Recent Research: Which Cities Are Poised to Generate New Discoveries?

Metropolitan areas with population densities of about 4,000 people per square mile tend to produce the highest rate of patenting, according to a recent article in the American Journal of Economics and Sociology. In a study of U.S. metro areas over a ten-year period, the authors found that metro population density has a significant positive correlation with patenting rates. At about 4,000 people per square mile, the benefits of agglomeration, such as knowledge spillovers and diverse labor pools, are at their highest, compared to negative effects of congestion, such as increasing costs of real estate and other scarce resources. Few U.S. cities, however, approach this level of population density. The authors advise against taking the averaged optimal density level as a basis for policy, but use their data to suggest that cities play a vital role in the innovation economy and that increasing urban density could lead to higher innovation rates in some U.S. urban areas.

Norman Sedgley and Bruce Elmslie examined the population densities and patenting rates of 302 U.S. metropolitan statistical areas (MSA) during the period from 1990 and 1999. Using a model that accounted for the many other factors that inform patent generation in a city, the authors found a strong positive correlation between population density and patent activity. This correlation follows a pattern of agglomeration benefits that increase as an MSA's density approaches 4,000 people every square mile, and then begins to taper off and decline as the negative forces of congestion begin to outweigh the benefits. The authors suggest that the knowledge spillovers that result from having a population in close proximity help produce new discoveries and are a strong economic argument for cities, even in the face of declining transportation costs and improved telecommunications technologies.

Harvard economist Edward Glaeser's recent bestseller Triumph of the City posits that cities spur innovation by facilitating face-to-face interactions that amplify the skills and knowledge of individuals. Glaeser urges against policies that encourage more dispersed, suburban metropolitan systems, since these arrangements subtract from the economic benefits of population agglomeration. Sedgley's and Elmslie's article supplement the innovation side of this argument with quantitative data on how urban agglomeration correlates with technological discovery. Though each metropolitan area represents a unique mix of resources and institutions that inform the local innovation system, population density appears to be linked to more innovative cities.

A 2007 article by Carlino, Chaterjee and Hunt, referenced by Sedgley and Elmslie, suggested that agglomeration's impact on innovation was more limited. That study used employment density of metro areas rather than population density to evaluate the impact of urban agglomeration. The authors found that a population density of 2,200 jobs per square mile represented the best balance of density benefits with the least amount of negative congestion effects. Holding other factors equal, a city with twice the jobs per square mile would have a 20 percent higher patenting rate. This suggests that a city with about 750,000 resident might represent the optimal patent generating environment.

Sedgeley and Elmslie urge further research on the mechanisms that link agglomeration, innovation and economic growth.. Their previous work includes examinations of the density of economic activity and its effect on innovation on a larger scale, at the state level. Future studies would pursue even more granular data, reflecting the diversity of urban density levels. They intend the current work as a counterargument to those who maintain that improved telecommunications technology reduces the need for geographic proximity. By linking innovation economics and growth theory to urban economics and economic geography, Sedgley and Elmslie seek to emphasize the economic importance of space.

Download Do We Still Need Cities? Evidence on Rates of Innovation from Count Data Models of Metropolitan Statistical Area Patents at: http://onlinelibrary.wiley.com/doi/10.1111/j.1536-7150.2010.00764.x/abstract.