Hypothetical Thinking and Information Extraction in the Laboratory

Abstract

In several common-value environments (e.g., auctions or elections), players should make informational inferences from opponents' strategies under certain hypothetical events (e.g., winning the auction or being pivotal). We design a voting experiment that identifies whether subjects make these inferences and distinguishes between hypothetical thinking and information extraction. Depending on feedback, between 50 and 80 percent of subjects behave nonoptimally. More importantly, these mistakes are driven by difficulty in extracting information from hypothetical, but not from actual, events. Mistakes are robust to experience and hints, and also arise in more general settings where players have no private information. (JEL C91, D71, D72, D82, D83)