Payoff Calculator Data: An Inexpensive Window into Decision Making

Payoff calculators provide a source of information about subjects’ decision making process that is cheap, frequently available, and rarely used. We study data from an experiment designed to look at a difficult coordination problem. The experiments were *not* designed to study payoff calculator use; the payoff calculator was included as a tool for helping subjects to understand the payoffs. Our goal is to show that data about payoff calculator usage can yield useful insights about subjects’ decision making. The main issue in the game is whether players will successful coordinate, and, if so, whether they coordinate at an efficient equilibrium or a safe one. We find that initial searches using the calculator have predictive power for the total surplus and probability of coordinating for a pair in the long run. Specifically, searches consistent with the efficient equilibrium reduce total surplus and the probability of coordinating. These conclusions remain true after controlling for a pair’s initial outcomes, indicating that the data about calculator searches has predictive power beyond the pairs’ initial outcomes.