Essays on dynamic non- cooperative games based on simulations and experiments
Not peer reviewed
MetadataShow full item record
This thesis studies ways to improve Non-Cooperative Game theory (NCGT) as a policy making methodology. NCGT allows to categorize a wide range of situations and to provide solutions. However, the effectiveness of such solutions depends on whether cooperative behavior can arise. The common presence of cooperative behavior in real life systems often threaten NCGT solutions reliability. Furthermore, most studies in NCGT literature fail to account for important dynamics of real life problems. Understanding such dynamics is of critical importance for policy analysis. We analyze three well-known non-cooperative games: the Cournot oligopoly, the public goods game, and the dictator game. To link this thesis with the existing literature, we analyze these games in contexts they have been applied before such as commodity markets and climate change conferences (COPs). We use simulations and experiments as means to test the solutions provided by NCGT in each specific case.
We start by using the Cournot oligopoly to study the case of deregulated electricity markets. In a first study, we use simulations to test the effectiveness of two capacity mechanisms (i.e. mechanisms intended to stabilize capacity investments in the market) under different uncertainty levels in the market. Contrary to theoretical predictions, capacity mechanisms present substantial differences in market stability, market welfare and sensitivity to uncertainty. We found the most market oriented mechanism to be the best option overall.
In a second study, we conduct an experiment to tests the findings of the first study. We found unexpected market reactions to one of the mechanisms leading to much worse results than what the literature suggests. We found the most market oriented mechanism to be the best one once more.
In a third study, we use an experiment to test Meadows’ dynamic hypothesis for the hog market cycle (Meadows ,1970). Contrary to economic theory and previous laboratory experiments, we found strong evidence of lasting price cycles.
In a fourth study, we carry out a public goods laboratory experiment to compare two procedures used in climate negotiations (COPs), one of which has been deemed as ineffective by the NCGT literature. We found no significant difference between the two procedures in terms of promoting cooperation. We also found significantly higher contributions than theory predicted.
In the fifth article, we study whether the Dark triad framework can be a good predictor of people’s decisions in the dictator game, and whether those decisions are consistent with theoretical predictions. Using a laboratory experiment, we do not find evidence to consider the Dark triad a good predictor in this case.
This thesis contributes to a better understanding of the existing limitations in NCGT as methodology for policy analysis. This should call for further efforts to understand these limitations in particular contexts and propose solutions to them. Also, the combined methodology proposed in this thesis should serve as a motivation to improve NCGT theoretical predictions in light of dynamic complexity and realistic decision making.