Hatfield for Tiebout

My friend and Salem Center colleague John Hatfield has written a characteristically thoughtful reply to “Tiebout Was Wrong - But Why?,” my plenary address to the Public Choice Society. Here’s Hatfield:

---

You argue that local governments are very inefficient in providing local services (“far below [the] efficiency frontier”). I think there are two definitional issues that must be clarified before we can evaluate this statement:

1. What is “local”? I think you would say that L.A. County is a “local government,” but it administers as many people as Switzerland or Sweden. Is that really local? In Tiebout’s formulation, we generally think of each locality as homogenous. L.A. County is not homogenous in any meaningful way, and neither is any “big urban government” in the United States. And Tiebout does not predict that such big urban governments will be efficient! And I am fine with saying that big urban government is very inefficient.

Saying that big urban governments are not considered by Tiebout does not make Tiebout’s claims vacuous, since many small suburban governments are reasonably homogenous. In fact, I would argue that this fact allows us to directionally test Tiebout’s model: Are small suburban governments like Evanston and Naperville or large consolidated urban governments like Chicago more efficient?

Relatedly, Tiebout’s model only works if local leaders feel there is a significant probability that they will be replaced if they fail to provide reasonable quality services; this is certainly not true in big urban governments, but it is at least somewhat true in suburban and rural local governments. (The probability of one vote mattering is, I agree, effectively 0 in all cases. But the relevant probability is the probability of mayoral/councilman replacement if the local government performed particularly badly.) Moreover, many suburban governments are effectively governed by a professional city manager—not a mayor (although a mayor may still exist). And these professional city managers expect, if they do well, to advance in the profession, get hired for bigger and higher-paying jobs, and so on.

2. What is “efficient”? I doubt you would say that the nuclear power industry is very inefficient, even though it does not do a good job of providing electricity at reasonable cost; you would say it is hamstrung by horrific regulation. And local government is a highly regulated industry! You argue at the end of your slides that “Localities can ignore inefficient state/federal laws via selective enforcement.” I’m not sure I understand: If the state imposes high taxes on certain activities by local government—and they do!—how can a local government ignore that? Or what if a local government wants to allow the development of a new industrial facility—just try ignoring the alphabet soup of federal agencies that regulate such activity! So the right definition of efficiency here has to be “efficient subject to state/federal regulation and law”—which is a much lower bar.

I am not arguing that the strong version of Tiebout (“Local governments will be as efficient as firms in a perfectly competitive environment.”) is true—Tiebout wrote a model and, like most models, his model has a lot of “perfectly spherical frictionless cow” assumptions. But the key point is that we should expect local governments to be more efficient than super-local (i.e., state and national) governments and—in that sense—I think his model has been proven out by the data. An easy example of this is that most people, when they have a choice, eschew “big urban government” and instead choose local government—do you and your friends with means live in D.C. proper or out in the ‘burbs? (Although D.C. may be a bad example, since it is a federal entity. But think of St. Louis vs. Clayton or Creve Coeur, or San Francisco vs. Los Gatos or Palo Alto, or Boston vs. Newton or Brookline, etc.)

In particular, you argue that local government is “far below [the] efficiency frontier” in its three main areas of responsibility, and you later discuss what I would consider a 4th major area of responsibility, local infrastructure:

1. Law enforcement: Do local (i.e., not big urban governments) do poorly on this dimension? I think that is a hard claim to make—sure, Baltimore or the South Side of Chicago are horrible, but those are big urban governments. And it seems to me that law enforcement outside of these areas delivers more or less what consumers want. If you disagree, let me know.

2. Schooling: Again, I agree that big urban school districts were a disaster on Covid. But many small local districts reopened much sooner (at least in Texas). On schooling more generally, big urban school districts are a disaster, but many suburban school districts are comparable to private schools—that is a big reason why people move to the suburbs when they have kids. You argue that “better schooling policies...[are] largely imposed by state governments.” I am not sure about this: Serrano v. Priest was a state court decision that was the beginning of the end of quality CA public schools (see Fischel, 2005), and Robin Hood has been a huge burden on local public schools in Texas (see Hoxby and Kuziemko, 2004). My local district pays 60% of its budget to the state; this is like facing a 150% average tax on school spending! And I would guess its marginal tax rate is even higher. And imagine if a local school district decided that it would not recognize/negotiate with the local AFT—good luck getting the NLRB to agree!

3. Local infrastructure: Actually, local governments do a pretty good job (as far as I can tell) on local roads and the like. It is true that highways (particularly in California!) need congestion pricing and a lot of refurbishment/expansion but that is a state government responsibility.

4. Housing policy: You and I may think local housing policy is really inefficient, but it is pretty clear at this point that local residents don’t agree. On this issue, local governments have been very responsive to the demands of their consumers/shareholders, even if such demands are misguided. (I think consumers’ demand for cigarettes is also misguided, but I don’t think the “cigarette industry is inefficient” because of this.) And Fischel (2005) argues quite convincingly (“the homevoter hypothesis”) that local governments work very hard to maximize local home values.

You could argue that local governments—if truly efficient—would find a way to make the value creation of building more housing work for their residents. But doing this would almost certainly require the state government to work with those local governments. For instance, if the influx of new residents requires us to build a new high school and we face a very high marginal tax rate on any new school spending, it might well not be possible to make a deal that makes local residents better off. And good luck getting the state to build/expand a highway to deal with the increases in traffic.

So in all of these areas, I find the claim that local government is “far below the efficiency frontier” dubious. It is definitely true that we are not at the efficiency frontier and purely private provision would be better in many cases. But local government does significantly better than national/state government, and I think that is the point that Tiebout was trying to make—or at least the point we should take away from Tiebout’s analysis.

Best regards,

John Hatfield