Monday, December 12, 2005
This is the second of two posts featuring a discussion between Michael Lewyn and Kurt Paulsen on sprawl. The first post focused on the definition of sprawl. This post talks about the larger issues of what causes sprawl and what, if anything, we can do about it. I had originally planned on breaking this down into a series of smaller posts, but because all of these issues are interrelated, I've decided to put everything in one long post. As an aside, you can find links to four of Lewyn's articles on sprawl here.
As before, Lewyn and Paulsen wrote their entries without seeing what the other had to say. Feel free to join the discussion in the comments, but please note that comments are held for approval, so there will be some delay in posting.
What creates sprawl and can we do anything about it?
Part of what makes sprawl a difficult public policy problem is that both kinds of sprawl discussed above ("how we develop" and "where we develop" is not just the result of one misguided policy (or of "the market", to use a common counterargument). Much like other major public policy challenges, it is the result of a whole bunch of policies, and so you need to work a lot of levers to get it right: kind of like crime in the 2nd half of the 20th century, which seemed unmanageable at first but was brought partially under control because a variety of actors made a variety of changes.
1. Where we develop
Causes of suburban development include:
a) highways that essentially subsidize (or at least encourage) long-distance commuting by making it faster and easier
b) zoning and parking regulations that, by reducing density, constrict supply of housing in older, already-developed areas and thus forces people to move further out in search of newer or affordable housing
c) FHA mortgage insurance (which generally supported suburban development only from the 1930s to 1960s, and thus helped send cities into a "death spiral" of decline which continued long after FHA policies changed)
d) public housing policies that packed public housing in cities for the poor, thus making cities less desirable places by concentrating poverty and related social ills in cities (This problem has at least partially been fixed by the HOPE VI program and Sec. 8 vouchers).
e) suburban zoning that kept out poor, thus keeping poor locked up in cities and making cities poorer and more undesirable
f) federal policies towards urban transit- both lack of support and unfunded mandates that canceled out federal support
g) state school assignment laws that force children to go to school in municipality in which they live- which means that city schools more socially/racially/economically diverse (and thus less attractive to middle-class parents) than suburban schools
h) school desegregation case law that exacerbated the effect of school assignment law, by providing that city schools had to desegregate (thus making city schools less attractive to white parents) but not suburban schools
i) artificial division of regions into city and suburb, thus causing city to be stuck with weaker tax base, poverty-packed schools, etc.
2. How we develop (or, given that people are going to move to suburbs why
are those suburbs so sprawling?)
a) zoning laws that keep housing away from jobs and offices, thus making it harder to walk to work or shops
b) zoning laws that artificially reduce density, thus keeping residential zones far from business zones
c) setback and minimum parking requirements that effectively require businesses and apartment buildings to be surrounded by parking (perhaps more parking than market would support, though this is empirical issue that is not fully researched), thus reducing residential density and making walking less comfortable
d) wide streets that make walking uncomfortable and reduce density
e) long blocks that reduce pedestrians' ability to cross street
f) minimum parking requirements make parking free, thus acting as fertility drugs for cars
g) cul-de-sac street design (often required by cities or FHA, though not always) which makes walking within residential neighborhoods difficult
If you are in a hole, stop digging. By and large, the solutions to the abovementioned policies are to stop the policies. (Some of the policies, of course, have been stopped long ago- which may explain why some central cities are beginning to recover from the ravages of sprawl). But here's a list:
1. Where we develop
a) Highway spending should be examined for its "sprawl impact." That means no more city-to-suburb highways (and perhaps even tear down a few that we have), no suburb-to-suburb highways in undeveloped or barely-developed parts of suburbia.
b) Transit spending doesn't really reduce this type of sprawl, but does mitigate its impact upon people without cars. So transit spending is often desirable (subject to cost-benefit constraints), and removing unfunded mandates on transit is desirable.
