If you believe that housing density is one of the most important policy areas today, this new and fascinating paper
by Rebecca Diamond and Enrico Moretti is a must-read. The authors have an amazing dataset - transaction-level consumption data for 5% of American households for a full year! - which they use to investigate how consumption levels vary across cities for households of different income brackets and education levels. The short answer is: a lot. San Jose
is twice as expensive for low income households as Natchez, MS
. Low income families in the least expensive cities consume 74% more than those with the same income in the most expensive.
Perhaps the most interesting findings show the relationship between education, geography and income. For college graduates, there is effectively no relationship between their city’s cost of living and their consumption. It seems that higher wages for high-skill jobs in the most expensive cities almost exactly compensate for higher costs. But this is not true for those with less education. Lower skill jobs in expensive cities don’t pay enough to offset the higher cost of living. On average a high school dropout moving from Natchez to San Jose and taking the best available job would see their consumption drop by 27%.
There are a number of important implications. One relates to the problem discussed above of building political coalitions for more housing: if the best educated don’t experience the costs of restrictionism, this is harder. Most interesting, though, is to imagine how remote work (see TiB 116
and TiB 171
might be changing this equation (the paper’s data is from 2014). Previously high-skilled workers couldn’t
increase their consumption by moving to cheaper cities; their income would have fallen accordingly. In 2022, that’s likely no longer true. That’s likely, in equilibrium, to have a profound impact on the economic geography of the developed world.