Methods in Economic Farm Modelling

Alexander Gocht



The objective of this thesis is to develop methods for the evaluation of agricultural firms using efficiency analysis and to develop and assess farm responses in mathematical programming (MP) models to changing political and economic conditions. The dissertation is structured in four main parts.

Chapter 2 extends Data Envelopment Analysis (DEA) by incorporating confidence intervals in the evaluation of the resulting point estimates. In the literature, agricultural farms are often evaluated and compared based on DEA, where causes of inefficiencies within a farm group are often analysed by regressing efficiency measures on other variables. However, when confidence intervals are taken into account, the results of this analysis show that neglecting the stochastic nature of efficiency measures cannot produce any valid conclusions about the real nature of inefficiencies. Hence, DEA efficiency measures need to be carefully interpreted, and further research is necessary before this methodology can be used as a standard approach for evaluating the efficiency of farms and other firms.

Chapter 3 analyses the responses of MP farm group models induced by a change in political and economic conditions. MP models are widely used as decision models in agricultural economics. In contrast to an application on the farm level with considerable modelling detail, an analysis of macroeconomic effects is often only reasonable if it is based on representative farms. However, only sparse information is available for the specification of aggregated representative farm groups. Furthermore, decision variables should reflect observed behaviour through a process known as calibration of MP models. Positive Mathematical Programming (PMP) has been developed for this purpose, a method that calibrates the objective function with the help of a non-linear costs component and determines simulation behaviour. The influence of the different proposed PMP variants on simulation results is compared ex post with observed values using the representative farm model FARMIS. This is done through 45 farm groups; these data were obtained from the German Farm Accountancy Data Network (FADN). Based on these farm groups, PMP calibration methods are applied for the year 1996/97, and a shock is introduced for observed gross margins of 2002/03. Comparison of the calibration methods reveals that the simulation strongly depends on the PMP method applied.

Chapter 4 develops an estimation method for the specification of crop-specific input coefficients in MP models. The lack of information about input allocations for different crop levels, e.g., fertiliser inputs for wheat or the level of pesticides used for sugar beets, provides a challenge for the specification of aggregated farm type models. In farm accounting records available for farm group models, often only total inputs per farm are reported. In aggregated MP farm type models, the explicit representation of input allocation plays an increasingly important role, for example in the representation of environmental effects such as nitrogen intake, and subsequently in the modelling of policy alternatives. In the past, crop-specific inputs were either implemented ad hoc in MP models based on management handbooks, or were based on total input levels that were estimated with input-output regressions. This chapter presents an approach that combines the regression approach with the estimation of a farm supply model using single farm data. The relationship between the MP and the linear regression model is defined, and an estimation approach based on the optimal condition of the farm is presented. The developed estimation approach is applied to Belgian FADN data, where input allocations for various crop levels are collected in the database. A comparison of observed and estimated data is possible to validate the suggested method. The results show that the developed estimation approach successfully models the observed values of input allocation, in contrast to the regression estimation. Furthermore, this approach leads to a crop-specific breakdown of variable inputs and a representation of the resulting farm type with a fully specified non-linear component.

Chapter 5 presents the farm type module developed in the modelling system CAPRI (Common Agricultural Policy Regional Impact). The integration of farm types into the modelling system CAPRI provides the chance to directly quantify the effects of market policies and developments on the farm level and to reduce the aggregation bias, resulting in an improved localisation of farm type related environmental effects. The farm types in CAPRI are based on data from the European Farm Structure Survey (FSS). For several reasons, these data are not consistent with the CAPRI database. One possible way to overcome these inconsistencies would be a simple linear up- and down-scaling of FSS to the quantity structure of the CAPRI database. However, this method could lead to a loss of information about the type and size of the farm group from FSS. To avoid this effect, an estimation approach is developed covering EU-27 that does not violate the type of farming or the economic size of the farm types.

Complete version (5 MB) Hier können Sie den Adobe Acrobat Reader downloaden


© Universitäts- und Landesbibliothek Bonn | Published: 20.01.2010