Adverse Selection in Multiple-Option Group Insurance. Executive Summary

摘要

The project investigated adverse selection in multiple option employment-based group health insurance by assembling a data set consisting of claims and eligibility information from 138 self-insured firms that offered only one insurance benefit and twelve large firms that introduced multiple option insurance in the study period 1984-1987. The project used four methods to estimate adverse selection: Method 1 compared the relationship between coverage and expenditures in the single option setting with that in multiple option settings; Method 2 looked at the relationship between expenses levels before multiple option were offered and plan choice; Method 3 compared expenditures levels for a given policy when it was offered in a single option setting to expenditure levels for the same policy in a dual choice setting; and Method 4 tested for the significance of a selection bias variable. Results indicated adverse selection for individual coverage but not for family coverage in Method 1. Methods 2 and 4 indicated adverse selection for both types of coverage, but results for Method 4 were highly sensitive to specification and were stronger for high option coverage. Method 3 did not indicate adverse selection. In all cases, the existence and extent of measured adverse selection varied substantially across firms. Groups did not use methods that were available to them to reduce adverse selection.