ON THE EVALUATION OF CONSERVATION COST-SHARING PROGRAMS. AN APPLICATION OF A MONTE CARLO EM ALGORITHM

Abstract

The 2002 Farm Bill has placed a great emphasis on programs cost-sharing the adoption of conservation practices on working land. Empirical evaluation of these programs, however, has received little attention. Reasons explaining this gap in the literature may be the voluntary nature of participation and the multi-objective quality of these programs, which complicate econometric analyses. An adequate assessment of cost-sharing programs requires the modeling of multivariate responses that frequently involve limited-dependent variables and other types of unobserved information. In this study I formulate an algorithm that solves such a problem and then I use this tool to evaluate two cost-sharing programs.

I begin this Dissertation by formulating a Monte-Carlo Expectation-Maximization (MCEM) algorithm that solves a variety of models involving unobserved information in systems of linear-in-parameter equations.

Subsequently, I use the MCEM algorithm to solve a multiple adoption model and evaluate the extent at which cost-sharing payments have influenced cropping operations in Maryland. I find that soil conservation practices expand cropping both at the extensive and the intensive margin. I also prove that farmers implement practices that provide on-farm benefits preferentially. Finally, I show that cost sharing has a perverse effect: since farmers prefer to implement practices that provide private benefits, the expansion in cropping induced by cost-sharing those practices reduces the extent at which practices that provide public goods are used.

In a second empirical analysis, I analyze policy implications of ignoring nutrient dynamics in the targeting strategy of programs cost-sharing soil fertility recovery. The analysis focuses on the phosphorus fixation problem in soils derived from volcanic ash. Using an optimal control framework, I conclude that, conditional on individual characteristics, the optimal fertilization path leads either to follow a low-yield fertilization strategy or to maintain a high-yield phosphorus level in the soil. Empirical estimations on Chilean data show that program impact differs among the two regimes and that program efficiency can be improved by targeting preferentially those farms financially or technologically constrained on the low-yield fertilization path instead of allocating the funds conditional on whether or not phosphorus stock is below or above an exogenously determined target level.