c) Housing Opportunity Zones- Zoning should be radically restricted, if not abolished, in central cities. Anyone should be able to build anything, anywhere, as long as it is (1) residential and (2) in the regional core (defined broadly but not too broadly- certainly anywhere in the city if the city has less than 100 square miles or so).
d) Some kind of reform of exclusionary zoning (though the line between "reducing exclusionary zoning" and "unfettered suburban development" is a blurry one. I think it is at least arguable that abolishing zoning and similar restrictions everywhere would do more good than harm by encouraging infill- but the question is an empricial one and there is no way of knowing for sure that this is the case).
e) Possibly urban growth boundaries in some regions- depends on effect on housing costs and on whether combined with liberalization of central city land use regulation. Some UGBs appear to have had a positive effect on sprawl (most notably Portland), others less so. Obviously there is a tension between (c), (d) and (e), reflecting the broader tension between discouraging suburban development and encouraging urban development. I prefer the latter, but have put the former of my menu of options because (i) not everyone's instincts are as libertarian as mine and (ii) the former might occasionally do more good than harm.
f) Break the link between residence and schooling in central cities, either through some kind of voucher system accessible to middle-class families OR through radically expanded use of "exam schools" (schools which, by screening out low achievers, draw middle-class families back into schools even in otherwise depressed school systems- City Honors in Buffalo, Boston Latin in Boston, etc.). I think equalizing school finance is of no value in sprawl control; plenty of central-city systems spend as much or more than
suburban school systems. (School finance equalization may be worth doing for reasons unrelated to sprawl- e.g., if it marginally improves performance of low achievers. But the low achievers helped by such a system are typically from low-income households and thus are a "captive audience" who live in the city because they cannot afford to live elsewhere).
g) Regional government or tax base sharing
h) Obviously, improved city government has a marginal positive effect- but we won't necessarily agree on what "improvement" means. Also, I think we should admit that effect of city government is MARGINAL. For example, NYC gained population (9 percent increase) in 1990s when Giuliani "cleaned up the city"- but also gained population in 1980s (3 percent increase) when city less well run. Even a Giuliani means a difference between a status quo city and a modestly improving city, NOT the difference between a status quo City and a St. Louis or a Detroit. Mayor Giuliani trapped in Detroit would have been like Johnny Unitas playing football with 10 high-school students on his team.
2. How we develop-
a) Deregulate, deregulate, deregulate. Abolish all regulations discussed above.
b) Cities could experiment with zoning that mandates more pedestrian-friendly development (e.g. maximum setbacks instead of minimum, maximum parking requirements instead of minimum). But command-and-control regulatory approaches run risk of increasing cost of doing business and driving out capital, and so municipalities need to be pretty careful about this sort of thing.
E. Municipal control vs. regional control
Because sprawl is caused by a large number of factors, some of them can be cured or mitigated without reforms related to land use or regionalism, others need regional support, and in still others, regional or state or federal power has done more harm than good (e.g. school desegregation). Obviously, the more of the above problems you work on, the less harm caused by sprawl. But this is not an all-or-nothing problem that depends on one key reform to be "solved" or "not solved."
Under standard urban economic models, the cost advantages of automobile travel permit lower-density development further from the city center. In this sense, automobiles "cause" suburban development. But does auto-dependence cause "excessive" suburbanization? Not necessarily. Automobile dependence may be partially caused by low-density, single-use development patterns, but it is just as likely that low-density single-use development patterns may result from the cost advantages of automobile ownership. I don't think there is anything inherently wrong about dependence on automobiles, given consumer preferences and the tremendous benefits of automobiles – and, subject to the conditions that automobile travel pay its true social cost (see below) and that those excluded from automobile ownership have equally viable and equally available transportation options.
