An Agent-based Model of Wood Markets in Switzerland

Abstract

The sustainable potential of the resource wood is currently not tapped in Switzerland, as the amount of wood harvested is smaller than the amount that regrows. There are many possible reasons for this, such as the fine-grained supply structure: in Switzerland, there are approximately 250'000 private forest owners who own on average a forest area of approximately 1.5 ha. This makes the sale of wood financially not very attractive for them. Also, public forest enterprises in Switzerland manage rather small forest areas compared to other European countries. Personal relationships play an important role in the Swiss wood markets, so potential business transactions are not solely evaluated based on economic criteria. In alpine areas, harvesting wood is associated with high costs. At the same time, wood prices are largely determined by international wood prices, which comes along with a dependency of the Swiss wood markets on the exchange rate EUR-CHF.
Computer simulation is one possibility to examine how these and other characteristics of the Swiss wood markets influence the availability of wood. The development of a computer model allows the simulation of different scenarios in order to gain insights that are otherwise difficult to gain. For example, the impact of the market participants' decision-making behavior on the quantities of wood available on the market can be investigated, i.e. which market participants can get what amount of wood at what time, and what prices they have to pay for it. The influence of the market structure, such as the existence of intermediaries or potential consequences of combining smaller forest enterprises to larger ones, can also be examined.
The characteristics of these markets make agent-based modeling a promising approach to simulate these markets. In an agent-based model (ABM), a system is modeled by describing its constituting entities, referred to as agents. In the given case, these agents represent the market participants. Each agent can be attributed with individual characteristics and behavior. When these agents are simulated, the aggregate behavior can be observed, e.g. the quantities traded on the market and the corresponding prices. This aggregate behavior emerges from the many interactions and decisions of the individual agents. The behavior of the entire system is often unpredictable, sometimes even counterintuitive, if only the behavior of the individual elements of the system is considered. It is only the interplay of these elements which makes emergent phenomena observable. The ability to observe emergent phenomena is a central strength of agent-based simulation. The bottom-up description of the system also allows the modeling of the structures and relationships at the level of individual market participants. These play an important role in the Swiss wood markets.
This dissertation was conducted according to the design science research paradigm. Concerning the specific requirements of this thesis, this means that the aim was to create a model that represents the Swiss wood markets with sufficient accuracy, so that it can be used to gain new insights into these markets.
This cumulative PhD thesis comprises four peer-reviewed journal publications, three of them are already published, the fourth is submitted. Each of these publications deals with a central step towards the overall goal: the first shows a simplified agent-based model of the sawlogs and energy wood market in the canton of Aargau. Simple scenarios demonstrate the suitability of the approach, but also reveal issues which need further research. The second publication focuses on how discrete choice experiments can be used to enhance the empirical foundation of the agents' decision-making behavior. The third publication describes the validation of the model, which now represents the sawmill, energy wood and industrial wood markets (all three further divided into one market for softwood and one for hardwood). Validating the model is an important prerequisite to use it for scenario analysis. The last publication in this thesis finally shows the simulation and analysis of various politically relevant scenarios. This includes an analysis of how the market is influenced by the presence of intermediaries or by the intensity of profit orientation of forest owners.
The developed model can be parameterized for different Swiss regions in order to simulate the regional wood markets. The necessary empirical data was gathered for the cantons of Aargau, Bern and Grisons. Currently, the model is parameterized and validated for the canton of Grisons. Particularly the last publication demonstrated that the model can be used to provide insights into these markets by simulating and analyzing different scenarios. However, the model itself is not the only contribution of this thesis. On the one hand, to develop this model, several problems had to be solved that could be of interest to modelers of other markets. One example is the problem of defining the geographical model boundaries in a market that is heavily influenced by the surrounding international markets. On the other hand, the creation of such a model already provides many valuable insights into the markets under study, since knowledge about them must be gathered and interpreted. This makes the journey a considerable part of the reward.