Microfoundations of Low-Frequency High-Impact Decisions

Abstract

We argue that for low-frequency high-impact strategic decisions under uncertainty decision makers define, experiment with and choose models before defining, experimenting and choosing actions. Framing decisions are choices of attributes, logical links and theories that enable decision makers to rank models by concentrating probability distributions. We develop a model that shows that decision makers with a high prior on their models do not experiment, but when the prior is lower than a threshold, they benefit from experimenting with more uncertain models in less familiar domains. As the prior on the less uncertain models declines, decision-makers are more likely to search for alternative models in less familiar domains, and either run larger scale more informative experiments (if they are not resource-constrained) or pick a decision rule that is less likely to reject the alternative decision problem (if they are resource-constrained). Our framework explains how and why decision-makers can make better strategic decisions from a wider exploration. We provide examples that illustrate the application of our framework.

Keywords: Decision problem, Experiments, exploration, framing, Strategy, Theory, uncertainty

JEL Classification: L21, L26, M13, M21