Inferring capitation rates from aggregate health plans' costs

Setting risk-adjusted capitation rates in health systems with centralized financing and decentralized delivery is one of the most intriguing policy issues. The common practice to set capitation group rates is based on individual data collected from either population surveys or medical records, using a single - and in most cases arbitrary - set of relative unit costs of services. This paper presents a method for estimating group-specific mean costs and capitation rates using a panel of aggregate cost data of the competing health plans and the composition of their populations. This method is used to estimate mean costs and capitation rates for the Israeli health care system. The limited data available severely constrains the range of estimable models, however, the results evoke some questions with regards to reimbursement and rates presently used, as well as to the methodology used to estimate them.