R&D alliances (Research Joint Ventures or other institutional forms) normally involve repeated, non-contractible actions (investments in R&D), and uncertainty regarding both success and the termination date. Accordingly, we model these agreements as equilibria of infinite-period supergames. Our approach is normative, namely that of finding optimal equilibria from the perspective of the firms involved in the agreement. The results show that repeated interaction allows for important gains in equilibrium pay-offs. The optimal solutions are still inefficient from the firms’ perspective, however. The sources of inefficiency include delay in investment outlays, suboptimal levels of investment, and abandonment of profitable projects. Lastly, we consider R&D cooperation between firms that also interact in the product market. In some cases, product market interaction is irrelevant from the perspective of optimal R&D agreements. In other cases, optimal agreements imply that firms behave more aggressively in the product market.
