Pur Autre Vie

I'm not wrong, I'm just an asshole

Sunday, July 26, 2015

Why I Disagree with the Ubermenschen

In a departure from my normal blog topics, I'm going to write about taxi regulation and regulatory policy generally. I'm going to start with a little side-discussion, which you can skip if you only care about the sexy Uber debate.

Side Discussion: Regulatory Policy and Distributional Concerns

In general, U.S. consumer regulation is aimed at making things extremely safe and extremely cheap. The classic example is food safety regulation. I can't find it now, but when New York City started giving letter grades to its restaurants, there was a humorous item in the New York Times in which low-end restaurants with good grades were asked for their tips, which were passed along to fancy restaurants with low grades. The suggestions were along the lines of, "Cook the meat until it is gray all the way through." (I'm probably exaggerating a little, but it was a funny piece.) Of course the same thing goes for cheese made with unpasteurized milk.

Note an important feature of this type of regulation: it enables consumers to choose the absolute cheapest option at the grocery store without trading away safety. In a libertarian society, people would be free to sell contaminated or adulterated food, and consumers would have to pay a premium for safe food. Being "frugal" would mean risking death for you and your family. In the U.S., all of the food is safe, and suppliers compete heavily on the basis of price and quality. There is no premium that consumers have to pay for safety. You can pick the cheapest milk at the store without getting sick. (I'm exaggerating slightly again—food isn't perfectly safe here, but it's reasonably safe. Also, I doubt there is much of a correlation between price and safety. There might even be a negative one.)

The same priorities are visible in airline regulation. It is dirt-cheap to fly in this country, and it is also extremely safe. But it is also very uncomfortable. It didn't have to be this way. Before President Carter deregulated the airlines, flying was comfortable and expensive for everyone. After deregulation, things were basically bifurcated. You could buy a coach ticket and suffer all of the indignities that entails, or you could pay a lot more and fly first-class (or business-class or whatever).

There's a good argument for the Carter approach. But let's point out an alternative. Instead of regulating solely for price and safety, the regulators could also regulate for comfort. (A very quick side note: the antitrust regulators, who can oppose airline mergers on competition grounds, may not be on the same page as the FAA in terms of keeping prices as low as possible. I'm going to ignore that complication.) Imagine that the FAA required a minimum 32" of legroom. (My understanding, based on very little research, is that this is the amount of legroom provided by Southwest and Virgin, but more than the legroom provided by Delta, American, U.S. Air, United, and Spirit.) Now picture a consumer who needs at least 32" of legroom to be comfortable, and who lives in a city that is best served (or solely served) by airlines that provide 31" or less. The FAA's hypothetical legroom regulation makes this consumer much better off: he can fly coach in perfect comfort, whereas in a no-legroom-regulation environment he has to pay for a first-class seat or fly in discomfort.

Who is worse off when we (hypothetically) regulate legroom? Anyone who prefers to pay less and get less legroom. So legroom regulation would result in a lot of consumers paying a little more, and a few consumers paying a lot less. (The latter are the consumers who, absent legroom regulation, would buy first-class seats.) Whether this is a net improvement or not depends on how you weigh these different preferences. Are you tall? Are you rich? Then you probably favor legroom regulation. Are you short? Are you poor? Then you probably don't. There are people who think cost-benefit analysis is well-equipped to make these kinds of distributional determinations. I think that's ridiculous, but that's a discussion for another day.

Back to the Main Discussion: Uber and Pricing

New York City taxis are supremely cheap. This is by regulatory fiat: if a taxi charged more than the going rate, even with the customer's consent, the operator would be punished. Taxis are also unpleasant in a variety of ways. They are not ideally comfortable, and they can be difficult to hail at certain times and in certain places.

Uber represents a simple proposition: the regulators have set the equilibrium in the wrong place. It would be better to establish an equilibrium in which it is about twice as expensive to get from point A to point B (and many multiples of that at certain times), but in which it is easier to hail a car (if you have a smartphone) and the car is more comfortable. Uber also represents the proposition that car service should not be used to subsidize mass transit, and that drivers should be able to decline to pick up certain customers. (Taxi drivers are subject to a surcharge, which funds the MTA, and are not permitted to decline to take anyone anywhere within the 5 boroughs.)

But let's focus on price, since Uber's stance on the MTA surcharge is basically indefensible. Who wins and who loses if Uber comes to dominate the industry? Well, the big winners are people who are willing and able to pay extra for comfort and convenience. These are affluent people who have smartphones. Who loses? Poor people and people without smartphones (there is presumably lots of overlap in these populations). It's the airline example all over again, except now it's not about how tall you are, but solely about how affluent you are and how much you are willing to pay for comfort/convenience. It's pure class war, and Uber is on the side of the rich people.

