Insurance for low-probability hazards: A bimodal response to unlikely events

Two insurance experiments using real-money consequences and multiple rounds to provide experience are described. In the first experiment, subjects bid for insurance to prevent a fixed loss of 4 at probabilities ranging from .01 to .9. Mean bids were near expected value except at the lowest probability of.01, for which a very bimodal distribution was observed (some subjects bid zero and others bid much more than expected value). A second experiment explored this bimodality at a probability of .01 with loss increased to4 at probabilities ranging from .01 to .9. Mean bids were near expected value except at the lowest probability of.01, for which a very bimodal distribution was observed (some subjects bid zero and others bid much more than expected value). A second experiment explored this bimodality at a probability of .01 with loss increased to 40. A similar bimodal distribution was obtained that persisted over 50 rounds of experience. These laboratory results are consistent with field evidence for low-probability hazards, for which people appear either to dismiss the risks or to worry too much about them.We would like to thank Alan Carlin, Ann Fisher, Risa Palm, and David Brookshire for their helpful comments on an earlier version of this article, and Rebecca Boyce, Julie Irwin, Glenn Russell, and Joy Smith for research assistance. We also gratefully acknowledge support from the University of Colorado Council on Research and Creative Work for human subject payments and from the U.S. Environmental Protection Agency, Office of Policy, Planning and Evaluation for support for research design and analysis provided as part of Cooperative Agreement #CR812054. All errors, opinions and conclusions are the sole responsibility of the authors.