In answer to my questions the other day about Aaron Renn’s idea that sprawl is driving to fiscal crisis for cities like Buffalo, Rolf Pendall from the Urban Land Institute sent along a link to a 2003 report he penned for Brookings: Sprawl Without Growth.
You can download the entire report in addition to reading the summary of the findings. Among the culprits are those I suggested in the post–greater individual incentives to sprawl–and some I didn’t think of (and should have) like high tax differentials between cities and outlying areas.
Pendall also grouched at me for suggesting that there is only direction of causation, arguing that urban systems are complex, adaptive systems–not mechanistic ones.
So what does it mean for systems to be complex and adaptive? Changes in any part of the system can prompt individuals to change their behavior and vice versa–that is, there isn’t one direction of causation, and multiple factors can cause the same change, or the same change can cause different subsidiary changes, and changes can either reinforce other factors of change (as I suspect is the case with fiscal crisis urban outmigration), or there can be multiple, overlapping, and conflicting factors.
I would argue that I didn’t suggest in my original post that there is only one direction of causation, but that’s clearly what he read, so let’s throw down over it. I did suggest, and I still do, that Renn has attributed causation to the wrong factor when suggesting that cities are broke due to sprawl. (Nonetheless, Renn’s probably right in that sprawl is probably reinforcing other factors that are driving central city fiscal crises even if it isn’t causing it per se.)
Since Rolf is one of those people I look up to, the comment struck a cord because the complex, adaptive system argument is everywhere. And it is both a) probably absolutely 100 percent correct and b) an excuse for extremely poor scholarship among urban planners.
As an urban planner in a school dominated by economists, I live in the whipsaw between the complex, adaptive argument and the mechanistic argument.
The complex, adaptive systems argument indeed allows us to see the many many nuances and factors in play.
It as often as not leads us to one descriptive, noninformative, blithering case history after another, where what’s deemed causal by the end comes down to whatever the planning scholar wants. Like “we need more transit, more urban growth boundaries, more of my pet policies because those all clearly worked here.” If, at any point, somebody raises a counter example of how, say, more transit accomplished very little, the response can always be “But these are complex, adaptive systems and you can’t generalize.” Well, you just generalized from your case. “But my case is exemplary.”
If I had a dime for every time an urban planner has given this talk in front of my economist and quantitive political science colleagues to be met with eye rolling and general disdain, I would be looking for a house in Beverly Hills now.
The other side of the whipsaw are the urban economists, who hope to find more generalizable knowledge and, thus, more mechanistic causal relationships.
As often as not, they have carefully controlled for every variable under the sun. They have found the cleverest instrumental variable known to mankind. They have used a marvelous fixed effect that has captured (in a general way) the nuanced differences from one administrative jurisdiction to another. They have thus produced a model that shows, definitively, that time costs matter when people select travel modes. No shit.
And my students wonder why I’m bitchy.
Both planners and economists are going to object to my characterization–and these are two, reductive extremes–but I stick by it. I hear this argument every damn day in some form. The economists use it to try to claim they are better than the planners, the planners use it to argue they are better than the economists.
As usual, I’m a person with a foot in both worlds with divided loyalties.
My preference, though I doubt there’s much merit, is to accept the belief that the world is complex, we can’t model everything (but neither can we describe everything–ask a historian; choices have to made about what is in and out of our memories) and that our knowledge of the phenomenon in any situation is miserably partial. Nonetheless, it’s worth making the effort to try to capture the complexities and exigencies using history on the one hand, and to try to look for the dominant levers to affect change on the other hand (seeking generalizability).
I have no big rationale for believing as I do, other than giving up on the idea that the truth is out there would put me out of job. The endgame of the first extreme is that nothing is truly generalizable; everything is happening for the first time and the only time, as everything depends on context. The end game of the second extreme is dancing on the head of a pin. Both are a lot work for remarkably little payoff in terms of knowledge.
One thought on “Rolf Pendall on upstate sprawl and a commentary on urban social science”
Three minor points with which I suspect you don’t disagree. First, the assumption of complex adaptive systems is a great starting point for simulations and Bayesian approaches, not just qualitative research. Second, quantitative studies with cause-effect models are also usually case studies. (One data set, one time period, limited populations, specific questions asked, etc.) A recent article by Rob Sampson (a quantitative guy) on the limitations (if not outright fallacy) of the “gold standard” of random assignment — in criminology, but more broadly applicable — is a terrific critique that lays this out further. See “Gold Standard Myths: Observations on the Experimental Turn in Quantitative Criminology” by Robert Sampson in Journal of Quantitative Criminology (26, 4), 2010. Third, you love reading history, partly because you can, in fact, generalize from its lessons despite the uniqueness of each set of events.
Comments are closed.