Working Paper No. 07-05

Academic Knowledge Spillovers Re-examined: a Look at the Effect of Exogenous Federal Funding

Eric Stuen

October 2007

ABSTRACT

Discoveries in basic scientific research at universities are useful in applied research in industry and sometimes lead to commercially valuable innovations. Many empirical studies have documented a positive relationship between academic research and innovation by firms. However, interpreting this relationship as a causal spillover from academia to industry is difficult since a substantial share of academic R&D is funded by industry. Proximity to industry also influences the quality of professorial talent at a university, and location decisions of both industry and academics may be correlated with other unobservables. Given the presence of such difficult identification issues, this paper uses a novel empirical method to re-examine the effect of academic research in particular metropolitan areas on commercial innovation produced in those locations. I exploit the fact that members of certain appropriations sub-committees within the U.S. Congress can influence the process of allocating federal research funds in favor of their constituents, which leads to ‘exogenous’ variation in research funding at particular universities that is plausibly uncorrelated with factors that affect industrial
innovation.

I construct a detailed panel of micro-data on patent counts, publication counts, doctorates granted and industry R&D expenditures at the metropolitan area/technology area/year level with which I find evidence that measures of academic scientific knowledge are positively related to industrial patenting. Using a city-year panel dataset of industrial patents and academic publications, I find an elasticity of patents with respect to publications from universities in that metropolitan area of 1.17, but that this elasticity is reduced to 0.95 when relying only on variation in publications attributable to “congressional favors”. This translates into an extra patent produced by industry for every 7 extra academic publications produced by universities located in that city owing to the extra research funds diverted to those universities. An elasticity of patents with respect to citations to academic publications of 0.55 is also reduced in the IV set up. These results provide evidence of spillovers unrelated to ordinary market transactions between firms and universities. Data on Ph.D. recipients from each university is used to examine another channel through which academia affects industry – the employment of students with frontier- level technical knowledge. My results show that the employment of new doctoral graduates in science and engineering by industry is also positively related to industrial patenting.

DOWNLOAD PAPER