Illegal pollution and monitoring of unknown quality : a signaling game approach
Werner Güth and Rüdiger Pethig
In this paper a game model is considered whose strategically interacting agents are a polluting firm that can save abatement costs by illegal waste emissions and a monitoring agent (controller) whose job it is to prevent such pollution. When deciding on whether to dispose of its waste legally or illegally the firm does not know for sure whether the controller is sufficiently qualified and/or motivated to detect the firm's illegal releases of pollutants. The firm has the option of undertaking a small-scale (deliberate) "exploratory pollution accident" to get a hint about the controller's qualification before deciding on how to dispose of its waste. The controller may or may not respond to that "accident" by a thorough investigation thus perhaps revealing his or her type to the firm. It is this sequential decision process along with the asymmetric distribution of information that constitutes a signaling game whose equilibrium points may but need not signal the type of the controller to the firm. In Part I of the paper the formal introduction of the game model is followed by an extensive discussion of four different equilibrium scenarios which are non-degenerate submodels whose (generic) equilibria are considered typical and especially interesting for the monitoring issue at hand. Having set up a rather complex game model the price to be paid is (as in many applications in other fields) the multiplicity of equilibria - even within one and the same equilibrium scenario. This multiplicity clearly weakens the predictive capacity of the model. To overcome it Part II addresses concepts of equilibrium refinement and selection on a fairly technical level. It is shown that the set of equilibria.is reduced - not to a singleton, though - by applying the refinement concept of uniformly perfect pure strategy equilibria. Unique solutions are obtained by reference to the equilibrium selection theoretic concepts of cell and truncation consistency, of payoff dominance and of risk dominance.
