Cry Wolf or Equivocate? Credible Forecast Guidance in a Cost-Loss Game

Abstract

Listen

Translate article

What is the most credible way to convey the risk in expert forecasts to the nonexpert decision makers who use the forecast? We test two ways to communicate this information: provide an unequivocal recommendation or equivocate by providing the probability of the uncertain event of interest. We use a simple game in which human subjects (the forecast users) decide whether to take a risk of a loss or pay a cost to avoid the risk. We find that the influence of forecast information differs depending on whether the information implies that the decision maker should take the status quo action, the preforecast optimal action, or the siren action, optimal only if the period-by-period forecast implies it. Unequivocal recommendations are more successful at inducing the status quo action. Equivocating by providing probabilities is more successful at inducing the siren action. Unequivocal recommendations to take the siren action are less effective because the false certainty makes them more vulnerable to the “cry wolf” effect. A forecast that combines probabilities with recommendations captures most of the gains of the separate approaches. A follow-up study examines whether an explanation of how risk quantities inform recommendations can substitute for providing the risk quantities in the forecast. The study finds little evidence in support of this proposition. Data are available at https://doi.org/10.1287/mnsc.2016.2645 . This paper was accepted by Yuval Rottenstreich, judgment and decision making.