Bryan Caplan  

Absurdities of the Tiebout Model

Research Incentives: Milton Fr... WSJ's "Roll Your Own" Deficit ...
Economists across the political spectrum embrace the realism of the Tiebout model.  The model's intuition: At the level of local government, there are many consumers (i.e. residents and businesses), many suppliers (i.e. localities), and low switching costs - all the key ingredients of perfect competition. 

The upshot: We don't need to worry about the efficiency of local government policy.  The quasi-free market will take care of things.  To retain population and business, local governments have to offer a competitive package of taxes and benefits.  Any government that implements policies that fail a cost/benefit test will by definition be unable to pass the market test.  At the local level, therefore, whatever is, is efficient.

To repeat, the Tiebout model enjoys bipartisan support.  Liberal economists like the Tiebout model because it exempts at least one level of government from criticism.  Conservative economists like the Tiebout model because it affirms the value of competition and decentralization.  On reflection, however, the Tiebout model has big absurd implications.

Absurdity #1: There will be no redistribution at the local level. 

If a locality even slightly overcharges any segment of the population (the rich, the childless, business, whoever) for the services it receives, that segment will flee.  Final resources redistributed: zero.  Defenders of the model often actually claim that this implication is true, but there's an ironclad counter-example: public schools.  Even today, public schools are heavily funded by property taxes - and the rate has nothing to do with the taxpayer's number of school-age children.

In a Tiebout world, there would be immediate blowback.  Childless residents would flee to their own separate, child-free, school-free, ultra-low-tax district.  Residents with one child wouldn't want to join them, but they wouldn't want to subsidize their more fertile neighbors, either.  They'd have their own separate one-child, small-school, low-tax-district.  And so on. 

On further reflection, the Tiebout model also predicts far lower taxes on businesses than on residents.  Unless there's slavery, business is inherently "childless."  Any locality that taxed business property to fund public schools would find itself bereft of business.  Business would then relocate a foot from the residential locality's border - and pay only the taxes necessary to pave the roads, patrol the streets, and put out the fires.  To forge a mixed-use district, local government have to charge businesses lower tax rates than flesh-and-blood families.

The one exception to the "no Tiebout redistribution" rule is redistribution that corrects local externalities.  Tiebout governments might tax emissions to clean the air or impose congestion tolls to clear the roads.  The catch, however, is that everyone who continues to reside in the locality must selfishly prefer the redistribution's indirect benefits to its direct costs.

In a Tiebout world, any social insurance program would have to act like a for-profit insurance company.  Rates would match risks, pure and simple.  This means that the poor would probably pay not just higher tax rates for welfare or unemployment insurance than Bill Gates; each poor person would probably pay more dollars of taxes for these programs than Bill Gates.  After all, what is the probability that Gates will ever go on welfare?  One-in-a-billion?

Absurdity #2: There will be no waste at the local level. 

In the Tiebout model, local governments only supply goods if they have a cost advantage over the private sector.   If a locality taxes a resident $1000 to buy the resident goods that the market could supply at $900, the resident will move away - and a competing locality will eagerly accept his patronage.  The only way to retain people and money, therefore, is to deliver the goods at the lowest possible cost.  If the private sector does the job best, local governments will embrace laissez-faire.

By this argument, if the private sector could supply education more cheaply than the public sector, there would be no public education.  "Unfair to the poor"?  That's not Tiebout talk.

Yet the NEA need not despair.  The mere fact that school vouchers haven't been adopted proves that public schools work better than any alternative - bloated budgets and illiterate students notwithstanding.  The same goes for zoning, local licensing, massive park lands that almost no one uses, etc.  Everything local governments actually do has to pass that quasi-market test, so who are economists to second-guess the result?

Coming soon: The Tiebout model is wrong in fact, but how can it be wrong in theory?

P.S. Comments somehow got turned off.  Now they're back on.

Update: David Friedman emailed me the following comment:
What's wrong with it is that land can't move. If the local government engages in exploitive taxation, pocketing the money, people move out. As they move out, land values fall. They stop moving when the drop in land value just balances the cost of the exploitive taxation.

