A Risk Sharing Rule for Peer-to-peer (P2P) Insurance for Hail Losses via Conditional Expectation

The fundamental concept is deriving a conditional mean risk sharing rule, i.e., assigning the individual loss to be the expectation given the total losses among all insurers. Consider there are n participants to an insurance pool with index i = 1, . . . , n. Each of them faces a loss Di, which is a non-negative random variable with (spatial) dependence. Recent literature showed that the conditional expectation of individual loss given the total among n participants can serve as a risk sharing rule with many attractive properties. However, the results are established under independence, which is violated in many real applications, such as damages due to natural events. Moreover, existing work simply assumed that the underlying model is known, which is unrealistic and limits the implementation. To overcome those critical limitations, we aim to extend the results under spatial dependence with heterogeneity when the model is unknown and not fully specified.

Journal article