Within the field of urban economics, the canonical model of urban spatial structure is the "monocentric" or Alonso-Mills-Muth (AMM) model. To oversimplify: this model describes a situation where employment is concentrated in a city center (Central Business District or CBD). Land values (and hence, the density or intensity of land use) decline in a predictable fashion with distance from the CBD. Different land use types (offices, industries, commercial, residential, etc.) each have "bid-rent" curves (how much rent they are willing to pay for proximity) resulting from the nature of production or their preferences. Different land uses have different demand for land vis a vis distance to the CBD. This model is illustrated conceptually below.
The Alonso-Mills-Muth (AMM) model holds up reasonably well in empirical studies, even with its unrealistic assumption of centralized employment and a single-centered (monocentric) metropolitan region.
There are two features of the real world to examine in defining "sprawl" by reference to this model. The first feature is employment decentralization, sometimes disparagingly called "job sprawl." Anyone who drives along highways notices that office-parks, warehouses, and manufacturing facilities have moved out to the suburbs and exurbs. Two recent papers by Ed Glaeser document the declining percentage of jobs within metropolitan regions within close proximity to the CBD. What caused this employment decentralization? (or, technically, a "flattening" of the office and manufacturing bid-rent curve) Glaeser demonstrates that declining transportation costs are a prime determinant. Transportation cost advantages of centralized employment (e.g. access to ports or railroads) have been reduced by trucks and highways. Businesses leave downtowns because the high land rents in the CBD are no longer cost-effective for them. Further, one advantage of centralized employment with residential centralization was the availability of a skilled and abundant work-force in easy commuting distance. If firms pay a worker her marginal productivity plus some compensation for the time and financial costs of commuting to the job, centralized employment and residences can reduce the wage bill of employers. Employment decentralization, however, has followed residential decentralization.
Generally, firm location will be based on proximity to its input markets including labor, land and raw materials and proximity to output markets. Declining transportation costs to move both goods and people should lead naturally to employment decentralization within metropolitan regions. Only those businesses for whom proximity to agglomerations of other businesses downtown (e.g. finance, investment, business services) provide a cost advantage are willing to pay the higher land rents downtown.
Of course, this is not the only story. Jobs in manufacturing and offices have not only left central-cities, but have also fled to southern states (because of anti-union laws and lax environmental regulations) and overseas.
The second feature of the monocentric model to highlight is residential decentralization. (Perhaps, if we are looking for the most neutral description of what most people mean by sprawl it would be "residential decentralization.") The figure below indicates a standard looking residential density function. In this figure, the density of residential development declines with distance to the city center.
Within the model, higher density development occurs closer to the CBD because of higher land rents, while lower density development occurs further away. Higher land rents lead to higher densities, while lower land rents lead to lower densities. If one believes that all lower-density suburban development is inherently bad, the real solution would be to increase land rents!
Households chose residential locations based on trade-offs between accessibility, amenities, and house/land size. If one lives further away from the city, one can afford a bigger house and more yard, but a longer commute. If one lives closer, one might be closer to employment and cultural amenities, but the unit-cost of housing is more expensive and houses are smaller. Households make these trade-offs and the choices of commuting time based on demographics, preferences and income.
All other things being equal, the model predicts that increases in population in a metropolitan region will cause an expansion of metropolitan size (total area of developed land). New entrants into a city can be accommodated by either growing up (increased densities) or growing out (new fringe development). All other things being equal, the model also predicts that increases in household income will lead to a "flattening" of the density gradient and increased suburban development at the fringe, because the demand for land and house is income elastic. As income increases, people demand more. This is one of the first questions I ask my land-use planning students when they complain about sprawl: "What is the income elasticity of demand for land and house size?"
All other things being equal, the model also predicts that decreases in commuting costs will expand a metropolitan area's size, as people can afford to live further out and consume more house and land for the same level of utility. Lastly, all other things being equal, increased land prices on the metropolitan fringe – from agricultural rents -- should restrict outward expansion of metropolitan areas.
Thus, the predictions of the model (which have held up remarkably well in empirical research) are: increases in population, increases in income, decreases in commuting costs, and decreases in agricultural rents should lead to employment and residential decentralization and lower-density development at the metropolitan fringe.
