Abstract: Using a general two-stage framework, this paper gives sufficient conditions for increasing competition to have negative or positive effects on R&D-investment, respectively. Both possibilities arise in plausible situations, even if one uses relatively narrow concepts of increasing competition. The paper also shows that competition is more likely to increase the investments of leaders than those of laggards. When R&D-spillovers are strong, competition is less likely to increase investments. The paper also identifies conditions under which low initial levels of competition make a positive effects of competition on investment more likely.

Abstract: The paper analyzes the effects of more intense competition on firms ’ investments in process innovations. More intense competition corresponds to an increase in the number of firms or a switch from Cournot to Bertrand competition. We carry out experiments for two-stage games, where R&D investment choices are followed by product market competition. An increase in the number of firms from two to four reduces investments, whereas a switch from Cournot to Bertrand increases investments, even though theory predicts a negative effect in the four-player case. The results arise both in treatments in which both stages are implemented and in treatments in which only one stage is implemented. However, the positive effect of moving from Cournot to Bertrand competition is more pronounced in the former case. JEL Classification: C92, L13, O31. Keywords: R&D investment, intensity of competition, experiment. *For helpful comments and suggestions, we are grateful to Michael Kosfeld, Adrian Müller, Max Pfister and to participants at the following conferences:

Abstract: The paper analyzes the e¤ects of more intense competition on …rms’incentives to invest in process innovations. We carry out experiments for two-stage games, where R&D investment choices are followed by product market competition. As predicted by theory, an increase in the number of …rms from two to four reduces investments. However, a positive e¤ect is observed for a switch from Cournot to Bertrand, even though theory predicts a negative e¤ect in the fourplayer case. This result re‡ects overinvestment in the Bertrand case. The results arise both in treatments in which both stages are implemented and in treatments in which only one stage is implemented.