Valuing Insurance Against Small Probability Risks a Meta-Analysis

The demand for non-mandatory insurance against low-probability, high-impact risks is relatively low compared to what is expected. From a dataset of 65 experimentally-elicited and survey-based estimates, we conduct a meta-analysis of contingent valuation studies. It appears that the average willingness to pay for insurance is about 87% of the expected losses. We use a meta-regression analysis to measure the extent to which specific moderators contribute to explaining heterogeneity across studies. While our results provide no direct support for hypothetical bias, the meta-regression shows that laboratory-based studies are associated with higher relative willingness to pay. Moderators such as information about loss probability, small probability levels, within subject design and show up fees positively influence relative willingness to pay, whereas respondents’ average income and age have a negative effect. Similarly, cultural sub-factors related to power distance and uncertainty avoidance seems to provide additional explanations on differences in insurance willingness to pay across countries.Our results, robust to model specification and publication bias, provide some recommendations for future research on stated preferences