| In this dissertation I create a state-level total factor productivity index for ten U.S. manufacturing industries for the period from 1977--1997. This index is used to estimate the size of innovative knowledge spillovers and the effect of distance on these spillovers. Innovative knowledge is proxied with patent stocks from the National Bureau of Economic Research (NBER) patent data. Intraindustry knowledge spillovers are estimated using the Office of Technology Assessment and Forecast's patent class-to-industry concordance. Distance is introduced first through a gravity-weighted specification that discounts the effect of distant patents and then through a non-linear specification that estimates the distance at which the impact of innovative knowledge is cut in half. The technology classes of the NBER patent dataset are exploited to model technology-to-industry spillovers. This method avoids the industry-of-origin/sector-of-use problem that undermines attempts to model interindustry spillovers with patent data. I find Chemical Related Technology and Mechanical Technology provide the largest spillovers. Using the intraindustry framework, I compare the effect of domestic patent stocks with patent stocks from the twenty-five foreign countries that register the most patents in the United States. I find a substantially more powerful spillover effect from innovative knowledge from the countries conduct research activity intensively, the G-7 nations of Germany, Great Britain, France, Japan, Canada, and Italy. Finally, after adjusting for agglomeration effects and for distance, I find that foreign innovative knowledge has the same effect as domestic innovative knowledge. |
