OBJECTIVES:  This study employs patient-level data to investigate the growth of health care expenditures by decomposing outpatient, inpatient and prescription drug expenditures into observable effect and structural change in a health care system that experiences the shift of payment schemes from cost-based to global budget system in 2002. METHODS: The Natioanl Health Insurance claim data which include around three-million patients was used to create a patient-level data set for the year 2000, 2005, and 2010 in Taiwan. This study employs the (recentered) influence function (RIF) unconditional quantile regression to decompose health care expenditures into observable effect and structural change. RESULTS: The payment shift results in negative structural changes on the growth of outpatient and prescription drug expenditures at the lower percentiles in the short run. In the long run, the observable effect still dominates the growth of health care expenditures at the higher percentiles. Specifically, population ageing accounts for the growth of inpatient expenditures and top prescription drug expenditures. The adoption of high technology mainly contributes to the growth of outpatient expenditures at the higher percentiles. CONCLUSIONS: Although cost control policy, like the global budget system, can lower the growth of health care expenditures in the short run, population ageing and adoption of high technology still likely drives the expenditures of high-cost patients in the long run.