A0749
Title: A new model for agricultural land use modelling and prediction in England using spatially high-resolution data
Authors: Pipat Wongsa-art - City, University of London (United Kingdom) [presenting]
Abstract: The focus is on better understanding farmers' responses to behavioural drivers of land-use decisions by establishing an alternative analytical procedure that can handle complex data structures and overcome various methodological drawbacks suffered by methods currently used in existing studies. Firstly, the procedure uses spatially high-resolution data so that idiosyncratic effects of physical environment drivers, e.g. soil textures, can be explicitly modelled. Secondly, the well-known censored data problem is addressed, which often hinders a successful analysis of land-use shares. Thirdly, spatial error dependence and heterogeneity are incorporated in order to obtain efficiency gain and a more accurate formulation of variances for the parameter estimates. Finally, the computational burden is reduced, and estimation accuracy is improved by introducing an alternative GMM-QML hybrid estimation procedure. The newly proposed procedure is applied to spatially high-resolution data in England and found that, by taking these features into consideration, conclusions are formulated about the causal effects of climatic and physical environment, and environmental policy on land-use shares that differ significantly from those made based on methods that are currently used in the literature. Moreover, it is shown that the method enables the derivation of a more effective predictor of land-use shares, which is utterly useful from the policy-making point of view.
