Kevin Drum asks why kids don’t walk to school anymore:
according to the CDC, only 31% of children ages 5-15 who live within a mile of school walk or bike. That’s down from 90% in 1969.There might be more than one initial impetus—irrational concerns about safety, heavier school backpacks making walking more difficult, busier parents using the commute as quality time, and the like. Once it gets moving, the phenomenon seems vulnerable to a self-reinforcing tipping phenomenon. By not letting your child walk to school because the streets aren’t safe, you take one more child off the sidewalks and incrementally exacerbate the problem of deserted streets.
But I still can’t figure out why. Why do parents ferry their kids around when there’s no reason for it? What’s the motivation?
Like the original Schelling tipping model of racial segregation , this explanation has some very attractive characteristics. It’s parsimonious, self-propelling and grounded in simple, disaggregated individual choices. It’s got all the desiderata of an elegant theory that satisfies the strictures of methodological individualism mentioned recently. It might be right. But there’s still a good chance that, empirically, it’s wrong.
For example, in many parts of Tucson you court death by walking to school, or anywhere else, because the city is built to accommodate cars and not pedestrians. You routinely have to cross 4-lane highways and often there are simply no sidewalks to walk on. The lack of children walking to school might then be explained by the design choices behind the built environment rather than by a nice tipping effect. The other potential explanations mentioned above are also exogenous in this way. But we might find that the appeal of the tipping explanation is so strong that it becomes conventional wisdom without anyone actually studying the problem.
Social Dynamics, a recent volume of studies of tipping and agent-based simulation models of various phenomena showcases some nice work on trying to capture the emergent character of many social phenomena. But lovely as these models are, we know empirically that many phenomena that can be formulated as tipping processes do not, in fact, happen in that way. Neighborhood racial segregation, for instance, has historically been actively enforced and collectively sustained, and is not simply the unpleasant byproduct of innocuous choices. Similarly, social movements that successfully propagate ideas or initiate collective action tend not to rely on contagion but are usually very well organized. (In my experience, although they may not describe the empirical process properly, Schelling-type models are good rhetorical tools for motivating people to admit that there might be a problematic pattern of racial or gender discrimination in their organization. This is because they give you the ability to say “There is this collective problem but it wasn’t caused by any of us making choices that were racist/sexist/whatever.” Very handy.)
To put it (somewhat unfairly) in disciplinary terms, economists are really clever model builders with no tradition of fieldwork, particularly in the more sociological areas that are now becoming popular amongst empirically-minded economists. They’re also strongly predisposed to explanations that can be put in terms of the collective consequences of isolated individual rational choices. Because these kinds of explanations can be really, really appealing, the need to go back and forth between the empirical data and the models is even more pressing than usual.
Update: I just remembered where I was channeling some of this line of thought from: I’d read a review of Social Dynamics by Michael Chwe a while ago, I think in the Journal of Economic Literature.