You might argue that it's okay to bifurcate service between cheap, uncomfortable cabs (for the masses) and expensive, comfortable Uber cars (for the affluent). (This is, of course, how it has always worked, except that the line between the two was drawn much further up the income scale.) Poor people will still be able to get rides at the same old price, the argument goes. But rich people can enjoy better service while paying through the nose. They should be allowed to make that choice.

But that assumes the market can support two equilibria. As Uber takes more and more market share, taxis will thin out and it will be harder and harder to hail them. The customer mix will also change: taxis will serve a disproportionately poor clientele. I assume, perhaps unfairly, that this means that taxi drivers will have to put up with customers who are less likely to pay and less likely to tip generously. Taxi service may become so shitty and unreliable that the equilibrium can't be sustained: the more taxi service shrinks, the more people are driven to Uber, and the dynamic could go all the way to zero taxis. (Bear in mind that the playing field is not level. Taxis are more heavily taxed than Uber and they are subject to a variety of burdensome regulations designed to achieve fairness. So Uber is fighting an opponent that will forever have at least one arm tied behind its back, making it relatively easy to drive taxis out of business entirely.)

Uber loves this prospect. Uber hates any regulation that favors a low-price equilibrium (low prices mean low profits), and Uber is happy to flout the regulations if the regulators aren't sufficiently... cooperative. Uber is also throwing its political weight around, gambling that the upper middle class can be rallied around its banner at the expense of working-class people, who are poorly organized. In American politics, you should generally bet that a coalition of upper-middle-class people is going to beat a coalition of working-class people every time.

That's fundamentally why I despise Uber. Uber is seeking to undermine a regulatory equilibrium that, while far from perfect, serves working-class New Yorkers very well. The best argument for Uber is that by operating what is effectively a taxi service outside the scope of taxi regulations, it will force the regulators to adopt better policy. (This is basically the same as the argument that it's good for some states to have weak environmental regulation, because this will lure employers away from high-regulation states and thus encourage states to improve—that is, weaken—their environmental regulation.) I'm skeptical, because Uber is emphatically not going to start complying with regulations once they permit e-hailng. In fact, the Taxi and Limousine Commission already has an e-hailing pilot program, and yet Uber continues to operate outside the bounds of taxi regulation. Uber's whole business model is inconsistent with price regulation. Uber isn't about e-hailing. It's about charging far more than taxis have traditionally been permitted to charge in New York, while paying fewer taxes and shouldering fewer regulatory burdens. So Uber won't ever comply with a regulatory regime that provides cheap taxi service to New Yorkers.

It is possible to argue that the TLC has gone too far in keeping prices low, just as it is possible to argue that the FAA should regulate comfort. Certainly some people would be better off if taxis were twice as expensive but considerably more comfortable and convenient. But that is the argument people should be making: taxi service is too damn affordable for poor people, and it sucks too much for affluent people. This debate isn't about innovation or "disruption" or free markets. It's about who wins and who loses when we regulate taxicabs. And it's about whether the city gets any say in what taxi service will look like for its citizens.


Blogger Sarang said...

I'm hazy on the empirics but there are two premises that bug me in this analysis:

1. Is it true that Uber is massively more expensive than taxis (assume it's not a surge -- these probably do not coincide with poor people's taxi needs)? In Boston my sense is that the opposite is true. This is also consistent with Mark Kleiman's data from Los Angeles (somewhere on samefacts.com). Do you know of any data on this?

2. How much of the taxi-using demographic actually consists of poor people? Again, NYC seems like someplace where you can do most things using public transit, so the main reasons I imagine poor people using taxis are (a) disabilities of various kinds, (b) bulk grocery shopping needs, (c) emergencies. Even NYC taxis are very expensive in an absolute sense if you use them regularly (like $20+), so the correct solution to (a) is not taxis but paratransit programs like (in Boston) the RIDE. So for the most part we're talking about rare/irregular expenses, and I'm not sure how these actually play out in terms of people's budgets.

2:30 PM  
Blogger James said...

It is difficult to compare taxi prices and Uber prices because they are based on different formulae (taxi cabs charge per mile but also by the minute in certain situations). At some point, I went to Uber's website and compared the price for a ride against a taxi route I have taken many times. The Uber price was about twice what I typically pay in a cab. I ran a few other comparisons as well and came up with similar results. I am not very familiar with UberX - my understanding is that for a while UberX was offering promotional low prices, but I don't know if that is still true. Of course once taxis are done away with, Uber can raise prices to the market level, which is presumably higher than the regulatory ceiling currently in place.