The implication, if the rulers are smart, is that they will produce an efficient bundle of services, but tax away the land value and pocket it.
I replied:
Also true.  I even published this exact model:

But I think there's a deeper problem which I'll soon post.

COMMENTS (16 to date)
Yancey Ward writes:

I am thinking of the increasing involvement of both state and federal governments in financing a lot of what passes as local governance.

Arthur writes:

Cities does not differ only in provision of public goods. Probably the problem is in his assumption 6 and/or 7.

mobile writes:

This sounds like a straw man on par with "most physicists embrace the realism of the frictionless surface model". In practice there are plenty of "frictions" working against perfect competition -- the high "switching costs" of moving to another locality, endowments (beaches, existing infrastructure) that cannot easily be moved to another locality that would manage it better, minimum size constraints (if a locality needs to have, say, 1000 households, it must cater to some average set of preferences which may deviate from some individual preference) -- that allow localities to extract rents from their residents and businesses.

Hadur writes:

There are a limited number of local governments, and it is difficult to create new ones. In most states, the legislature would have to create a new municipality or town or county or whatever.

JFA writes:

There is also the problem of bundling. Localities don't just offer tax codes; they don't just offer public schooling; there are a variety of services that people want, and people can't get those things if they live in single-service providing localities.

wd40 writes:

Many of Bryan's clever insights can also be applied to ordinary markets. There are no sedans that seat three in the front row. This is an implicit redistribution from large families to smaller families. When someone has specialized skills, such as those associated with being a university professor, that person does not get the exact combination of job and location attributes that the person would like. And even in some private universities, administrators get paid more and the faculty get paid less than they would if the faculty could start a similar school across the street. None of these examples are really arguments against the market, and, for similar reasons, Bryan's examples do not undermine the Tiebout hypothesis.

Furthermore, city governments typically do not undertake provision of those goods and services that can be provided more cheaply in the private sector. That is why we do not see city governments producing or selling new automobiles, clothing or computers. So Tiebout explains a lot, even if at the margins it does not explain some things such as education that Bryan believes would be more efficiently served by the private sector.

As far as redistribution goes, a great deal of it has been shifted to higher levels. States have taken on a heavier burden of supporting local schools and the federal government has taken on much of the burden of welfare, more generally.

Finally, if rich people prefer to live in places where the poor have subsidized health care, then such subsidization is efficient even if a private unsubsidized health insurance plan would not be viable.

Alexei Sadeski writes:

Am under the impression that property tax rates are set at the state level, not local. Disbursement of some or all of the funds may be determined at the local level, yes, but rates are not.

Jeff H writes:

I don't think this passes the Turing test. You've put up a straw man argument. The comments by mobile, Hadur, and JFA, seem to show that your assumptions don't match the theory as used by most economists other than those holding to the original Tiebout paper. (Admittedly, it's been a while since I've read that.)

The biggest things urban economists tend to talk about is bundling and non-economic factors. Decisions aren't made in a vacuum. Where do you live if you can only get the job you want in community A, but schools are better in community B, but zoning is better in community C, but family lives in community D, taxes are ideal in community E, and so on? You compromise on some issues.

As a result there are always local level issues that people don't agree on 100% but can live with. Or they work to make it the communtity they want.

You need to view the model in that light otherwise you're not fairly representing the position.

MatthewH writes:

Tiebout doesn't predict either of these outcomes. It predicts that the amounts will be efficient, relative to the preferences of the residents.

Residents don't particularly care for redistribution, but they do want a little, so they have limited redistribution. Shocker, Peterson argues this in City Limits.

Parents want good schools because good schools attract rich parents who buy houses, making everyone in the neighborhood richer -thus everyone supports schools. Fischel's Home Voter Hypothesis

And you can have plenty of waste provided residents are willing to put up with it in return for really nice ammenities, a point supported by both Joel Kotkin's The Next Hundred Million and Ed Glaeser's Triumph of the City.

