Japan introduced a surcharge in October 2024 on brand-name drugs when patients request them despite the availability of generics. The surcharge is waived when a physician designates the drug as medically necessary, yet little is known about how this exception is used. Using 160 million pharmacy dispensing claims, we find that such exceptions accounted for 1.7% of dispensings and were highly concentrated: among 73,090 physicians, 64.0% never used them, while the top 1% accounted for two-thirds. Drug category, specialty, institutional type, and patient composition explained little of this variation, which was stable over time and persisted even for the same patient and drug. Their concentration among a small number of physicians suggests that targeted measures may be more efficient than uniform regulation. 

Using medical and long-term care claims data from a Japanese prefecture, we examine expenditures during the last 48 months of life. Medical costs rise sharply in the final year before death; relative to age-specific average spending, they are disproportionately high among younger decedents and remain above average not only in the final year but for several years beforehand. Long-term care costs, by contrast, show no such concentration in the final year: they increase with age at death and stay persistently high over multiple years. These patterns suggest that analyses confined to the last year of life miss much of the end-of-life cost trajectory—particularly its age gradient—once medical and long-term care costs are considered jointly. 

Cost-sharing reforms can shift behavior beyond the patients who pay. Using 180 million pharmacy dispensing claims, we study Japan's 2024 surcharge on brand-name drugs with generic alternatives. Generic dispensing increased in both surcharge-eligible and exempt categories, 1.92 pp more in eligible categories. Responses extended beyond own-price incentives: generic-name prescribing increased despite physicians not bearing the charge; dispensing responses showed no monotonic gradient in surcharge amounts; and generic adoption spilled over to exempt categories (+3.85 pp), concentrated among the most-exposed patients and physicians. Targeted cost-sharing can shift defaults across stakeholders beyond its formal scope, and reporting requirements can change prescribing behavior.  

Japan has sought to contain social security expenditure growth within the increase attributable to population aging, on the view that growth from non-demographic sources should be addressed through policy. Yet the composition of those factors remains poorly identified. Combining three national statistical sources, we decompose the non-demographic increase in per capita health expenditure over 2014–2019 and 2019–2023 into incidence or utilization, duration, and unit price. Although total growth was comparable across the two periods, non-demographic factors accounted for most of the increase in 2019–2023, driven primarily by surgical procedures in inpatient care and injections in outpatient care—pointing to the diffusion of catheter-based interventions and the wider use of high-cost biologic drugs.