Poor is of course a relative term. Do poor people take airplanes very often? Probably not. It is still thought to be desirable to keep prices down, even at the expense of comfort. You're certainly free to argue this is the wrong choice. By "poor" I really mean people who prefer cheap cab rides to expensive Uber ones.

I wonder what motivated Kleiman not to test the TLC cab-hailing apps. Possibly the adoption is so low at this point that there isn't any point.

3:17 PM  
Blogger Sarang said...

Not sure about not TLC app (you could ask him in comments?), but in any case Kleiman does not find a significant fare differential. For the couple of Boston routes I checked (home to Logan, home to Fenway) UberX is about half the price of a taxi. This is also consistent with what I've heard from people in the Boston and SF markets.

I think it is relevant whether a taxi is a luxury good or not, so don't find your notion of "poor" sensible. Are poor people also those who buy generic tablets instead of ipads? Taxis are not analogous to airplanes because (1) there is no plausible alternative to flying across the country, (2) planes are expensive enough in absolute terms that small percentage differentials are important. (By the way, I will observe that the same people who are Uber users consistently take the cheapest flights regardless of comfort.) There are some situations in life when one has no alternative but to use a taxi, but these are rare enough (and the dollar amounts for either taxi or Uber are modest enough, i.e., mostly under $100) that I do not find the potential hardship narrative plausible for NYC, DC, or Boston. Perhaps most of the people who prefer cheap taxis to expensive Ubers (accepting that assumption arguendo) are merely those who use taxis much more frequently? Maybe raising the price just makes them walk half a mile and take the subway instead?

4:28 PM  
Anonymous Anonymous said...

(This is James.)

A few points.

First, I will note that NYC is an incredibly poor city, one of the poorest in the country. There are plenty of poor people here, and I'm sure that like most people they benefit tremendously from having access to cars from time to time. I think it makes sense to try to keep things as cheap as possible for them, even if affluent people might prefer a much higher price point. I feel roughly the same way about most regulated monopolies.

But if you really think that it should cost twice as much to take a car in NYC, then it seems to me the right regulatory response would be to double the metered rate and allocate most/all of the extra revenue to the MTA. This would raise, I believe, about $2 billion a year, which would go a very long way toward solving the MTA's funding problems. You could also allow yellow/green cabs to take hails via smartphone, providing the convenience of Uber. This seems to me a vastly better solution than allowing Uber to take over the market while providing $0 to the MTA.

Alternatively, if UberX costs no more than a metered car, you could simply order UberX to meter its cars and pay the surcharge. This is no burden to UberX because it is already charging the metered rate on average.

Second, I don't think it makes sense to view UberX pricing statically. If taxis are no longer available in NYC, then UberX will have considerable pricing power. I don't think it's safe to assume that in those circumstances Uber would feel bound to charge only the metered rate. (Though again, I'm happy to allow UberX to operate however it pleases if it's willing to obey taxi regulations, including the metered rate and the surcharge.)

Third, I wonder if UberX could afford to charge such low rates if it had to buy medallions to support its fleet. You can argue with the wisdom of requiring medallions, but they have the effect of keeping cars off the road, which is a public good.

10:59 PM  
Blogger Sarang said...


I agree that any taxi-like service should subsidize the MTA. For that matter I believe that the roads to enter New York should be heavily tolled and the proceeds distributed either to the MTA or to poor New Yorkers. I am not trying to defend Uber, just trying to understand your weirdly fundamentalist take on pricing. If you are not opposed to innovation in the taxi/rideshare/equivalent industry tout court, surely you should be open to pricing models that are different from metering, as long as they are fairly taxed. In fact if you wanted to make the pricing system less regressive you would probably want to go from metering toward a relatively flat rate, because I bet short-distance taxi riders are disproportionately rich.

I don't think medallions at current or near-current limits are sensible. They would be if taxis were even like 20% of the cars on the road in NYC but they are nowhere near that level. And in any case it would be better to do it via a tax because that would raise revenue.

12:34 PM  
Anonymous Anonymous said...

(James again.)

One thing I didn't get into enough in the post is that I think the status quo is a fragile equilibrium, and I also think that it is probably not feasible (and maybe not desirable) to have a continuous distribution of prices.