It is in this context that one can define suburbanization as "natural." As pointed out in Rybczynski's review of Bruegmann's book, suburbanization is a natural feature of metropolitan regions almost everywhere and at almost every time. Increases in population, or increases in income, or declines in commuting costs lead to greater residential decentralization at lower densities. It is as true of Paris as it is of Atlanta. The issue, of course, is whether this suburbanization is "excessive." Suburbanization, in and of itself, is not "sprawl" nor is it in and of itself necessarily evil.
The United States in the post-war period has sustained increases in population, increases in income and declines in commuting costs? Voila: metropolitan spatial expansion at lower densities. Quod erat demonstrandum.
(One surprising result when I present this lecture to my students has been recent number of students who will thus argue that population control and immigration restrictions are necessary to stop "sprawl". Interestingly, if one Googles the term "sprawl", the first website listed is www.sprawlcity.org which makes the same claim. I don't have time here to refute this argument, but I do wish to repudiate it with all the vehemence with which I am capable.)
Aha -- not so fast. Within the urban economics paradigm, the implicit assumption is that this monocentric model (or its poly-centric cousins) represents something like the "free market." However, within urban transportation and urban land use, there really is no such thing as a free market. Governments intervene in and regulate urban transportation and land markets by necessity. Imagining an unregulated, free market in urban land and transportation is merely an academic exercise.
Let's return to the transportation cost argument. What caused the decline in transportation costs, particularly for automobiles and trucks? While some reduction might be attributable to better automobile technology, most of it is attributable to government investment in highways. While it is true to say that suburbanization has occurred in every country, suburbanization is more auto-oriented in America because of more investment in infrastructure to support private automobiles relative to public transportation. The outward expansion of urban areas facilitated by declining transportation costs are due to government investment in highway infrastructure. Thus, while some might conclude that "highways cause sprawl," it is more precise to say that "highways facilitate suburbanization."
Within a "market" paradigm, urban development and urban transportation are subject to a wide range of "market failures." Jan Brueckner points out 3 major market failures which may lead to "excessive" spatial expansion of metropolitan areas. First, there is a failure to account for the social value of open space. Open space, natural lands, agricultural vistas, scenic beauty, protected wildlife habitat, etc. all have a value to society and future generations. Because this value is not usually expressed in a private land prices, development at the urban fringe appears artificially cheap. The solution to this is simple: a land development tax, the proceeds of which are used for land preservation, parks and open space and the like. This is somewhat happening around the country: hundreds of municipalities, counties and many states have seen overwhelming voter approval to raise taxes in order to purchase open space. Of course, as any land use planner knows: why actually pay to preserve land when you can simply down-zone it?!
The second market failure is the failure to account for the infrastructure costs of new development. To the extent that new development fails to "pay its own way" in community infrastructure (water, sewer, roads, schools, etc), it appears artificially cheap. Again, a solution seems to have been recognized in many places around the country: development impact fees (or concurrency requirements or adequate public facilities requirements).
The third market failure is the failure to account for the social costs of traffic congestion. I would expand on Brueckner's formulation to include the failure to account for the social costs of automobile travel generally. Let's face it: people drive too much, and they drive too much at the wrong times of the day on the wrong roads. People do this because the marginal cost is below the true social cost. The true social cost of driving includes the infrastructure costs of roads (including maintenance, repairs and law enforcement), air pollution, and congestion. Congestion occurs because each driver imposes a minimal congestion cost on everybody else who is driving on the same road at the same time. Every driver knows the problem with congestion: too many other cars on the road when I want to get to work!
