Dynamic Competition, Strategic Knowledge Sharing, and Innovation Policy

My dissertation consists of two essays on dynamic competitions with endogenous knowledge sharing. Chapter 1 develops a multi-stage R&D competition in which firms can acquire the rival's knowledge by poaching its employee. During the competition, firms can attempt to poach their rivals' employee (through a poaching wage offer) to learn from their rival. Firms can also attempt to retain their employees (through a retention wage offer) to prevent their knowledge from spilling over. The motivation for this research comes from the fact that there is variation across U.S. states in the enforcement of post-employment covenants not to compete (CNC). Employees bound to an enforceable CNC are restricted from competing with their former employer for some time after the employment ends. California's statutory law invalidates virtually all forms of CNC, whereas this covenant is enforceable to some degree in most U.S. states. I show that the poaching wage offer exceeds the retention wage offer and employees depart for an outside firm when the inside firm's R&D success rate, relative to the magnitude of profit loss from joint innovation, is sufficiently low. I also show that these two parameters determine the firms' optimal competitive environment. In particular, when engaging in an R&D competition, the firms' profit-maximizing environment can either be (i) an environment in which employees can move between rival firms without restriction (e.g., California), or (ii) an environment in which employees can be legally restricted from making such moves (e.g., Massachusetts). Chapter 2 develops a dynamic contest model in which players compete by accumulating knowledge. During the contest, each player can increase her rival's knowledge level by sharing her knowledge, which, in turn, builds her reputation for being helpful. Players cannot commit to reciprocate in the future and each player's payoff (at the end of the contest) is increasing in her knowledge level but decreasing in her rival's knowledge level. The motivation for this research comes from the fact that competitors often help out their rivals during the course of such contests without any contracts or commitments for reciprocation. For example, employees competing for a promotion typically work as a team, students competing for grades spontaneously help each other learn, and participants competing in crowdsourcing contests often share valuable information to the public. The model is used to characterize the dynamics of knowledge sharing and participation in various types of contests. In particular, I show that the key to knowledge sharing lies in the degree of asymmetries in the players' initial knowledge level. For example, the level of participation in a contest that requires the rival's help (to obtain a positive profit) is not monotonic in the degree of such asymmetries. I also show that, in terms of knowledge sharing, a contest with fixed duration can outperform a contest with a fixed goal. The results can be applied to research joint ventures, intra-firm knowledge management, and admission decisions of academic programs.