A Bayesian Approach to Modeling Multivariate Multilevel Insurance Claims in the Presence of Unsettled Claims

A Bayesian model for individual insurance claims is proposed which accounts the multivariate multilevel features of claims, including multiple claimants same event, each whom may receive benefits under different coverages. approach makes it possible to account missing values in covariates and partial information contained open files, thereby avoiding sampling bias induced when unsettled are ignored. For a given claim, combination coverages payments made modeled as type with multinomial regression. The presence legal expert fees follows logistic regression type. non-zero claim amounts then log skewed normal regressions linked by Student t copula. framework yields predictive distribution paid, parameter risk process risk, while handling files. illustrated Accident Benefits car from Canadian company.