Essays on dynamic contracts : allocation of ambiguity and delegation

My dissertation studies the design of contracts in different contexts. It contains two theoretical investigations about contracting under ambiguity: in the context of research partnerships and venture capital financing; and an experimental study to examine delegation of decision rights within organizations. The first chapter studies contract design for innovation under ambiguity. Outsourcing of research is a large and growing trend in knowledge-intensive industries such as the biotechnology and software industries. I model innovation as an ambiguous stochastic process and assume that the commercial firms and research labs differ in their attitude towards ambiguity. I characterize the optimal sequence of short-term contracts and examine how the features of this contract facilitate ambiguity sharing: the dynamic moral hazard problem is mitigated under ambiguity; experimentation stops earlier than is socially optimal; the project may be liquidated even after being granted a patent. I find that redesigning the patent law can not implement the Policymaker’s desired optimum. The second chapter analyzes venture capital investment under ambiguity. A central feature of venture capital financing is the extensive use of control rights as an instrument. In this chapter, I present a model of venture capital financing where investment is allowed to depend on an intermediate ambiguous signal. I show how the presence of ambiguity explains the allocation of control rights if the investor is more ambiguity averse than the entrepreneur. In the third chapter, I discuss how delegation of decision rights can be used as a signal of trust that can be reciprocated by cooperation. First, I theoretically show that in a principal-agent framework, using delegation as a signal is the only Perfect Bayesian Equilibrium that survives forward induction criterion. Then I use experimental methods to test this theoretical prediction. I find that the players do not use delegation very often, thus the forward induction logic is not supported by the observed data. However, once the players are given information about the past sessions, they choose the forward induction equilibrium more often. This suggests that information affects the formation of beliefs and equilibrium selection in Bayesian games.