Essays on the economics of knowledge spillovers

This dissertation is an empirical analysis of the relationship between international trade and the diffusion of technical knowledge. It uses patent citations and surveys to measure knowledge flows, and examines the relationship between trade and these knowledge flows at both the aggregate and micro levels. It begins by characterizing the types of knowledge diffusion that patent citations measure through a comparison of citations with firms' responses to a survey (the Innovation Survey, or CIS-I) about their acquisition and dissemination of new technology. The types of learning for which citations are a proxy include learning through RD collaboration, technology licensing, equipment purchases, and mergers and acquisitions. Using aggregate bilateral patent citations as a measure of knowledge diffusion, I find that international trade is positively and significantly correlated with knowledge flows after controlling for a variety of factors one would expect to affect technological diffusion. Other variables that are significantly associated with bilateral patent citations include language similarities, geographic proximity, investment flows, and similarities between countries in the mix of patented technologies. Looking at firm-level data, I find that exporters and importers cite more foreign patents than similar non-exporters and non-importers. Exporters' patents are cited abroad more often than those of non-exporters, and foreign inventors cite patents held by French importers more often than patents held by non-importers. The results suggest that firms with more important or more valuable patents make more citations to foreign patents and are cited more often abroad, highlighting the role of the firm's existing knowledge base in learning about foreign technology. Results from the Innovation Survey corroborate the findings discussed above and help shed light on the specific channels through which diffusion takes place. After controlling for endogeneity and selection bias in the relationship between trade and citations using instrumental variables, the relationship between imports and citations remains statistically significant but the independent effect of exporting is eliminated. While firms do not see an increase in citations relative to similar firms after entering export markets, they do cite more foreign patents after beginning to import. I suggest that importers and exporters differ in the ways they learn about foreign technology, and that this may explain why importing appears to have a causal effect on learning as measured by patent citations while exporting does not. Evidence from the Innovation Survey supports this claim, since the types of new technology acquisition most highly correlated with patent citations are the ones most often associated with importing.