Two-person bargaining experiments with incomplete information (Q1329543): Difference between revisions

This book is a report of the experiments performed at the Bonn Laboratory of Experimental Economics. All the chapters are about one simple bargaining problem containing only a `whiff' of uncertainty. The restriction to one bargaining problem and the simplicity of the model are the strength of this writing. Let us start with a description of the bargaining problem at hand. Two players bargain about the division of one hundred points (approximately 12 DM). With 50\% chance each of the players has a high reservation value (an amount they obtain if the negotiations break off) or a low reservation value \((=0)\). The players only know their own reservation value and the a priori given 50/50 chance distribution. During the bargaining process the player in turn has three options: to accept the last offer, to reject the last offer and make a new one or to leave the negotiation table. If the negotiations are broken off, the players obtain their (discounted) reservation value. If an offer is accepted, the players obtain the (also dicounted) payoff they agreed upon. If a new offer is proposed, it is the turn of the other player to make a move. In the experiments the high reservation value \(\alpha\) is taken between 30 and 70. The book consists of three parts. The first part gives a game theoretical analysis of the problem at hand. It describes what the agents are supposed to do (according to the theory); it forms the background with which the actual behavior of the negotiators in the experiments is compared. In the second part one finds an account of experiments done by students unexperienced in game theoretical analysis. They are matched with each other randomly and obtain their type according to the 50/50 distribution. Then they bargain according to the forementioned rules. They cannot communicate during the bargaining process in an other way than by accepting an offer, proposing an new offer or breaking off. In total 1440 plays of the game are observed, equally distributed over the five different values of \(\alpha\), the type and first mover/second mover combinations. The results and an analysis of the results are given in the chapters 6-8 of the manuscript. In the third part of the book more experienced students are asked to write a program for a strategy, i.e. a program that gives an answer what to do in every situation that can occur during the negotiations. These (31) programs are used to simulate the bargaining process (the programs `play against each other') and the outcome of all combinations of programs are collected and analyzed. Finally, in an interesting chapter 14, one finds a repeated game (or evolutionary) approach. It is investigated how individuals with a given strategy survive if the payoffs in earlier rounds increase or decrease the fitness (or relative frequency) of a strategy. At the start of the simulation experiment four populations (High/low reservation value, first/second move) are defined, each containing initially a same number of individuals with each of the 31 strategies. Members of the first-move populations are randomly matched with members of the second-move populations. The payoff in each round of the process determines the frequency of a strategy in the next generation of the population. If a strategy is doing well in some round of the game, his relative frequency will increase in the next generation, if will decrease, if it performs badly. By organizing such a tournament it is possible to find out which strategies `die out', which strategies `survive' and what are the stable distributions over the strategies (if any) in the long run. In my opinion this book and the experiments of the Bonn Laboratory of Experimental Economics will be a benchmark for other experiments in economical behavior in the next decade. The restriction to exactly one bargaining experiment and the thorough analysis of the results makes this book to a `must' for everybody interested in experimental economics.