Bio-economic farm modelling for integrated assessment of agricultural and environmental policies: towards re-usability and improved empirical validity

Keywords: integrated assessment, environmental policy, agricultural policy, market liberalization, bio-economic model, farming systems, mathematical programming, maximum entropy estimation, data envelopment analysis, agricultural activity, land use, future studies. The main objective of this PhD thesis was to develop and evaluate a generic bio-economic farm model that can be used under different biophysical and socio-economic conditions for integrated assessment of a variety of agricultural and environmental policies. The functionality of the generic bio-economic farm model developed in this thesis was illustrated with an analysis of the impacts of the 2003 reform of the Common Agricultural Policy in the European Union for arable and livestock farms in a context of market liberalisation. In bio-economic studies, estimation of model parameters related to increasing costs because of limited machinery and managerial capacity, decreasing yields because of land heterogeneity and risk aversion is often not possible because of lack of data. Not including or misspecifying such parameters can have negative consequences on the forecasting performance of the model. In this thesis, methodologies based on Positive Mathematical Programming and Maximum Entropy estimation were proposed and implemented to recover unknown parameters underlying the actual decision making of farmers and to improve the forecasting performance of the model. The proposed methods relax a number of arbitrary assumptions of existing calibration methods and enhance representation of the actual decision making. The forecasting capacity of the models calibrated with the proposed methods was tested in ex-post experiments in which the models were calibrated with historical data of a particular base year and used to forecast policies and price changes of the following historical years. Results of these ex-post experiments showed that the proposed calibration methods improve the forecasting capacity of the model. For meaningful assessment of future policies using bio-economic models, a comprehensive set of alternative activities must be identified. Combinatorial procedures and filtering rules have been used in the literature to generate a set of activities that can be evaluated in bio-economic models. One very important limitation of combinatorial procedures is that the number of generated activities can easily explode. However, many of these activities are inferior with respect to their input-output relationships and they will never be part of the solution of the bio-economic farm model. In this thesis, a method based on Data Envelopment Analysis was proposed to identify and select alternative agricultural activities, representative for specific policy questions that can be used in bio-economic models. The Data Envelopment Analysis method reduced the number of alternative agricultural activities generated by existing combinatorial procedures by 95%, arriving at a number that can easily be applied in bio-economic farm models. The proposed method was applied to a problem of alternative nutrient management in Flevoland (the Netherlands).