One, Two, (Three), Infinity, … : Newspaper and Lab Beauty-Contest Experiments

Abstract

Acknowledgments: We wish to thank Ernan Haruvy for his comments on a preliminary version. We acknowledge financial support from the Spanish Ministry of Education through grants SEC98-1853-CE and DGES PB96-0300, and the EU-TMR Research Network ENDEAR (FMRX-CT98-0238). We thank the Spanish newspaper Expansión and the German magazine Spektrum der Wissenschaft for letting us use their platforms to run our experiments, Richard Thaler for giving us his data from the Financial Times experiment, and Gary Charness, Sjaak Hurkens and Bettina Rockenbach, for running an experiment in their classes.1 Abstract "Beauty-contest" is a game in which participants have to choose, typically, a number in [0,100], the winner being the person whose number is closest to a proportion of the average of all chosen numbers. We describe and analyze Beauty-contest experiments run in newspapers in UK, Spain, and Germany and find stable patterns of behavior across them, despite the uncontrollability of these experiments. These results are then compared with lab experiments involving undergraduates and game theorists as subjects, in what must be one of the largest empirical corroborations of interactive behavior ever tried. We claim that all observed behavior, across a wide variety of treatments and subject pools, can be interpreted as iterative reasoning. Level-1 reasoning, Level-2 reasoning and Level-3 reasoning are commonly observed in all the samples, while the equilibrium choice (Level-Maximum reasoning) is only prominently chosen by newspaper readers and theorists. The results show the empirical power of experiments run with large subject-pools, and open the door for more experimental work performed on the rich platform offered by newspapers and magazines.J.E.L. classification codes: C7, C9

On the persistence of strategic sophistication

“…The observed distributions of guesses show clear spikes consistent with Level-1 and Level-2 play. This finding is robust to the structure of the game [31] and varied populations [10]. 6 Stahl & Wilson [62] study Level-k behavior in ten 3×3 matrix games.…”

“…Many papers find strong support for Level-k play in certain games using behavioral data alone [62,63,54,31,42,10,18] or behavioral data augmented with lookup data [24,23] or eye-tracking data [19,65]. For some games, however, the Level-k model does not appear to organize the data well [43,44,26].…”

Incentives for subjects in internet experiments

“…Table 1 provides a summary of the four experimental treatments. 2 The bulk of the experiment was conducted as an internet experiment (setting net). In net, subjects played on the internet, in a location of their own choice (home, office etc.…”

Who Thinks about the Competition? Managerial Ability and Strategic Entry in US Local Telephone Markets

“…First, although journalists like to play up unobservable characteristics such as charisma and leadership as driving CEO success, the 1 Camerer and Johnson (2004) track how long subjects looked at competitor payoffs and find that measured strategic ability is positively correlated with time spent looking at competitor payoffs. Bosch-Domenech et al (2002) ask subjects in a beauty contest game to explain their choices and find that people explain their actions with logic based on thinking steps. 2 Collard-Wexler (2008) discusses the realism of Bayesian Nash equilibrium in applied work and argues that a lack of industry norms and experience, combined with simultaneity, provide an ideal platform for adding behavioral assumptions.…”