Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Since 1979, the Medicare program has relied upon hospital cost function parameter estimates to develop a hospital-level and, subsequently, patient-level reimbursement system. This paper demonstrates that the resulting payment adjusters are derived from biased estimates of the impact of patient and hospital-specific characteristics on individual treatment cost. Because hospitals appear to differ in their treatment cost functions, aggregation bias is present in such hospital cost models. Given this misspecification of the aggregate cost relationship, policymakers may be wise to consider patient-level cost analyses in their design of a patient-based prospective payment system. Copyright 1992 by MIT Press.