Last week, the Bureau of Economic Analysis released updated figures for gross metropolitan product, and I’ve been playing around with the numbers a bit. One thing that stands out: most metropolitan areas enjoyed substantial output growth between 2001 and 2008, but there were major divergences in the performance of output per capita.
Consider this: the Washington (25%) and Phoenix (29%) metropolitan areas saw similar output growth over the 2001 to 2008 period. But Washington’s output per capita rose during that time by 15.2%, while Phoenix saw output per capita growth of just 2.8%. Houston and Dallas enjoyed output growth over that time frame of 22% and 24%, respectively, while New York and Los Angeles grew by 21% and 20%. But New York and Los Angeles had per capita growth of 18% and 17%, respectively, while Houston and Dallas had per capita growth of 3.6% and 5.1% during that time. Raleigh and Atlanta actually saw declining output per capita from 2001 to 2008, of -3.7% and -6.0% (although to be fair to the hometown, output per capita in the Durham-Chapel Hill metro rose by 26%, so the Triangle as a whole did pretty well).
The metropolitan areas with high levels of growth in output per capita tend to be the older coastal cities (on both coasts), rapidly growing cities in the mountain west, and slow growing cities in the plains states. The metropolitan areas with low levels of growth in output per capita tend to be sunbelt boom towns and rust belt cities (and sunbelt rust belt cities like Saint Louis and Birmingham).
Of the ten largest metropolitan areas in the country we have six that enjoyed above average per capita output growth since 2001 — New York, Los Angeles, Washington, Philadelphia, Boston, and San Francisco — and four that had below average per capita output growth — Chicago, Dallas, Houston, and Atlanta. And while all ten metro areas were above average in terms of per capita output, four of the faster growing metro areas — New York, Washington, Boston, and San Francisco — were also richer than the slower growers in 2001. In other words, the per capita output gap between the rich and the less rich widened over that time.
What does this mean? Well, one obvious potential explanation is the filtering hypothesis I’ve discussed here before. The thing that stands out about Chicago, Dallas, Houston, and Atlanta is their status as metros with very elastic housing supplies.
Housing remained affordable in these places as prices rose considerably in the other large metros. Those rising housing prices would tend to encourage marginal families to move to more affordable metro areas (beginning, one presumes, with those getting the least benefit from being in a high priced city). Similarly, high priced metros would tend to attract new residents with high expected incomes. The expensive cities therefore “select” for very productive workers, while the cheap cities attract workers that were close to the bottom of their old cities’ productivity distributions, but which might be a little above, at, or below the average productivity level of their new cities.
In this explanation, the driving force behind the divergence in per capita output levels is the people. You move the people, you get different metropolitan productivities.
Another potential explanation is that population growth in one kind of city increases productivity, while population growth in another kind of city does not. The key here would seem to be density, as there is a documented relationship between metropolitan productivity and density in the economics literature. The problem with this theory is Chicago, which works for the housing supply theory but not for the density theory.
But one could pursue a hybrid of the two theories by suggesting that it isn’t simply the most productive workers who opt to stay in expensive cities, but the workers whose productivity is most dependent on city-specific externalities. That is, we can imagine two individuals living in New York that are equally productive and equally well compensated, but one of which has a productivity level that falls by a third if he leaves New York, while the other’s is city independent. Prices for homes in New York would have to rise by much more to convince the worker dependent on New York-specific factors to leave.
But we’re also imagining that city-dependence is externality-oriented, which means that when the independent worker leaves New York, he doesn’t reduce the benefit to others of being in New York, but he also doesn’t increase the benefit to others of being in the place to which he moves. Price filtering, in this context, reinforces the externalities present in some cities; it will tend to push metropolitan areas apart, into those dependent on positive spillovers and those that aren’t.
One can then imagine that there is some optimal level of price filtering for cities. Imagine a population of people divided into fixed numbers of externality-oriented and “independent” workers. If New York has a housing supply that is perfectly elastic, then too many “independent” workers will choose to live there, adding to the negative congestion externalities of the place but not contributing to the positive spillovers. This would then reduce the return to being in New York for the externality-oriented workers. In fact, they may self-select into metro areas with tight housing supplies, in order to keep out those who would free ride on the positive externalities.
I’m still thinking through all of this, but that’s what I got out of the latest round of metro GDP numbers.
This article was written by