What If It Rains?

Follow Up To: Full Service

Gas stations in New Jersey do not have awnings to protect customers from the rain. Let us presume for the moment that due to some combination of inertia, regulations and initial costs, it is not practical for them to create such awnings in the short term. Now suppose that New Jersey legalizes self-service. 90% of the time it is not raining, and the customers gain $0.25 of net utility from the cheap gas that comes from self-service. However, 10% of the time it is raining or snowing or something else unpleasant, and in those cases customers would on average be $5 better off if someone pumped their gas for them so they didn’t have to get out of the car. Some even get to avoid ruining their $60 haircuts.

A few stations switch to self-service. Ninety percent of the time, the sun is shining, and most customers choose to save a marginal amount of money and go to the self-service stations. With business drying up, more and more stations switch over, until most stations are self-service.

Then the rain comes, and everyone gets wet and miserable. The customers complain about how good things were in the old days with full service and say there ought to be a law!

One chain tries the obvious, which is to stay full-service, take a small loss when the sun is shining, and then charge an extra $4 per customer when it rains to make up for it. The customers go ballistic. Profiteer! Exploiter! Price gouger! Boycotts are announced and state senators are lobbied. Governor Christie warns that this is illegal activity under his interpretation of New Jersey law, the same way gas stations were not allowed to raise their prices after hurricane Sandy resulting in no one being able to buy gas without huge lines. The public cheers and his presidential campaign gets a boost, and the last of the full service branches go down. Self-service wins, everyone is miserable, and legislation gets reintroduced to return New Jersey to full service. Christie boasts that the repeal, passed over his veto, was a huge mistake, and adds mandatory self-service to his national platform, but loses anyway because he’s Christie.

What went wrong? Fundamentally, the problem is that consumers are unwilling to let prices change to meet supply and demand. Allocating scarce resources to those who value them most, and rewarding those who provide those resources so that people are encouraged to keep potentially scarce resources on hand, makes you history’s greatest villain. In the case of rain and gas stations, a few hairstyles get ruined and people are annoyed. During peak hours, movie seats and restaurant tables go to whoever was willing to make plans far enough in advance, rather than whoever actually values the experience. During real natural disasters like hurricane Sandy, it means the stores all run out of food and bottled water, and the gas stations run out of gas, and everyone spends all their time on lines, because the people who have real need can’t buy.  This isn’t getting better.

When it rains in New York City, people go out onto the sidewalks with umbrellas. They charge a premium price. This is a very good thing, because otherwise they would not be standing by waiting for it to rain, and if I get caught in the rain without an umbrella, I can get out of it. Technically, however, they’re breaking the law, and people get pretty mad about this, calling it ‘taking advantage of people.’ I often wonder about that term, it also came up last night at the rationalist meetup: Companies are taking advantage of people! You don’t say! Is that supposed to be a bad thing?

This is not a call for people to understand economics, grow up and pay these hardworking Americans their due for keeping emergency supplies on hand. If I thought that would work, I would totally do that, because that would be a huge win. You never know; please do share that message! Maybe Uber’s hard work wearing the black hat will warm people up to the idea a little.

The problem I’m actually wondering about is related but less clear, and has no obvious right answer that I can find. What people regularly buy allows profit to be made on a regular basis, and that is how you pay the overhead to run a business like a gas station or a store. If something is being bought all the time, it also means that a lot of places are offering to sell it. That item in general (e.g. gas or milk) provides a ton of utility if the alternative was to go without, but the utility of another place to get gas is only large if the next gas station is far away. If it is across the street, the only utility is that they are keeping each other honest on price, so you might save a few pennies.

The sheer amount of gas purchased is causing placed to have great incentive to hijack sales from each other. When this means price competition or quality competition, that’s great; without price competition we’d pay a lot more for most things. When that means a race to the bottom on price because everyone only has enough attention to compare prices, resulting in a sacrifice of everything else (for example, on airlines, which have migrated their fees into secret compartments) or the loss of a little extra service when that service would have been efficient (in the sense that it’s easier for an employee to do a thing, or their time is less valuable than yours, or something similar). Why in the world are customers now bagging their own groceries? It makes no sense… unless people are being trapped by comparisons into looking at headline prices and this is forcing the hand of grocery stores.

