Anticipated post-innovation collusion encourages R&D effort, but realized collusion later yields deadweight losses. In balancing this trade-off, Bertrand industries sometimes outperform Cournot; sometimes not. Both usually out perform perfectly collusive industries. The optimal level of collusion is often less collusive than Cournot duopoly...
This report has been published in The RAND Journal of Economics- Vol. 24, No. 2 (Summer, 1993), pp. 157-197
Quirmbach, Herman C., "R&D: Competition, Risk, and Performance" (1991). ISU Economic Report Series. 7.