For purposes of thinking about demand-side cost-sharing, the existing evidence on spillovers has two main weaknesses. First, it is focused mainly on how the spread of managed care affects practice patterns for those not in the managed care program. But while managed care often has different demand-side cost-sharing, it also has very different supply-side payment incentives. Furthermore, while managed care plans had substantially lower average cost-sharing than fee-for-service plans in the 1990s, as Table 1 shows, in recent years average cost-sharing in fee-for-service plans has dropped (perhaps reflecting in part a different mix of patients remaining in the smaller fee-for-service pool). Second, the existing evidence focuses primarily on short-term spillovers, while the largest spillovers may arise in the long run, driven by changes in system infrastructure and capacity, as well as in workforce composition.