Even more toxic is the rush to hijack sales based on location. If I have a source of milk two blocks away, that is good enough for anyone, and I’d prefer to have a diversity of other things available for purchase at other places. Instead, someone opens up another drug store one block away… often a copy of the same store that is also two blocks away! The reason they do this is to hijack sales by being marginally closer, and therefore better, than the competition. The result is that a huge percentage of store fronts end up being effectively wasted on drug stores and banks, and endless copies of the same chains, when far more utility could be had with something more unique, but which would get less business.

I have no idea what to do about this, even in theory. But it is clearly a problem, and it would be great if we could solve it.

This entry was posted in Death by Metrics, Economics, Impractical Optimization and tagged , , , . Bookmark the permalink.

9 Responses to What If It Rains?

  1. glenra says:

    Are you using a non-standard meaning of “utility”?

    If there are lots of places to buy milk then you as a consumer get lower prices, shorter trips, better service including but not limited to less time waiting in line (because there are lots more lines and people-making-change in parallel for the same kinds of products distributed all along the street rather than a big single line at one central location).

    All the duplicate stores are likely to compete away most of the producer surplus, so they’re not especially profitable. On the margin, a bunch of them will be just BARELY profitable.

    If “far more utility could be had” by a store selling something unique, presumably some of that utility could be captured as producer surplus – enough to be MORE profitable than that marginal now-out-competed drugstore in a less-favored location. So why doesn’t the marginal drugstore shift into selling this new unique good or service /alongside/ whatever they were selling before (like they did when a thousand stores all decided they needed to offer “paninis” or “wraps” alongside more traditional sandwich types) or why doesn’t the Unique Thing Company start buying up marginal drugstores and converting them entirely to Unique Thing stores?

    My guess is that there’s actually “far more utility” in having lots of access to drugstore/sandwich shop/supermarkets and banks.

    I would also submit in particular that a source of milk two blocks away might NOT be “good enough for everyone” where “everyone” includes people with age-related handicaps.

    • TheZvi says:

      The standard economic model says you are correct; the prices will lower until the customer is exactly indifferent, thus eating away all the surplus and making things more profitable, and service will increase.

      However that is not what actually happens. Service does not get better, because they staff in proportion to demand; waiting in line is basically a constant. Sufficiently small locations (e.g. delis) do reduce it slightly because chance of zero wait goes up since you need staff of one but only need half a guy. We do know that, perhaps partly for this reason, the surplus is not competed away (and cost to the place goes up due to increased spoilage). Instead, prices in dollars go up rather than down, since people make their choice on location often enough that it is better to capture those who are price indifferent; the supermarkets are the only ones who compete seriously on price.

      Supermarkets are really different from each other, and do compete on price (not location by location, but in general they do) so I do think more supermarkets are good, but they are being hurt by the drugstores hijacking common items. I think this is even causing more people to do delivery more often because there’s no other food naturally in the house.

      Drugstores need to be chains and standardized, for various reasons, and doing non-standard things with a business is way harder than it naively seems like it should be.

      The problem is it’s easy to observe that extra locations don’t cause competition on useful metrics other than location; the idea is that the human is lazy and doesn’t actually choose the utility-maximizing solution – the decision algorithm is not equal to the utility algorithm. We are not homo economus.

      Given delivery services and the ability to choose exact location, yes, I’d say having universal two-block coverage, with the median at one, really is enough.

      In conclusion, it’s not that I am using a different definition of utility; I’m instead saying that in practice things don’t do what the economics basic models say they will, because the “correct” choices are unnatural and decision algorithms are flawed. Ideally I’ll flesh this out more later.

  2. glenra says:

    When airlines weren’t allowed to compete on price they flew lots of nearly-empty planes most of the time so as to compete better on convenient routings and times; the ubiquitous corner delis that have nearly-empty stores with no line most of the time seem to me are doing much the same thing – making life more convenient for me (which is worth something) at substantial cost to them. Location convenience /IS/ a form of quality competition. It’s reducing a price – the price in time and aggravation to get our purchases.

    So I guess I’d like to see some /evidence/ that “in practice things don’t do what the economics basic models say they will” alongside the assertion. Is this something you can measure?

    Maybe the right question is “compared to what?” What would the world look like if this weren’t happening? If you were the zoning czar able to decree by fiat that corner store delis and banks are only allowed every other block, are you SURE we’d see “more unique” storefronts interspersed in all the remaining spaces rather than a dead zone? Are there any towns that operate the way you think they “should” in this regard? As Czar, would you also ban apparent clustering of other businesses, like a diamond district, furniture district, music district, fashion district?

