Why Do Cities Matter? Local Growth and Aggregate Growth

This is a working paper by Chang-Tai Hsieh and Enrico Moretti that examines how growth in cities relates to the growth of the nation writ large. The upshot is that, according to their model, if workers were allowed to move more freely to high productivity cities, such as New York and San Francisco, the US could have a substantially higher level of GDP. Let’s walk through the argument.

To start, we can safely assume that there is variation in the productivity of individuals, which is to say we have differing marginal products of labor (MPL). But this productivity does not exist in a vacuum; ultimately a worker’s output also depends on the environment into which they are inputting labor. A mathematics PhD may have a very high potential MPL, but if they are inputting that human capital into a retail cashier production function, there won’t be much difference in output from someone with only a high school diploma.

The second part is that there is geographic, or spatial, variation in productivity as well. It’s well documented that high density urban areas have far higher productivity than rural areas. Now, this could be due to sorting: high productivity workers tend to move to cities so of course cities have higher average productivity. But there is research that suggests density itself leads to higher productivity, as explained in this paper, through phenomenon like knowledge spillovers.

So we have variation in productivity across individuals and across space. In equilibrium, we would see workers distributed spatially such that the marginal product of labor would be equal across the country. Now of course we will never actually reach equilibrium because the economy is made of so many moving parts, but this gives us a frame of reference to look at data and see what direction things are moving in. If the US were moving closer to equilibrium, we would expect to generally see convergence in the spatial distribution of wages; recall that it is assumed that wages and the MPL are linked.

What the authors find is the opposite, however: after controlling for factors like worker characteristics, the geographic variation in wages across cities in the US has increased since 1964. This suggests that differences in worker productivity across cities are growing.

The natural follow up question is, why? Land use regulations seem to be the culprit. Over the past several decades, there has been a huge increase in demand for finance and tech skills, which are concentrated in New York City and the Silicon Valley areas. In other words, these local labor markets have undergone a large positive shift in demand. However, stringent land use regulations have made the labor supply curve in these areas highly inelastic, meaning the demand has primarily resulted in an increase in wages instead of employment.

The key here is to realize that the consequences of this are lot merely local; this affects the country as a whole. There are workers with high potential MPLs that are prevented from fully realizing them, and are stuck in lower productivity jobs in lower productivity areas simply due to the fact they cannot physically move to the region that would best use their talents. This means that the US has a lower level of output, and social welfare, than in the counterfactual reality where housing policy was more flexible.

I’m increasingly convinced that these stringent land use regulations and related NIMBYism are responsible for much of the economic pain of the past few decades. Mobility in the US has declined significantly since the 1980s, and the increased social sorting such policies encourage has only exacerbated racial and class differences, leading to a horrible generational feedback cycle and overall loss of opportunity.

Moving the locus of land use policy away from the local level is low hanging fruit with huge potential upside, and yet the national conversation on economic growth pays essentially zero attention to this. This needs to change.




Fabula delenda est

In the coming days and weeks there are going to be many words about ‘the meaning’ of this election. But in the spirit of the name of this blog, we should be cautious in believing any of them. The fact is that the margin in this election was so close that interpreting what was essentially a coin flip as portending a seismic shift in American politics is folly. Nate Silver makes the point here that if one in one hundred voters, just 2% of the voting electorate, had shifted to Clinton, we would be crafting an entirely different narrative.

Furthermore, political science models that were based on fundamentals, such as the state of the economy and length of incumbent party control, predicted a Trump victory (why people didn’t take such models seriously is a separate issue). As Larry Bartels argues, this should suggest to us that in many ways this election was entirely normal.

That being said, of course this election still raises many questions:

Continue reading “Fabula delenda est”