As a thought experiment, imagine a whole rainbow of taxi colors, with each charging some multiple of the metered rate. Yellow could charge 1x, orange 1.5x, red 2x. Then the "market" would decide the mix of prices that consumers want. But the point is that the "market" wouldn't necessarily work very well. As the number of yellow cabs thins out, the waiting time increases, and the number of people willing to wait for them would further shrink. You could get an equilibrium with all red cabs, even if only a minority of people actually prefer red cabs.

This is roughly what I fear with Uber. The collective preferences of New Yorkers may be ill-served if the market fragments and then the low-price options disappear. Those low prices aren't a robust market outcome, they're a regulatory mandate. There's no assurance they will be available once we have decided that it is "innovative" to get rid of our taxi regulations.

If you prefer a "market" outcome, then fine - regulations were never going to command your loyalty. But if you think there is anything worth preserving in the status quo, then you have to consider whether it will survive when Uber has displaced the taxi cabs. If taxi pricing were resilient against competing equilibria, then I would not be very concerned.

Medallions are probably not the optimal policy, but I'm not sure they're inferior to leaving congestion totally unpriced, which is presumably what Uber wants. Proper congestion pricing would be great, of course, but isn't on offer.

1:00 PM  
Blogger Zed said...

I think it's wrong to congestion-price or ration taxis if you're not going to also do it fairly to cars. First there are vastly more cars than taxis on the road, and second, to the extent that there's a case for either subsidizing or price-controlling taxis it is because they decrease the cost of not owning a car relative to that of owning a car.

I don't think yellow cabs are "low-price," I think they are hideously expensive, if you compare them to the disposable income of the average New Yorker (or to the cost of public transit). For (say) people earning less than 50k/year they are some combination of luxury good and emergency expense. I don't see how market pricing vs. yellow cabs is plausibly going to burden people who take under ten taxi rides a year. Say ten taxi rides ~300 at current rates and ~500 under Uber. That wedge isn't obviously problematic. Obviously this is a Fermi-problem level of analysis and if you have evidence that working-class New Yorkers (defined in the normal way, not circularly as people who like cheap taxis) are more reliant on taxis than I think, you should share it.

3:50 PM  
Anonymous Anonymous said...

Before going into the model or argument at all, I think many of the stated facts are incorrect. All the reasoning which follows from the incorrect facts is fruit of a poisoned tree.

Cabs are very pricey. The relevant Uber comparison is uberX. UberX is what the app defaults to and is what competes with cabs. Uber non X competes with fancy blackcar services such a Legends Car and Limousine (a locally owned fancy car service serving the Brooklyn area for over 20 year and my preferred choice for airport travel). In my experience, UberX without surge pricing is priced similarly to cabs or slightly cheaper (note my experience is not a randomized controlled trial or even a large study, so it means very little).

Poor people, and even middle class people do not take cabs regularly. They take the subway if they are middle class and busses if they are poor. Cabs are for coming home from an anniversary dinner, or going to the airport, or even from arriving at a bus stop late enough at night that the subway is scary. They are also used for emergencies. Cabs are pricey luxuries or expensive necessities, not everyday amenities.

Cabs are not ubiquitous. The model of street hailing requires streets that are highly trafficked. More, it requires streets highly trafficked by people who can afford cabs and are likely to take them. In order for cabs to make the most of their time on the road, they want to go back and forth between two areas that both meet the above criteria. Poor (and to some extent middle class) people do not live in areas where this is the case. Even on very busy streets in queens, it is hard or often functionally impossible to find a cab. On the non busy streets, it is outright impossible. The same is true in many areas of Brooklyn, even relatively affluent areas. In park slope, it is easy to find a cab at night, but very difficult in the daytime. One of the major advantages of Uber (or similar services like lyft) is that e-summoning works in areas where street hailing is not viable.

To some extent, we have weak data points about if New York can support multiple equilibrium points. Wall Street types and lawyers (hem, hem) have been taking black cars home from work for decades. They do not step outside at 11 and hail a cab, a black car is called or electronically summoned and they step out and get in when it arrives. I know banks have been using websites to call black cars for over ten years, probably longer. That market coexisted with cabs. The black car market has also taken the relatively affluent (and even upper middle class) to and from airports for decades as well. So the existence of ordered cars does not necessary obviate the market for hailed cars (which are much more convenient if you are in the relevant areas).

As to the point about it being ridiculous that Uber is not paying the same taxes, it is obviously correct. Uber should pay the same taxes as taxis, black cars, etc. There is, or should be, no disagreement among right thinking people on this point.

11:06 AM  

Post a Comment

<< Home