closely resembles demand, which is exactly the kind of efficient allocation we'd hope for.

awch• 11 years ago

After reading some of the surge-related threads on UberPeople.net, it sounds like drivers say "don't chase the surge" because surge pricing immediately results in drivers saturating the surge area, causing the surge pricing to disappear.

I would be interested to learn more about the profit-maximizing strategies Uber drivers employ, such as whether or not to "chase surge." Any idea where to look?

dnautics6• 11 years ago

The best thing is to just learn your clientele. I was regularly driving 30-ride days on Lyft where 75% of my rides were at elevated pricing. I do have a PhD in chemistry and a degree in math though, so I had somewhat of a career in analytical thinking and pattern recognition prior to becoming a driver.

If you have questions, feel free to drop a line (contact in profile), I drove in two cities for about a year (I do contract coding now)

ufmace2• 11 years ago

Don't know how it actually goes, but if "Don't chase surge" is a saying in some communities, it's probably because chasing surge is such an accepted thing in the driver base as a whole.

cowpig• 11 years ago

This reads like a press release from Uber

mra1385• 11 years ago

The topic is interesting but this analysis/write up is very weak. I have a bunch of question but here are some:

1. Why only two data points? That doesn't really provide us with enough information to make a judgement about ubers surge pricing algorithm.

2. Why not look at an example where an increase in demand is not predictable (ie end of a concert) if this issue was mentioned as a weakness in the first example in the study.

3. Is it really the case that NYC cab drivers are sitting at home monitoring surge pricing at 11pm before deciding whether to get in their car and drive to Manhattan or wherever there is high demand for rides in the city to start giving rides? It seems most likely the case that drivers would assume there would be some surge and would already be on the road esp in a place like NYC on New Year's Eve. It's not clear to me that this data point really works. I can't imagine a driver is sitting with his family celebrating nye and waiting for surge pricing somewhere in NYC before hopping in his car to give a ride.

vishveshs• 11 years ago

Nice analysis! I think the effects and implications of surge go beyond UBER itself. For instance, in India, when a surge happens most of the UBER drivers tend to move away from the area. Primarily because the competitor's prices are way lower and people would just not call an UBER at all. In this case defeating the whole purpose of Surge.

Would be interesting to see other implications of Surge

The Effects of Uber's Surge Pricing: A Case Study [pdf]

Related Stories