Part III Policy analysis 11
7 Retirement payment, phasing out, and a decoupled
5.5 Conclusions from the analyses
of this comparatively strong result, the shortcomings of the chosen approach should be mentioned. First, it was assumed that farm agents could not learn, i.e. their managerial ability could not improve (or decrease) in the course of a simu- lation run. Accordingly, a farm initialised as a farm with low managerial ability will remain so during the entire simulation.
Second, the definition of managerial ability and the way of operationalising it are admittedly very simple and it is questionable whether it is possible and ad- missible to narrow down such a diverse concept to only a single factor and a single effect. Measuring managerial ability real systems is in fact an ambitious task that often involves such diverse subjects as psychology as well as produc- tion frontier approaches. Furthermore, the interpretation of managerial ability as the ability to realise costs and produce more efficiently is certainly not generally applicable although there is some evidence that management contributes more to the efficiency of production than for example organisational structure (GA- LUSHKO and BRÜMMER 2003).68 However, factors like the motivation of em-
ployees or personnel management play a role as well. However, as for family farms in the region Hohenlohe considered in this study, the chosen interpretation of managerial ability seems more appropriate.
5.5 Conclusions from the analyses
The simulations in this chapter underline the importance of analysing the sensi- tivity of a model from different points of view. This chapter followed three dif- ferent approaches to exploring the reference situation given by the Agenda 2000. The first approach applied the DOE methodology and a simple regression metamodel to observe the impact of parameters expected to have significant ef- fects on the simulation outcome. It showed that, for example, the size of the re- gion has no significant effect on results whereas interest rates, on the other hand, together with technological change and managerial ability significantly influ- ence model results.
The second approach analysed the effect of multiple random initialisations on simulation results. Results of 20 independent model replications did not show any significant abnormalities or extreme values, although these can never be ex- cluded fully if more replications are made or parameter values changed.
68 See B
ALMANN and LISSITSA (2003) for a collection on the topic of large farm manage-
Third, an analysis of the effect of heterogeneous managerial ability was under- taken which focussed on the adjustment of different groups of farms. From this analysis, it becomes apparent that different farm agents follow different adjust- ment strategies depending on their individual situation regarding managerial ability, but also with regard to aspects like factor endowment, farm type, or lo- cation, which were not analysed further. Adjustment strategies will differ across farm agents. Hence, it can be expected that some policies will lead to more ad- justment processes in one group of farm agents than in another. A policy change thus is expected to affect different farms in different ways. This could be an in- teresting result for policy makers because it shows that policies made based on an average farm do most likely overestimate or underestimate the effect of the policy on a particular group of farms or even individual farms.
Furthermore, results show that different and important insights can be gained from simulating different possible framework conditions such as for different interest rate scenarios. Most of these framework conditions could be interpreted as possible futures of the target system. Although, in this chapter, only interest rates were considered, the merit of analysing the effect of different framework conditions becomes obvious. First, this procedure allows the modeller to get a feeling for how the model behaves under different (plausible) conditions, which may eventually make it easier to detect errors. Second, it suggests that the model reacts quite differently under different conditions. As for policy analysis, it could therefore be interesting to carry out simulations not exclusively for one plausible parameter constellation, but also for other plausible constellations. This may be of interest if the introduction of new policy measures is concerned as it is certainly in the interest of policy makers to introduce policies that are equally effective for different future scenarios.
Part III
Policy analysis
6 Analytical framework and assumptions for measuring
policy impacts
6.1 Introduction
The goal of this chapter is to set the general analytical framework for the policy experiments presented in chapters 7 and 8. In particular, this chapter provides the reader with information on the tools used to analyse results from policy ex- periments (sections 6.3 and 6.4). General assumptions underlying the policy experiments are presented (section 6.5). The analytical framework for policy analysis is presented in Figure 6-1. Three steps can be distinguished in setting up the policy experiments: definition of experiments and setting of assumptions, simulation, and the analysis of simulation output representing policy impact.
Figure 6-1: Framework for analysing the effect of agricultural policies on re- gional agricultural structures
Experimental design
(assumptions) Simulation
Output = policy impact
1. Structural development 2. Efficiency
3. Income
4. Budgetary effects
I. II. III.
Source: Own figure.
Results are investigated mainly from four perspectives: structural development, efficiency, income, and budgetary effects. The analysis of structural develop- ment is based on a range of indicators of which the most important are defined in section 6.2. An important analysis tool to investigate differences between farms and to show the distribution of results is Kernel density estimation, which is explained in section 6.4. The second perspective focuses on analysing Hohenlohe's agricultural structure with respect to efficiency (section 6.3). To measure individual efficiency differences between farms, a Data Envelopment Approach (DEA) is applied. Economic efficiency is measured based on the con- cept of economic land rent. To investigate income and equity issues, income
distribution among farms is analysed using Gini indexes and the Lorenz curve (section 6.4).