This scrappy, thoughtful book makes the provocative claim that the causes and consequences of sprawl are less about the excesses of private markets than the excesses of planning.
More to the point, Levine says that unrestrained zoning in favor of modern single-family, suburban housing has so distorted land market decisions that much of the research on the effects of compact development on livability and travel misses the forest for the trees. "But improved understandings of the relationship between land use and travel behavior will not resolve the controversy because travel behavior studies are not designed to shed light on the more fundamental question of why there is so little alternative development to begin with." (p. 23/24).
Suppose Marlon and I study the purported transportation benefits of a feature of the new urbanism, conclude that earlier studies making such claims are deeply flawed, in turn find those benefits are either difficult to verify, case-specific, or potential problems -- and in the end thus question this feature's utility as a dependable source of traffic relief. The problem with this approach, in Levine's view, is that our conclusion is implicitly with respect to an unreliable benchmark -- the status quo -- which itself is overregulated and thus a useless metric.
You have to like the ethos of the story so far. It's consistent with two established policy reform literatures: (1) the theory of the second best, which demonstrates that preexisting distortions change the rules for correcting market failures, and (2) its counterpart in theories of public sector failure, as in Wolf's classic, Markets or Governments (though neither are cited that I noticed). The basic logic is sound. In Levine's terms, what is the default: The status quo, or a less regulated landscape in which developers are more free to test the market for compact developments? (Plus, I can't help but notice the back cover has highly laudatory blurbs from Wachs, Orfield, Calthorpe, Fischel and Downs. One of these guys could be wrong about something, maybe, but not any two.)
Levine's core argument is that compact development is undersupplied. But evidence that this amounts to a problem is not featured until late in the book. (Several early chapters do demonstrate that local land authorities often vex developers and otherwise reduce densities from market levels, and that exclusionary zoning is ... exclusionary.) One example is work indicating that compact developments command a market premium (e.g., Song and Knaap, in the most downloaded Journal of Urban Economics article of 2003).
The book finally presents supportive evidence of its own in chapter 8. That carefully compares the residents of Boston and Atlanta, where the former enjoy more diverse neighborhood/travel environments and thus presumably are less constrained in their travel decisions. Controlling for the mix of preferences, the results indicate that Bostonians are indeed better able to match their housing/neighborhood/travel preferences to their housing choices. Atlantians, with fewer options, would benefit from having more.
We are thus left to conclude that because of credible evidence that many residents of Atlanta would like more compact neighborhoods, that (a) they should be widely permitted (not provided) on that basis and that (b) studies of the transportation effects of neighborhood design are missing the point. Wait, where did that last one come from again?
The theory of the second best doesn't say that you need to correct the preexisting distortion before addressing new ones, though that is preferred, only that you must account for it. Levine's story does not appear sufficiently nuanced to recognize this, which gives his blanket dismissal of analyses of the transportation effects of urban design the appearance of overstatement.
The book's discussion of "self-selection" bias seems off. Levine is right that we often try to separate out sorting when attempting to measure the pure effect of a treatment, such as the siting of a transit station. (To measure the effect of a new station on the ridership of the average neighborhood resident, we want to control for those who moved there because they are more inclined than the average person to take trains.) He claims, however, that the self-selection of transit users to move near the station is in fact evidence of the treatment in some cases, not a bias to be eliminated. Yes and no. We still need to separate out this sorting in order to measure the treatment effect on the average resident.
As much as I thought I learned from this book, I can't keep myself from believing that when a developer makes traffic improvement claims for their subdivision plan, and those claims -- whether based on wishful thinking, a sincere faith in the general applicability of the simulation results of Walter Kulash, or marketing savvy -- are shown to be false or simply cannot be verified, that this should raise questions about whether the project will make traffic better or worse. Even if I accept that higher densities and more walkable neighborhoods would lessen the frustration of numerous Atlanta residents, and a good many others besides, I don't follow the implication that the possibility of some development patterns worsening traffic is neither here nor there. Apparently I will have to reread Zoned Out.