Deterministic and Stochastic Prisoner's Dilemma Games: Experiments in Interdependent Security
This paper examines experiments on interdependent security prisoner's dilemma games with repeated play. By utilizing a Bayesian hierarchical model, we examine how subjects make investment decisions as a function of their previous experience and their treatment condition. Our main findings are that individuals have differing underlying propensities to invest that vary across time, are affected by both the stochastic nature of the game and even more so by an individual's ability to learn about his or her counterpart's choices. Implications for individual decisions and the likely play of a person's counterpart are discussed in detail.
We appreciate helpful discussions in designing the experiments and comments on earlier drafts of this paper by Colin Camerer, Vince Crawford, Rachel Croson, Robyn Dawes, Aureo DePaula, Geoff Heal, Charles Holt, David Krantz, Jack Ochs, Al Roth and Christian Schade. We also benefited from helpful comments from participants at the Workshop on Interdependent Security at the University of Pennsylvania (May 31-June 1 2006) and at the Workshop on Innovation and Coordination at Humboldt University (December 18-20, 2006). We thank George Abraham and Usmann Hassan for their help in organizing the data from the experiments. Support from NSF Grant CMS-0527598 and the Wharton Risk Management and Decision Processes Center is gratefully acknowledged. The views expressed herein are those of the author(s) and do not necessarily reflect the views of the National Bureau of Economic Research.