| We study alliances in the biotech industry. Typically such alliances occur between a small biotech firm and a large pharmaceutical company, and they are worth several billion dollars across the industry. The biotech owns proprietary technology capable of developing a certain product, but doesn't have enough resources to do so independently. On the other hand, the pharmaceutical company has sufficient resources to develop product, but lacks the technology.;We model this alliance using the principal-agent framework, with the pharmaceutical company as the principal and the biotech firm as the agent. A key variation between our model and those usually studied is that the effort of the agent does not translate into immediate payoffs for the principal. Rather it only affects the time distribution of eventual success or failure of the project and thus indirectly influences the principal's payoff. This modeling twist is essential to capture the hit-or-miss nature of biotech product development.;We adapt the method of Sannikov to incorporate this model feature, and thus to design an optimal incentive-compatible contract using a Hamilton-Jacobi-Bellman (HJB) equation. We numerically compute solutions to this equation, and use these computations to illustrate the structure of the optimal contract. In particular, we establish the superiority of our contract over the myopic contract, which is commonly used in practice, from the perspectives of the pharmaceutical company, the biotech firm and the general public. We also show that it is not in the agent's interest to increase the noise of the observation beyond a level, even if he could, establishing the robustness of our contract.;Our analysis and simulation results show that the principal should compensate the agent based on a noisy observation, and that closer monitoring is generally better for both parties, as well as for the general public. This leads us to the hypothesis that contracts that provide more intermediate payments will be more successful. An empirical test using real-world data on alliances purchased from Deloitte Recap (http://iq.recap.com) lends support to the hypothesis. |