And if I can briefly flog my own dissertation, given the same starting population and wealth, something I found, but hadn't expected to, was that MSAs have more cities even if the population and wealth are the same as non-MSA counties. I'm exploring that now, but my lead theory is that Tiebout was right, and the ability of the MSA to break up bundling makes them more efficient -allowing for smaller and more specialized jurisdictions.

egd writes:

Alexei Sadeski writes:

Am under the impression that property tax rates are set at the state level, not local. Disbursement of some or all of the funds may be determined at the local level, yes, but rates are not.

At least in the US you have it backwards. Property tax rates are set locally (I pay nearly twice the property taxes my neighbors across the city line do) while disbursement is mostly mandated by the state - a large portion of property taxes go to schools.

Chris H writes:

If frictions are keeping people from moving from less to more efficient governments then the size of local governments should matter. We should see more redistribution in cities and counties with larger land areas than smaller ones. Population wouldn't matter much because if a high population area had inefficient government, but it was only a short trip to get their from another more efficient jurisdiction, the friction losses will be relatively small. Does anyone know of any research trying to correlate the land area a government has jurisdiction over and the efficiency of that government in avoiding redistributive actions?

Steve Sailer writes:

Government in Chicago is traditionally more expensive and crookeder because Chicago, going back to the construction of a canal linking the Great Lakes to the Mississippi watershed in the first half of the 19th Century, has been a chokepoint in the national transportation system.

OneEyedMan writes:

The tendency of retired families to cluster together where they are a voting majority is a pretty nice example of Tiebout in action. At best, this says that Tiebout doesn't explain everything.

Immobile factors go beyond land or even transportation choke-points. The agglomeration externalities of cities limit are a scarce and immobile factor that limits competition.

A big part of the story why Tiebout is imperfect is the big fixed costs of moving and incentive problems among leaders. There is very little return to leaders of running the perfect city but there are many costs from angry and loud constituencies. If we can barely control CEOs with 1-5% of a firm's upside why expect better from a mayor getting none.

Ken B writes:

@Alexei Sadeski:
Where I live the local mill rate (as it is called) is set at the county level.

Slocum writes:

'mobile' is right -- switching costs are not low at all AND there's a coordination problem besides. You can move to the boonies for lower taxes, but until businesses move as well, you'll be trading taxes for transportation costs that may be greater than the tax savings. If businesses are the first movers, then they impose those transport costs on their employees and customers -- can they lower prices or raise salaries by enough to compensate?

Of course, instead of moving within a region, you could relocate entirely, which means much higher switching costs, and you have to find a new region where the same dynamic isn't in play.

But it's not the case that people and business don't migrate out of high-tax, high-cost, high-rent-seeking localities to lower-cost neighboring cities -- obviously they do. But the switching costs mean that it's necessarily a slow process. So two thirds of the population of Detroit has now sprawled out into the rest of the metro region, but this process has taken more than 50 years (and isn't finished yet).

kashof writes:

I don't know much about the Tiebout model, but I did work for a very illuminating year a while back doing economic development consulting.
Two basic rules:
1)Politicians want to stay in power
2)Voters don't want to pay taxes

I think the ideal locality for a politician is no residents except for the politician and his/her immediate family, lots of businesses that pay taxes to his/her government. There's a city in CA like this.

Result, politicians are constantly courting businesses to move to their locality, ie the ubiquitous tax breaks. I imagine that large companies are much more mobile because their customer base is much more spread out. They get the tax incentives. Smaller businesses that can't move: dry cleaners, restaurants, doctors, etc pay higher taxes.

Another result: Political homogeneity. The people who don't like the policies move to a nearby locality with people of the same outlook. This is no skin of the politician's back because it increases their ability to stay in power.

Visitors get hosed. Obscene taxes on hotels, rental cars (especially if you rent at the airport), etc.

Anyway if competition was so perfect how did Detroit get hollowed out and the deserts of NV and AZ filled with people.

Comments for this entry have been closed
Return to top