Agricultural structures have been shaped by a variety of factors including economic, cultural, historical, political, technological, and geographical conditions. Moreover, agricultural structures are not static. Structural change can be characterised as an evolutionary process consisting – among other things – of constant adjustment to changes in demand, supply, and technological progress. The development of competitive and efficient agricultural structures has been one of the central goals of agricultural policy making in addition to ensuring a fair standard of living for farmers. To achieve these goals, the agricultural sector in most industrialized nations has long been the subject of government interventions. However, many agricultural policies have worked counteractively to these goals by creating distortions in the use of resources. Against this background, the emergence of new and innovative modeling methods such as agent-based models, in addition to ever-increasing computing capacities has offered new possibilities to model adjustment reactions and to quantify the impact of agricultural policies. This thesis takes up these new methodologies and applies them to the modelling and evaluation of agricultural policy impacts on regional structural change in Baden-Württemberg. The thesis contributes to a deeper understanding of structural change dynamics and factors causing structural change. The starting point of the analysis is the hypothesis that Hohenlohe's agricultural structure displays structural inefficiencies and that structural adjustment in the past has been impeded by existing agricultural policies. Based on these assumptions, it is studied whether and to what extent policy changes can facilitate structural adjustment towards a more efficient and competitive agricultural structure. At the centre of the thesis is the development of the spatial and dynamic simulation model of regional agricultural structures AgriPoliS (Agricultural Policy Simulator). The core of AgriPoliS is the understanding of a regional agricultural structure as an agent-based system, i.e., a system of interacting heterogeneous agents. The model extends previous work by BALMANN (1995, 1997). AgriPoliS is a normative spatial and dynamic model in which a number of individually acting farm agents interact with each other subject to their actual state and to their individual environment. Farms can engage in a number of production activities, invest into buildings and machinery, operate as part-time farms, or leave agriculture altogether. The initial situation of the model is calibrated to the agricultural structure of the region 'Hohenlohe' in Baden-Württemberg in the reference year 2000/2001. The calibration aims to map the farming structure in the region and the variety of prevalent farms and production activities. The main data sources used are accountancy data (FADN), regional statistics, investment data, as well as technical production coefficients. The full implementation of the Agenda 2000 by the end of 2002 is taken as the reference scenario. In a first set of simulations, AgriPoliS' behaviour in the reference scenario is investigated. In particular, parameter values for technological change, interest rates, region size, and managerial ability are varied. Accordingly, high interest rates together with heterogeneous managerial ability and high technological change induces a particularly strong structural change in the region. Moreover, results show no strong variations in response to different initial conditions. Finally, an impact analysis of managerial ability on farm survival shows that farms with higher managerial ability are more likely to survive structural change. In the policy simulations, emphasis is put on policies that are likely to support the development towards more efficient and competitive agricultural structures. In particular, a retirement payment, a stepwise cut of coupled direct payments, as well as several scenarios with decoupled direct payments are simulated (de-coupled single farm payment, single area payment, and partially decoupled payments). Key results are the following: - Structural effects: A fully decoupled single farm payment granted independent of agricultural production shows to have strong effects in the short-run. Shadow prices for production factors fall dramatically. Thus, farms have stronger incentives to spend less on leasing land and to look for alter-native uses of the complementary factors labour and capital. This means accelerating structural change. In addition, marginal land may no longer be managed. A basic land management premium of 50 € per ha prevents most land from falling idle. A stepwise reduction of payments leads to a gradual increase of adjustment pressure resulting in land prices falling over time. On the other hand, a single area payment if implemented as payment entitlements per hectare results in no significant changes compared to the reference scenario. However, in the long-run, results do not differ greatly between scenarios. - Income effects: As for the single farm payment and the retirement payment, unprofitable farms, and farms with a growth potential, benefit already in the short-run. Farms in the first group benefit because they are rewarded for leaving the sector. This takes away pressure from the land market, as more land is available for lease. Surviving farms benefit because they have the opportunity to lease land at lower prices and to realise size effects more easily. Most of all, farms with a high share of rented land benefit from falling land prices. Over time, a stepwise reduction of direct payments also transfers into lower rental prices. In case of this policy, farms that previously existed because of direct payments are successively pushed out of the sector. Losers of policies leading to lower rental prices will be land-owners as a drop in rental prices implies land values to decline. - Efficiency: A Data Envelopment Analysis shows that on average efficiency increases significantly, if an incentive is provided to unprofitable and inefficient farms to leave the sector. If no incentive is provided or direct payments are coupled to production, inefficient farms remain in production. This is also the case for a uniform area payment. However, since inefficient farms are relatively small, the production share of inefficient farms is also relatively small, such that the output-weighted structural efficiency shows only small differences between policy scenarios. To assess the efficiency of production in the region, economic land rent is used as a measure for the allocation of factors. In all scenarios, average economic land rent increases with time despite of lower rental prices in the case of fully decoupled payments. Considering adjustment costs, efficiency gains are highest and adjustment costs lowest for decoupled single farm payments. The results show agricultural policies to affect structural change in many ways.