Towards a capitation formula for competing health insurers. An empirical analysis

In many countries the concept of capitating health care insurers is receiving increasing attention. The main reason is, that capitation may induce health care insurers in a competitive environment to concentrate more on cost containment. However, if the adjusters on which capitation payments are based, are too global, there may be ample room for risk selection by the insurers whilst also an unfair distribution of funds over the insurers may result, thereby undermining the objectives of capitation. The prime motivation for the present study is, that the Dutch government, as part of proposals for a new, market oriented structure of health care system, is considering to capitate insurers on the basis of global parameters like age, gender and location. Our analysis based on panel data of some 35,000 individuals, shows that the proportion of variance in annual health care expenditures that can be predicted (R2) by such a global capitation formula, is only 0.024. This is less than 1/5 of our estimate of the theoretically maximum achievable R2 which amounts to 0.138, implying the existence of abundant selection opportunities, e.g. on the basis of past expenditures or other health indicators. Alternative capitation formulae incorporating prior-year's costs and reaching about 3/5 of the maximum obtainable R2, effectively remove the profitableness of selection on the basis of past expenditures. The findings suggest, however, that selection via (chronic) health status may still be profitable to some extent. Therefore, we also analyzed data from the Dutch Health Interview Survey (N approximately 20,000) which comprised better health indicators. It appeared that a capitation formula based on the global adjusters mentioned above as well as three health status indicators and several background characteristics, yields an R2 of about 0.114, which probably accounts for 3/4 of our estimate of the maximum obtainable R2. The main conclusion is, that in the short term information on prior expenditures, which is available in the files of most insurers and thus may be used for risk selection, should be included in the capitation formula. For the more distant future, the formula should be expanded with indicators of chronic health status, possibly based on diagnostic information from previous, non-discretionary hospitalizations.