Essays in experimental economics on contract design
This dissertation contains three related essays which examine contracting environments with moral hazard. I use laboratory experiments to study how across treatment variations affect contractual outcomes including the types of contracts that principals design, the overall efficiency of the contractual relationship, and the surplus distribution between the principal and agent(s). In the first chapter, which is joint work with Steve Wu, I investigate relational contracting within a bilateral relationship. Specifically, I explore how contracting is impacted by a reduction in the agent's market power as proxied by an exogenous decrease in the agent's expected outside option. Surprisingly, principals did not lower promised payoffs to agents. Instead, contracts are restructured to shift more strategic uncertainty onto agents. Thus, agents are no worse off under successful relational contracting but are significantly worse off when there is a breakdown in the relationship and/or performance outcomes are not favorable. Additionally, agents are more willing to engage in trade despite strategically riskier contracts thereby increasing efficiency via trading volume. An implication of these findings is that standard monopsony models may overestimate efficiency losses from a reduction in volume of trade if production occurs under a contract. In the second chapter, I explore how the relational contract is impacted by the potential for the principal and agent to dispute over the agent's performance level. Both parties publicly observe the agent's performance value during each period in the baseline treatment. In the treatment of interest, the principal privately observes the value and must send the agent unverifiable feedback concerning the value of the performance signal. Therefore, parties may have conflicting beliefs about the agent's performance in the latter treatment only. Despite the lack of contracts that facilitate relational contracting under the private signal, no efficiency losses, as measured by contract acceptance rates and agent effort provision, occur across treatments. Furthermore, principals honor their promised performance bonuses at similar rates. As in the first chapter, however, principals increase their discretion under the private performance signal by shifting more strategic uncertainty onto the agents, but they do not lower the agents' promised payoffs. Agents only experience significant payoff reductions across treatments when the principal under reports a good performance outcome and fails to pay the promised bonus. Principals, however, tend to provide accurate performance feedback so that the agent's overall welfare is unchanged across treatments. These results imply that relational contracting may be relatively no worse off in environments where the performance measure is subjective. In the third chapter, I study a contracting problem with two agents whose production technologies are independent but verifiable. My treatment variation compares the types of team incentives that principals design across static and dynamic contracting relationships. Most principals implement cooperative compensation schemes, which compensate agents the most when both perform well. Furthermore, a larger proportion of observed contracts support cooperation in one-shot relationships. Assuming parties are self-interested, this finding contradicts the comparative static prediction where the proportion of contracts favoring cooperation should not decline when the relationship is repeated. The across treatment difference, however, is statistically insignificant although the types of incentives designed are significantly more variable in repeated relationships. Furthermore, the agent's expected compensation is similar across treatments. I also illustrate how agents who are inequity averse to the principal's earnings, but not to their peer's earnings, can explain why contracts favor cooperation and agents earn higher rents in one-shot interactions.
Wu, Purdue University.
Off-Campus Purdue Users:
To access this dissertation, please log in to our