The solution to excessive driving is obvious to any economist, yet not so obvious to any politician: user charges, gas taxes and congestion tolls. User charges per mile of road should pay for acquisition of rights-of-way, construction, maintenance, repair, enforcement, etc. The charge should be per-mile, but could be proxied in a gasoline tax. Proper gasoline taxes should encourage less driving and more fuel-efficient, lower-polluting cars. Congestion tolls should vary with the time of day and the roads chosen. Every day I drive, an electronic sensor takes money from my credit card and sends it to the Pennsylvania Turnpike. Adjusting electronic tolls for congestion is technologically simple, yet politically unrealistic.
If the social cost of automobile travel is not fully borne by each traveler, it is artificially cheap and people drive too much. And, because transportation costs are a significant component of metropolitan spatial expansion, metropolitan expansion must therefore be excessive. The spatial expansion of developed land is excessive because open space conversion is artificially cheap, infrastructure is inadequately priced, and private automobile travel is artificially cheap. If we had the political capacity to address these issues, we might make some headway in curbing the excesses of spatial expansion. If our main concern in "sprawl" is land consumptive low-density, auto-dependent development, we should certainly stop subsidizing "sprawl." Ok, where is the courageous politician to propose development taxes, higher gasoline taxes and elimination of the mortgage interest deduction?
There is far more to sprawl than market failures and the natural expansion of metropolitan areas. Cities have been losing populations also due to irrational fears. Fear of crime and fear of racial integration have certainly led many whites to flee central cities and establish bastions of exclusion and privilege in the suburbs. Let's not kid ourselves that "sprawl" in the United States is natural. White flight and racism have played significant roles in sprawl, to our great shame.
There are also policy and planning failures. One policy frequently mentioned as contributing to sprawl is the federal tax treatment of mortgage interest paid. Richard Voith in a series of papers has demonstrated that the mortgage interest tax deduction reduces the operating costs of homeownership by approximately 15 percent per year. Because demand for land and house is income elastic, this subsidy allows people to consume more housing than they otherwise would. This demand stimulus should make metropolitan spatial areas excessive.
However, the real planning and policy failure causing sprawl is zoning. Consider the simple model posited above. The model ignores politics. Yet, every metropolitan region in this country contains large number of general purpose local governments with land use regulation powers. Land use regulation (of which zoning is the 900 pound gorilla) is almost completely a local affair. The fragmentation of metropolitan regions into hundreds of autonomous local governments with zoning power surely complicates the story above.
Consider the figure below. Again, we see the familiar declining density gradient. But, in this case, I have illustrated the location of a hypothetical "Upper Snobville Township." In the pure market model above, the density within this municipality might be 8 housing units to the acre in the portion of the town closer to the CBD, and 6 units to the acre in further out portions of the township.
In this illustration, the dark line represents the "market" density, and the dashed line represents the maximum permissible zoning under the Upper Snobville Township ordinance. Saying that "zoning causes sprawl" is like saying that overeating causes obesity. It is true, but not particularly interesting. The real question is what "causes" zoning? How does zoning arise? Well, the current residents of Upper Snobville, the homeowners who are voters (the "Home-voters" to use William Fischel's term) certainly do NOT want 8 units to the acre in their community. Why? First, by artificially reducing development densities, they create a "scarcity effect" within their community. Second, by reducing density, they create an "amenity effect." Third, by increasing minimum lot sizes, they make sure that their neighbors are of a higher income class. Fourth, by reducing the total number of housing units, they reduce the total numbers of school kids, thereby reducing fiscal burdens. The "Home-voters" who control municipal zoning thereby use zoning to protect their property values and confer upon themselves a nice capital gain.
Upper Snobville's neighbors are no fools: their home-voters utilize the zoning powers as well. On and on it goes. The result? To paraphrase William Fischel: the American way of zoning causes metropolitan areas to be too spread out. Each municipality zones for less density than would occur under the "market" and the result pushes housing demand further away from the center at lower densities.
This is exclusionary zoning. It need not been merely aimed at keeping out the poor and racial minorities (although it often is). I consider it exclusionary zoning when a municipality significantly reduces development densities below what might occur within a market framework. The net effect of exclusionary zoning is excessive spatial expansion at low densities.
