Innovation Spillovers and Technology Policy

Abstract

— We examine a model of R&D competition and cooperation in the presence of spillovers. However, unlike virtually all the literature, we treat these spillovers as endogenous and under the control of firms. We show that it is then essential to make a number of distinctions that are ignored in the literature. In particular we need to distinguish between the amount of R&D that firms do and the amount of spillover they generate; between information sharing and research coordination; between each of the latter and cooperation; between substitute and complementary research paths; between firms being located in the same industry or in different industries. These distinctions matter because, as we show, coordination can arise without cooperation (different industries, complementary research, research design) while cooperation need not induce information sharing (same industry, substitute research, information sharing). In many cases, however, allowing cooperation is sufficient to induce full infonnation sharing/research coordination, in which case the justification, if any, for a technology policy that takes the form of an R&D subsidy lies in encouraging firms to undertake more R&D. Our analysis suggests that cooperative arrangements between firms may often produce too little R&D, and therefore that R&D subsidies can be justified — but not to correct information problems, but other market failures in the amount of R&D firms choose to do.

We are grateful to CEPR for a research grant under their Competition and Markets Programme which enabled us to collaborate. Earlier versions of this paper were presented at seminars at CREST-Ensae, OECD, Insead, University of Leuven, WZB Berlin and at the French Economic Association Conference in Nantes (June 1995), and we would like to thank seminar participants for their comments. We would particularly like to acknowledge the many useful comments we received from Raymond De Bondt, David Encaoua, Bruno Jullien, Patrick Rey and two anonymous referees.