A Semi-parametric Analysis of Technology, with an Application to U.S. Dairy Farms

This article proposes a semi-parametric stochastic frontier model (SPSF) in which components of the technology and of technical efficiency are represented using semi-parametric methods and estimated in a Bayesian framework. The approach is illustrated in an application to US farm data. The analysis shows important scale economies for small and medium herds and constant return to scale for larger herds. With the exception of labor, estimates of marginal products were close to the value expected under profit maximization. Finally, the results suggest important opportunities to increase productivity through reductions in technical inefficiencies.