The decision to evacuate facing uncertain hurricane landfall track, intensity, and lead time

The failure to evacuate on time (i.e., evacuation under-response) can tragically increase the cost of a hurricane; unnecessary or premature evacuation (i.e., shadow evacuation) can also make a hurricane more costly than might otherwise be necessary. Due to the constantly evolving storm characteristics over the days prior to landfall, the decision of whether and when to evacuate from an impending hurricane is a complex dynamic decision under uncertainty. Consequently, an understanding of the evolving probabilistic risk of the storm in terms of its track, intensity, and lead time is a critical aspect of the evacuation decision. This paper models these various aspects of a hurricane's risk within a dynamic model of a household's hurricane evacuation decision in order to ultimately better understand the decision to evacuate. The optimal behavior is solved using a dynamic optimization problem in which the value function is approximated as a discrete function of the state variables that convey the relevant track, intensity, and lead time risk information. Our baseline results indicate that from an economic perspective what may be deemed evacuation under-response and shadow evacuation are actually rational outcomes of the dynamic framework. These patterns are robust when sensitivity analysis of our cost of evacuation function is carried out, suggesting that the problems of evacuation under-response and shadow evacuation stem from the dynamic uncertainty that decision makers face when considering an approaching hurricane.