New methods and data are enabling us to finally get causal evidence for – and possibly against – Marshall’s three classic pillars of agglomeration economies. I will present two sets of results, first, a causal model of knowledge spillovers, and second, recent work that agglomerations ultimately discourage exploration of new technologies. A great deal of effort has been expended in responding to Krugman’s 1991 challenge that knowledge spillovers leave no paper trail. Here we combine the classic approach that uses citations with the quasi-experiment of inventor or scientist death, for collaborative documents where we can compare the difference in citations in regions of the deceased vs. a still-living co-author. I will present results for regional economics and the diffusion of science for a simple model and a test of absorptive capacity using a more complex model. The work takes us a step closer to a visualization and causal estimation of one Marshallian mechanism of agglomeration. Enrico Moretti recently provided causal evidence that clusters lead to greater inventor productivity. Here we reproduce those results with a larger dataset but go on to show that the simple productivity results are driven entirely by the exploitation of known technologies; clusters actually decrease the exploration of new (and more valuable) technologies. We present a simple economic model, evidence, and a brief discussion of the policy implications. The papers are collaborations with Ben Balsmeier, Sonja Lueck, and Gustavo Manso.
The full presentation is available on Youtube: LINK