The thesis begins by illustrating what is meant by strategic behaviour. Three fish farmers living by a lake have to decide whether to go ahead with jointly investing in a wastewater treatment plant. If the fish farmers benefits are private knowledge it is quite likely that the participants will try to keep this information private for strategic reasons. This is just one instance where the provision of so-called public goods is uncertain. There is a vast literature on voluntary public goods provision in similar situations (often labelled social dilemma situations). The main lesson is that strategic behaviour in such situations has a high probability of causing inefficient solutions. Methods from cybernetics and game theory are used to develop an agent based simulation approach which can contribute to further investigation of such dilemma situations. Two foundational chapters introduce some concepts from game theory and environmental economics. Chapter 2 contains a critical review of the rationality assumptions, typically used in standard game theory, and discusses evolutionary game theory as an alternative. Chapter 3 summarises definitions on public goods, external effects, Pareto efficiency and relates the work to the Coase Theorem. Chapter 4 deals with the conception of an agent based simulation tool. Game theoretic models are translated into models which focus explicitly on public goods and describe these goods by state variables. Moreover several models for the adjustment of agent behaviour are provided. Predominantly this concerns imitation models but in addition an adaptation procedure based on the sampling dynamics is implemented.
Main part of the thesis are two applications of the simulation tool. Chapter 5 demonstrates on a simple two-player example how the simulation tool can be deployed. The so-called inspection game has no equilibrium in pure strategies but it has a unique Nash equilibrium in mixed strategies. But this equilibrium has almost no predictive power. No convergence is observed in the simulations but significant fluctuations. Potential interventions by the legislator are also discussed, establishing that such interventions may have counterintuitive effects. Some of these findings can be explained also analytically by which is shown that Lotka-Volterra equations may be derived for specific parameter settings. This reconfirms the close connection between imitation models and the replicator dynamics of evolutionary game theory.
Chapter 6 refers back to the introductory example. So far the problem has involved just three fish farmers. Since it is of specific interest to investigate size effects (increasing player numbers) a more general problem description is formalized. This formal model is related to the Coase Theorem. First investigations concern the ability of the fish farmers to solve the problem on a voluntary basis. In other words: the question is if a spontaneous organisation of agents might result in a more or less efficient solution. The simulation results accord pretty well with suggestions made by Olson (1965). With a larger number of fish farmers, it is almost inevitable that the project will fail. In which case, the potential surplus will completely dissipate. On the other hand, some relief is found with increasing heterogeneity, an issue which was also discussed by the author at a WEHIA conference in Essex in 2005. However, the overall findings for large and medium player numbers remain discouraging. As long as there is no way to curtail the tendency towards understatement, no improvement can be expected. Therefore the final part of the thesis concentrates on “mechanism design” and the simulation of suchlike mechanisms. Simulations are made for the Clarke-Groves mechanism as well as for a mechanism suggested by Rob. The latter mechanism also satisfies the individual rationality condition. One common problem of the Groves schemes, as well as the Rob mechanism and indeed most other mechanisms, is that they induce a truth-telling equilibrium only in theory. Moreover, the compensation (or tax) schemes are often quite complicated. Thus it does not surprise to find that in experiments subjects often fail to find these equilibria. Kawagoe & Mori (1999) demonstrate this “understanding problem” for the pivotal mechanism. When mechanisms or other procedures taken from game theory are found to “work”, one must not forget that the idea of a fully rational individual is usually an underlying assumption. To what extent will a mechanism “work“ under conditions of bounded rationality? That is the question to be answered by means of simulation.
The concluding Chapter 7 refers back to the notion of strategic behaviour, and the need for correct evaluations of public goods is emphasised. Two striking thought experiments show that decision-making in the context of strategic interdependence is quite different to individual decision-making. The first experiment highlights the ambiguity of additional information, whereas the second demonstrates how additional options can be harmful. Again it turns out that these problems are ever-present in social dilemma situations.
|Number of pages||169|
|Publication status||Published - Sep 2007|
|Series||Online-Dissertationen, Universität Stuttgart (Germany)|