This book focuses on a simple public good game: each player (agent) decides on contributing his endowment to the provision of a public good; only if sufficiently many players contribute the public good will be provided, in which case every player profits; if a player decides to contribute he will loose his endowment also if the good is not provided. The aim of the book is to describe experimental tests of this game, and the resulting conclusions.
Chapter 2 of the book reviews some theories that apply to the public good problem. Psychological theories distinguish between cooperative and individualistic agents. Economic theories focus on expected value or utility maximization, but also on alternative approaches such as prospect theory. Game theory considers Nash equilibria or, in case of heterogeneous players and incomplete information, Bayesian Nash equilibria. In order to take the possibility of error into account also the so called quantal response equilibrium is discussed, characterized by a “rationality level” parameter. Finally, learning models are introduced, and some theoretical results for this particular game are derived. A distinction is made between belief learning (such as already fictitious play) and reinforcement learning. The latter is based on an evolutionary idea, such as replicator dynamics.
Chapter 3 introduces basic experimental tools, specifically, to test the personal characteristics (individualistic or cooperative) of individuals participating in the public good game, before they actually participate in that game. Chapters 4 and 5 then describe the experiments and resulting conclusions with the focus on decision rules and on beliefs and learning, respectively. Some conclusions are that individual characteristics play an important role in the results, and that there is more support for “naive” Bayesian updating than for mere static Nash equilibrium.
Chapter 6 concludes the book with an attempt towards a positive theory of public good games, based on the obtained experimental results.