    (Note that another way of saying “the human is lazy” is “location has surprisingly high utility”.)

    • TheZvi says:

      I like the airline metaphor, and sure, ‘humans are lazy’ can mean that location has high utility, but it’s also that humans in the moment, when making decisions, are much lazier than they would prefer to be taking the long/outside view, and they make decisions they would, if you asked their system-2, correctly predict they would regret. Also, they will choose metrics they can easily evaluate (location) over metrics that are harder to evaluate, when making choice. Airlines have gone the other way and now compete dramatically on headline price, because that’s the filter easily available to consumers, so they then pick up the dollars later in fees and squeeze out things customers really would prefer to have, or sell them back to you at huge markups. The way you present a decision makes people decide differently.

      The standard economic model assumes that you will properly evaluate the factors related to your decision, and that simply is not the case in practice. Certainly a huge percentage of NYC residents would explicitly prefer to have less drugstores in their areas, and I have never heard anyone think they needed more (in a city, as opposed to a small town).

      Alternatively, you can create a simple toy model that gets you the result under a standard set of assumptions: Presume there is a chain of stores, and you can choose to open a store at any location as a franchise (reasonably close to accurate). The chain tells you what prices (the create marginal profit P) you can charge and what your store must look like, so the only choice you can really make is where to go, and some minor stuff like your hours (again, pretty accurate). Assume that consumers buy, with a uniform distribution over time and space (for simplicity), from such stores with quantity Q, and will always choose the store nearest to their current location (or on their route, or whatever, again reasonably accurate). Consumers get U unit of utility net of the fixed price set by the chain, minus E for each extra block traveled. You can easily see that if U>>E (so that almost all customers buy under equilibrium, rather than give up), the number of stores if F(P*Q), and is unrelated to U or E.

      You can say, I think E is high, so E(E)*Q matters. It matters some, sure, but to give you an idea of how far things went: I walk out of my last apartment. There is a Duane Reed (they are all the same, except for size):
      1: Across the street to the North.
      2: A half block West.
      3: A block East.
      I do not think any plausible calculation says that this is necessary. Likely, they open so many in part to box out potential competition from CVS or RiteAid (or at the time Walgreens, which has since bought Duane Reed).

      So, as I said, I don’t have a good solution to the issue, but yes I think we have reasonable evidence that store fronts would not lie vacant; the equilibrium rent might decline, or it might not, but the market would certainly clear, at least in big cities. There are lots of valuable uses for such an asset, and driving down the price would enable additional purposes to be viable. New York back in the 80s and 90s looked more like this, and I think the best big example today is probably Portland.

      In general of course appointing czars and doing zoning ends up doing far more harm than good; this problem is far from the worst thing that could happen when you start zoning this way. That’s why at most I’d be inclined to do is to tax. Designing the tax properly is a hard problem. Certainly one can impose a tax based on how close you are to another franchise of the same chain, perhaps with X-squared decay, and I think that would be useful, but it misses much of the target (e.g. pseudo-chains like the Delis or Generic Chinese Restaurant, which is much closer to a chain than people realize even if it isn’t actually a secret chain outright). Hitting the full target is one of those libertarian impossible nightmare problems.

      As for the districts, that’s the opposite; I think those are great. What the districts are doing is, they are making it easy to compare and shop around; you know that if you need a diamond you can go to the diamond district and try five stores in three blocks, if you want nice flowers you can go to 29th or whichever street that is and look around. I see a lot of value in that, even if it lowers the utility of living on that exact street, which it likely does.

      • glenra says:

        > The standard economic model assumes that you will properly evaluate the factors related to your decision

        Not really. The model has predictive power because it assumes people /in aggregate/ will tend to behave /as if/ they were doing something along those lines – it doesn’t mean people are individually making that calculation. Plus change happens /on the margins/ and you’re not the customer that this arrangement is optimizing for. So imagine that it’s SNOWING and you live in a side street apartment a block or two or three away from this CVS cluster. For THOSE people the width of the cluster means trudging one less block each way to the nearest store in the cold or rain or snow. Might be worth it, no?

        I suspect the underlying problem here is that “interesting” stores have gotten priced out of your neighborhood and you’d be happier moving somewhere where the rents are cheap enough that companies can afford to take more chances. Maybe Park Slope where there was a “Superhero Supply Depot” on the main strip. Or Greenpoint where every other block has a polish restaurant, a bakery, a meat market, a nail salon, and a dance/yoga studio.

