Three essays in contracting under asymmetric information

This dissertation analyzes contractual choice in a principal agent framework with asymmetric information. Chapter 2 sets up a theoretical model to analyze contract choice in agriculture in a two period principal-agent framework. The principal does not know the agent's true type but has prior beliefs over types. We show how multiple contracts, namely wage, rent and share contracts, can co-exist and all three can persist over time. The model is driven by the rate at which the agent discounts the future. For a sufficiently low discount factor, the landlord will always offer a land-lease contract to the tenant. In Chapter 3, using data from three Indian villages, we test the predictions of the theoretical model. With age as a proxy for the discount factor we find that the oldest tenants tend to be sharecroppers, while the youngest tenants tend to be wage cultivators, and those in the intermediate range have fixed rent contracts. We also find that plots with higher per acre value are usually leased out for tenant cultivation. The theoretical model also makes use of the idea that in a dynamic principal-agent relationship characterized by incomplete information on the principal's part, the principal uses any information revealed by the agent to extract the agent's informational rent in future periods. The agent, knowing this, tries to avoid revealing his true type early in the relationship so as to evade more demanding schedules in the future--the "ratchet effect". In Chapter 4, we test for this phenomenon using experimental techniques. The agent produces an output for the principal and can have either a high or a low cost of production. The principal only knows that each cost level is equally likely. The principal's aim is to extract the maximum possible surplus from the agent by setting production quotas and the principal does not commit to a long term scheme. We find little evidence of the ratchet effect. Agents play the game in a naive fashion and reveal their types even when such type revelation is not optimal; and the principal frequently does not exploit such type revelation.