Revisiting the NYT coverage of Chetty et al. 2013 on social mobility and urban form

So I finally got around to reading the Chetty et al. paper that prompted this writeup from the New York Times about the connection between sprawl and opportunity, which I wrote about here. To say that the original paper doesn’t say anything about sprawl would be an understatement. Go read the original study.

The smarties over at the Sol Price Center for Social Innovation noted the discussion became a sprawl brawl between Paul Krugman and the folks over at Cato. Now, listen, I don’t go agreeing with the people over at Cato often, but Krugman is wrong about Atlanta and Detroit. The original study he references seems to have jack diddly to do with actual urban form. Detroit and Atlanta usually score much differently on sprawl measures, and, if you look at the original map and rankings, Charlotte is right there by Atlanta. While everybody has been hating on Atlanta, Charlotte has, like the top-ranked city for upward mobility, Salt Lake City, been a darling of New Urbanists.

Here’s the actual finding from the white paper: cities in states with more progressive income taxes and greater federal tax expenditures per adjusted gross income have more social mobility. Oooooooooo this is interesting, because Chetty et al. don’t use spending as the measure of tax expenditure. They use the mortgage interest deduction as a federal tax expenditure, which it is. Let’s get more specific: they find positive correlations between local tax expenditures via the Earned Income Tax Credit and the mortgage interest deduction and greater income attainment among their second generation. It’s a clever design question if the model is a mite simple.

I’m trying to wrap my head around what this has to do with ‘sprawl’ in the way that the NYT and subsequent spins had made it, and I’m coming up empty. If anything, the mortgage interest dedication is uniformly hated among anti-sprawl advocates who have always cited it fuel for McMansions.

I’ve questioned this association for a long time, though smart people like Richard Green tell me that it does prompt larger houses. I have argued here that the interest deduction does not itself make houses or lots bigger. Instead, it allows people to spend more on housing. Some people choose to spend that increment on apartments in Manhattan; some people chose to spend it on their McMansion. What the deduction does is allow for more consumption across all preference sets. If you have a million dollar mortgage, your deduction depends on how much you owe, not what you bought. Jonathan Levine’s point, that $1 million condos are less ubiquitous than they should be, matters, in that there are more big houses at $1 million than condos at $1 million, but that problem isn’t resolved by getting rid of the mortgage interest deduction. It IS resolved by removing development restrictions on density and letting markets suss out what people prefer to use their extra bit of housing money on.

The study seems to say that regional housing markets that allow for home ownership relative to income increases intergenerational mobility, and that would fit into the Atlanta and Detroit contexts very well in thinking about how barriers to African American home ownership in cities become, in turn, barriers precisely to the intergenerational social mobility outcomes that are of interest. It may be that there is a connection between sprawl and barriers to home ownership, but that’s not what we have here, at least not yet, and I’d have to a really really good empirical study that sprawl dampens home ownership—theory is not in favor of that argument.

And it’s rather annoying to have people trumpeting this as a ‘sprawl’ story when there are other, important questions that stand before urban form in the authors’ work.

H/T to the wonderful Peter Gordon for sending me the paper.