        The reason I brought up districts: we’re agreed that diamond stores benefit from clustering together where the customers know to go and the cops and religious authorities can efficiently watch over the trade. And that furniture stores benefit from clustering together where the streets are wide and the customers know to go and the delivery operators know how to safely park the truck, and so on. But this is a problem! Not for the utility of the people living on THOSE streets, but for the utility of the people living on YOURS. To wit: if most of the jewelry stores and furniture stores and theaters and music stores and flower stores do best if they self-segregate into their own districts, what’s left to populate YOUR neighborhood?

        Answer: Banks, CVS stores, restaurants, and deli/grocery bodegas.

        P.S.: If you DO want to move to Greenpoint, better move fast. Just last week an AMAZING weird only-in-brooklyn thing 4 blocks from my apartment called “The Brooklyn Night Bazaar” had to close so its warehouse space could be converted into…a BMW dealership. :-(

  3. TheZvi says:

    I’m actually quite happy with my neighborhood; where I am now there isn’t even a particular problem, and where I was before was actively getting better rather than worse. I like to optimize, but this isn’t an especially urgent problem for me at the moment in the Lower East Side, which has basically avoided the problem, whereas the Upper West Side where I grew up has very much failed to dodge.

    I agree that the clusters do drive certain things out, but I still maintain there are lots of things (and in fact my current location shows this; the only actual complaint I have is that if you go east, there isn’t anything because there’s a giant river, and without an especially good park-front to go with it).

    To me, the underlying key question is, under what circumstances does the marginal change model cause perverse outcomes, and can anything be done about it? Where does the market fail to optimize? Why does this happen and where does the standard model break down? I think a lot of the breakdown is that people are unwilling to pay what rarer things are worth to them when buying the marginal thing, because the price is ‘unfair’, whereas the price of common things does not get competed down to zero profits because of a combination of implicit and explicit collusion and the understanding that fixed costs are a thing. You can see people’s preferences are contradictory because e.g. they move to locations to have access to things, which implies a very high utility by revealed preference, then if those things are even a little “too expensive” they do not buy them. Whereas the effects on property/rent values say that the things are vastly cheap, the same way that they show how much a decent school is worth.

    And very sorry to hear about Greenpoint. That sucks.

    One solution is actually to have a single owner for a reasonably large area, and have them make the decisions with rent/property values in mind, perhaps even a whole semi-planned city. The results would be very interesting.

  4. SB says:

    I think this is just a special case of the general problem of competition. It is naively inefficient almost always to have competing firms. The marginal cost for Apple to make more iPhones to replace Samsung phones would be much lower than the total cost of Samsung’s phone business. Why does Darwin tolerate the fixed-cost overhead for a zillion different burger joints?

    I think we have sort of empirically decided that the massive inefficiencies of competition in market economies are desirable relative to the massive long-run inefficiencies of control economies. But… there’s no reason that should necessarily be true in all possible worlds.

  5. Pingback: Self Service Roundup | Don't Worry About the Vase

  6. Banananon says:

    > Fundamentally, the problem is that consumers are unwilling to let prices change to meet supply and demand. Allocating scarce resources to those who value them most, and rewarding those who provide those resources so that people are encouraged to keep potentially scarce resources on hand, makes you history’s greatest villain. […]
    > During real natural disasters like hurricane Sandy, it means the stores all run out of food and bottled water, and the gas stations run out of gas, and everyone spends all their time on lines, because the people who have real need can’t buy.

    If I am willing to spend a fifth of my budget on bottled water, and you are only willing to spend a tenth of yours, who actually values the water more? If you have three times as much money as I do, does that mean you do? Does Bill Gates or Warren Buffett (or for that matter El Chap Guzman) implicitly deserve things more than you or I?

    I think people have a strong kneejerk reaction against “taking advantage of people” because they implicitly believe that things like bottled water and gas provide similar utility to everybody. From this perspective, dynamic pricing is another factor which makes life harder for those who are already disadvantaged and are indeed ‘gouging’ people.

    Which is not to say markets are bad — they’re clearly the most efficient, reliable resource allocation method we know of. So what should be done? Maybe Basic-Income-something-something? Government as coordinating mechanism? Figuring out an adequate solution is much harder than diagnosing the problem.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s