WP 2015-03 Risk Perceptions in the Virtual Wilderness
ABSTRACT: In economic decision making most probabilities are formed in a compound manner through the interaction of multiple attributes of events, each of which have likelihoods that are unknown to various degrees. We consider how subjectively formed risk perceptions are affected by the dispersion of the underlying objective, compound probability distribution. Our methodology relies on virtual reality simulations of physical cues of the risk, allowing us to bring together the natural stimuli of the field and the control of the lab. Our application is an important example of a risk with serious economic consequences: the management of wild fire risk. This is an important natural setting where the risk is compound, depending on many random physical processes and where the formation of risk perceptions necessary for risk management is therefore complex. We find that increasing the dispersion of the underlying objective risk leads to higher subjective probabilities of the worst outcome occurring, consistent with increased pessimism. We compare the risk perceptions of experts in this domain with non-expert residents that are affected by the risk, and conclude that experts are not always better than non-experts at estimating the risks. Experts appear to be locked in by their strong priors based on stimuli outside those presented in our naturalistic virtual reality. With a global environment that produces increasingly extreme phenomena, training experts to be less anchored on their prior experiences will become important.