Preference for Insuring Against Probable Small Losses: Insurance Implications

A series of laboratory studies of insurance decision making shows that people buy more insurance against events having a moderately high probability of inflicting a relatively small loss than against low-probability, high-loss events. Two explanations are discussed, both contrary to traditional utility theory. One postulates a utility function convex over losses. The second asserts that people refuse to protect themselves against losses whose probability is below some threshold. This research provides insight into other, often puzzling, facts about people's insurance behavior. Relevance for public policy is discussed. What determines whether people will act to protect themselves from the severe consequences arising from a low probability event? The answer to this question is vital for understanding how people cope with the risks from accidents, diseases, and natural hazards and for helping them to manage their lives more effectively in the face of such risks. The importance of this question has motivated a major research project concerned with one class of risks, those from natural hazards (floods and earthquakes), and one class of protective mechanisms, insurance.' The project staff, under the direction of Howard Kunreuther, interviewed 2,000 homeowners residing in flood-prone areas, half of whom were insured and half uninsured. In addition, 1,000 homeowners residing in earthquakePaul Slovic, Ph.D., Baruch Fischhoff, Ph.D., and Sarah Lichtenstein, Ph.D., are Research Associates; Bernard G. Corrigan, M.A., is Research Programmer/Analyst; and Barbara J. Combs, M.A., is Research Assistant, all with Decision Research in Eugene, Oregon. Slovic, Fischhoff, and Lichtenstein have worked together for a number of years on psychological studies of judgment, risk taking, and decision processes. They have recently reviewed this field for the Annual Review of Psychology. Slovic and Fischhoff are consultants to the National Research Council's Committee on SocioEconomic Effects of Earthquake Predictions. Slovic is a member of the socio-political resource group of the National Academy of Sciences Study of Nuclear and Alternative Energy Systems and is Adjunct Professor, Department of Psychology, University of Oregon. 1 This research was supported by NSF-RANN Grant ATA 73-03064-A03 to the Wharton School of the University of Pennsylvania under subcontract to the Oregon Research Institute. We thank Howard Kunreuther, Daniel Kahneman, and Amos Tversky for their invaluable contributions to this project. Frank Conklin supplied us with a prototype for the farm game. Comments and reprint requests may be addressed to the authors at Decision Research, 1201 Oak Street, Eugene, Oregon 97401.