Part III Policy analysis 11
7 Retirement payment, phasing out, and a decoupled
1.3 Structure of the study
The thesis is organised in three parts, following the three sub-goals. Part I of the thesis introduces agent-based systems in general as well as the use of the ap- proach as a tool for understanding and modelling regional agricultural struc- tures. Chapter 2 motivates agent-based models as a conceptual framework for modelling in (agricultural) economic research. The chapter furthermore provides a formal definition of agents and agent-based systems. Specific features of
1.3 Structure of the study 7
agent-based models relevant for modelling regional agricultural structures are discussed, and applications of agent-based models are presented. Particular fo- cus is put on applications relevant in agricultural and resource economics. Fi- nally, chapter 2 points out some theoretical and practical challenges of agent- based modelling. Having provided the theoretical background for agent-based modelling in chapter 2, chapter 3 presents AgriPoliS. The agricultural structure created in AgriPoliS consists of a number of spatially arranged individual farms, called farm agents, that act individually and interact with each other subject to their actual state and to their individual environment consisting of other farms, factor and product markets, and the technological and political environment. Farm agents pursue the goal of household income maximisation, they can en- gage in a number of production activities, invest into buildings and machinery, operate as non-professional farms, or leave agriculture altogether.
Part II of the thesis presents an application of AgriPoliS and formal testing and
sensitivity analysis to validate the model. In chapter 4, AgriPoliS is calibrated to the agricultural structure of the region 'Hohenlohe' in Baden-Württemberg in the financial year 2000/2001, which is taken as the reference year. The calibration aims to map key characteristics of Hohenlohe's farming structure as well as the variety of farms and production activities. In AgriPoliS, farm agents are defined based on farm accounting data of real farms in Hohenlohe. In addition, regional statistics, investment data, as well as data on technical coefficients (e.g. KTBL) are taken to represent the structure of agricultural production in the region.6 The full implementation of the Agenda 2000 by the end of 2002 is taken as the refer- ence policy scenario. Chapter 5 further investigates and tests the behaviour of AgriPoliS under the reference policy. To shed some light on AgriPoliS from dif- ferent points of view, three separate simulation experiments are conducted. It is aimed to learn more about specific characteristics of AgriPoliS and its behav- iour. First, the impact of different parameter constellations for technological change, interest rates, the region's size, and managerial ability are analysed with respect to their impact on results. In particular, the statistical technique of De- sign of Experiments (DOE) is applied. This procedure involves the systematic variation of parameter values and the subsequent identification of result patterns. Second, repeated simulations with different random numbers aim to show that simulation outcomes are robust against variations of initial conditions. Finally,
6
The Kuratorium Technik und Bauwesen in der Landwirtschaft (KTBL) is a German insti- tute, which regularly identifies and publishes data, e.g., on technical coefficients of pro- duction, production technologies, gross margins.
an analysis of the impact of managerial ability on farm survival is conducted.
Part III of this study is devoted to the simulation of the mentioned agricultural
policy scenarios with AgriPoliS and subsequent analyses. The study is based on an application of AgriPoliS to the Hohenlohe region. The analysis of simulation results aims at identifying the policy impacts from different perspectives: farm size, production, technical efficiency, economic efficiency, income, and gov- ernment outlays. For all scenarios, the policies defined under the Agenda 2000 serve as the reference policy scenario. Chapter 6 introduces different indicators and analysis techniques. A popular technique to investigate efficiency differ- ences between different farms based on observed data is Data Envelopment Analysis (DEA). Chapter 6 introduces efficiency analysis and different effi- ciency concepts. It presents a DEA model to study differences in technical effi- ciency between farms as well as the structural efficiency of the region. The model uses simulated data from the policy experiments. Furthermore, analysis techniques such as Kernel density estimation, Gini coefficients, and Lorenz curves are introduced which are subsequently used to analyse indicators derived from the policy simulation experiments. Chapter 7 shows results of policy ex- periments comparing the impact of a retirement payment with fully decoupled direct payments and a successive phasing out of direct payments. The approach followed is to contrast the impact of three policy alternatives, each of which is thought to have substantial impact on structural adjustment. Based on the simu- lation and the analyses of these policy scenarios, the goal is to identify some fundamental structural dynamics and adjustment patterns, which are discussed in the summary of the chapter. Chapter 8 follows a different intention. Based on the elaboration of more fundamental adjustment patterns in chapter 7, different ways of decoupling direct payments are analysed. In particular, a fully decoup- led single farm payment is analysed against a single area payment and partially decoupled payments. This set of policy experiments reflects, in a more general way, policy concepts that will be introduced in the European Union's common agricultural policy from 2005 onwards. At the end of this chapter, simulation results are summarised and discussed
Finally, chapter 9 summarises the thesis. It places the obtained simulation re- sults in the context of policy analysis as well as of the methodology of agent- based modelling. Future directions of research are pointed out as well as possi- ble extensions of the current version of AgriPoliS.