Estimating Cropland Use in a Multi-County Region of the Southeastern United States

In this thesis, a model to analyze land use in a multi-county region of the Southeastern United States is presented. Farmer planting decisions are assumed to follow a non-stationary first order Markov decision process. The non-stationary transition probabilities are estimated as a function of the prior year‟s land usage and a set of exogenous variables using annual county level data from 1981 to 2005 using the maximum entropy method suggested by Golan et al. (1996). The transition probabilities are applied to each county‟s prior period crop production to estimate crop production in the current period. The model is graphically validated. A discussion is included on difficulties encountered in estimation of the model. Acreage elasticities are estimated and used to analyze the marginal effects of the explanatory variables on crop land use.