Spatial inequality is concerning, but we don't yet fully understand all forces that contribute to it. In this new paper (with @pa_balland et al), we find that spatial concentration is larger for complex economic activities. (thread) nature.com/articles/s4156… #EconTwitter 
We use four datasets summarizing publications, patents, and employment by industries and occupations for US Metro Areas. Not surprisingly, economic activities in these four datasets clusters in cities.
We can measure this concentration using scaling laws, like the ones described by @BettencourtLuis in (pnas.org/content/104/17…). In agreement with previous findings, we find that large cities have more patents, employment, and publications (per capita!), than smaller cities.
But when we look within each category, we find that activities that look intuitively more complex (e.g. patents in computer hardware) scale more super-linearly than intuitively less complex activities (e.g. patents in pipes & joints).
We formalize the correlation between complexity & spatial concentration using different measures of complexity, such as the number of authors in a paper, the age of patents, or the education level of industries and occupations (plus more measures in the appendix).
All cases show strong correlations between the complexity of economic activities and their concentration in large cities. Finally, we look at the dynamics of spatial concentration for different activities using a time series of patent data.
We find that the spatial concentration of technologies has increased over time more for complex technologies. We also find that technologies that used to be complex (e.g. mechanical) are losing some of their concentration in large cities (which peaked around the 40s).
These results are interesting because they speak directly to a poignant aspect of today's spatial inequality. That between knowledge hubs (e.g. San Francisco, Boston, London), and places involved in simpler economic activities.
If complexity is connected to the concentration of activities in large cities, then the increasing knowledge intensity and complexity of our economy will push to amplify differences between superstar cities (with complex activities and high salaries) and the rest.
Of course, we've had a hint about this for a while (Audretsch & Feldman AER 1996 showed that innovative activities tended to be more concentrated in space: people.unica.it/stefanousai/fi…).
Yet, by looking at four different datasets, and within each one of them, we now know that this concentration is more pronounced for more complex innovative activities, and also, true for non-innovative activities.
So if complexity & spatial concentration go hand-in-hand, we need to ask how to include more people in large cities (e.g. transportation, affordable housing, etc.), or how to promote knowledge diffusion to help develop less complex places. (Full PDF at: static1.squarespace.com/static/5759bc7…)
