The Structural Estimation of Principal-Agent Models by Least Squares: Evidence from Land Tenancy in Madagascar

We develop a method to structurally estimate principal-agent models by ordinary least squares (OLS). We set up a general principal-agent model which explicitly incorporates the wealth levels of each party and the opportunity cost to the agent of entering the contract. This yields an optimal contract that is linearized by way of an Nth order Taylor approximation. This in turn imposes N(3N-1)/2 restrictions on the parameters and yields an empirical test of the canonical principal-agent model. In the application, we consider the case where N = 2 and apply our method to a sample of land tenancy contracts in rural Madagascar. Empirical tests lead to consistent failure to reject the hypotheses derived from our structural model, which lends support to our structural method as well as to the canonical principal-agent model.