The other effect of exclusionary zoning is the observed pattern of residential segregation by race and class. Because where one lives determines access to a wide range of public goods and access to jobs, the result is an unequal – and unjust – geography of opportunity. This, to me, is the great evil of the current pattern of low-density, exclusionary zoning and the real reason why "sprawl" is bad.
What is the solution to exclusionary zoning, residential segregation and its attendant lower-density, fragmented land use pattern? It depends on what one sees as the motivation behind exclusionary zoning. William Fischel has written extensively on this issue, and has proposed something like "House Value" insurance. The idea is simple: if home-voters (for whom the value of their home is their prime asset) no longer feared reductions in property values associated with more dense development, they would have less motivation for exclusionary zoning.
I am less optimistic about this solution. I prefer to think of it like this: if there is too much of a bad thing, we can tax it to get less of it and use the revenues to mitigate the negative effects. I would propose something that would function like an "exclusion tax" or an "inclusion bonus." Municipalities which did not permit adequate densities would have to pay. Or, municipalities would be incentivized to allow higher densities in that they would receive compensation from the state. I know that, in practice, implementing an exclusion tax would be politically difficult. One partial proxy for my idea has been New Jersey's system of mandated fair-share affordable housing where municipalities can trade away half of their fair-share obligations for payments to other municipalities. In practice, suburban municipalities "sell" their affordable housing obligations to urban municipalities. If we can create markets in pollution tradable permits, why not exclusion tradable permits? Municipalities would be allocated their fair share of both affordable housing and overall development density. They could either accommodate such allocations in their zoning, or they could "sell" their obligations to other municipalities. Presumably, urban municipalities, which already have density and affordable housing, would receive payments to obviate suburbs of these requirements. Properly structured, this market could elicit a suburban jurisdictions' true willingness-to-pay-to-exclude. Suburban jurisdictions would either have to pay to exclude or accept higher density development.
I am not naïve. I know that forcing people to live in integrated (both racially and socio-economically) neighborhoods would be resisted. However, as a practical matter we need to structure the entire system of public finance so as to 1) NOT subsidize low-density, exclusionary suburban development, and 2) incentivize higher-density, mixed-use, integrated development. I realize this might never happen, but one can dream!
As a more practical solution to "sprawl," I am in sympathy with proposals by William Fischel and Jonathan Levine who have argued for what I call "market solutions to planning failures." This would involve restoring to owners of undeveloped lands their development rights: Let developers and real-estate markets build the higher-density, mixed use developments they can profitably build without the arbitrary interference of exclusionary-minded suburban zoning! What could be more American and more "pro-property rights" than that!!
Some critics of urban planning would have people believe that the natural market outcome is low-density, single-use, auto-dependent development. After all, isn’t that what everyone really wants? Really? How about putting that to the test by letting developers build higher-density, mixed use developments (which they would) and see if the market will bear them out.
We will never stop metropolitan spatial expansion. Americans have a love for single-family, detached, owner-occupied housing and for automobiles. We also claim to like fragmented local governments and protections of private property.
People should be free to live wherever they want and be free to live according to their preferences for the type and quality of life they consider worthwhile. However, this freedom MUST be consistent with paying their true social cost for those choices and consistent with maximizing the ability of everyone else to live according to their preferences for a high quality life. To me, the issues involved in the sprawl debate are primarily about justice.
Let me conclude and summarize with some of the questions posed for this discussion. What is sprawl? "Sprawl" is excessive, exclusionary suburbanization. Is "sprawl" bad? So defined, yes. Why? Because it is overly land consumptive and therefore environmentally destructive, and because it unjustly creates segregation by race and class and an unequal geography of opportunity. What causes sprawl? Among others, exclusionary zoning. Can we do anything about it? Yes! Are we likely to? No.
[Comments are held for approval, so there will be some delay in posting]