Working Paper

Dynamics in Risk Taking with a Low-Probability Hazard

Nov 8, 2016 | Andrew Royal

Summary

An experiment using financial stakes to simulate an infrequent environmental disaster reveals how people learn about and mitigate disaster risk. Subjects exhibited rationality along some dimensions of behavior and irrationality along others.

Key Findings

  • Subjects that experienced more frequent disasters were more likely to invest in risk mitigation. On average, subjects integrated information conveyed by their past experiences in a manner consistent with rational Bayesian updating.
  • Given sufficient experience, subjects’ median subjective perceptions of their disaster risk converged to their true objective risk.
  • Subjects exhibited an irrational tendency to place greater weight on experiences with disasters that directly impacted their earnings than those that did not.
  • Subjects were less likely to mitigate risks that they encountered repeatedly over multiple periods. This may be attributed to an inertia effect whereby one’s current choice is biased towards one’s previous choice made in a similar context.

Abstract

The experiment reported in this paper identifies the effect of experience on revealed risk attitudes by examining “one-shot” insurance choices made by subjects faced with a low-probability risk and their choices when they are faced with repeated exposure to an identical risk. I find that subjects engaged in greater risk taking when making repeated choices. Estimates from models that extend the standard utility and Bayesian frameworks offer evidence that subjects’ risk taking was influenced by whether previous outcomes were experienced directly and by inertia across repeated choices, though on average subjects appear to have otherwise weighted experienced outcomes in a manner consistent with rational Bayesian inference. I discuss how these and other insights apply to market behavior and policy in the presence of infrequent environmental hazards.