Vol. 66 (1997), No. 2, pp. 151-175 Journal of Economics Zeitschrift for NationalSkonomie 9 Springer-Verlag 1997 - Printed in Austria Subsidizing Cooperative and Noncooperative R&D in Duopoly with Spiliovers Jeroen Hinloopen Received December 4, 1996; revised version received July 24, 1997 Comparing the effect on private R&D investments of allowing firms to co- operate in R&D with that of providing R&D subsidies reveals that in general the latter policy is more effective than the former in promoting R&D activ- ity. Analyzing the implementation of both policies simultaneously reveals that subsidizing cooperative and noncooperative R&D leads to the same market outcome. The preferred R&D-sfimulafing policy is to subsidize optimally an agreement according to which firms only share the outcomes of their indepen- dent research. Keywords: research and development, subsidies, cooperation, Cournot duopoly, spillovers. JEL classification: L43, 032. 1 Introduction Due to the uncertainty surrounding markets for innovations, to negative pecuniary externalities, and to the lower private than social rate of return to R&D, it is widely held that the market for innovations fails in the sense that market forces do not induce firms to invest in R&D an amount considered first-best. 1 To reduce this market failure, authorities have three policy tools at their disposal. Granting a successful innovator a patent over its discovery is traditional practice. However, the provision of patents implies a nonoptimal dissemination of research outcomes (Arrow, 1962) and in practice it appears to be very difficult to restrict technological spillovers (Mansfield, 1985). A second policy to encourage private R&D spending is to allow 1 See Katz and Ordover (1990) for a theoretical discussion. Recent em- pirical evidence is given by Jones and Williams (1997). For an overview, see Hinloopen (1997, chap. 1).