Performance pay under capitation in rural China: an artefactual field experiment with primary care physicians

Performance pay may address underprovision under capitation, but physician responses can depend on the incentive structure and clinical context. Cross-setting evidence from controlled experiments remains limited. We examined whether adding performance pay to capitation changes service provision among practicing primary care physicians in rural China and whether responses differ by bonus level or patient illness severity. We conducted an artefactual field experiment involving 118 primary care physicians in Xintai, Shandong Province, China. Each physician chose service quantities for nine standardized patient profiles, first under capitation and subsequently under capitation plus performance pay. Physicians were randomized to low or high bonus schedules designed to increase the maximum net experimental payoff by 5% or 20%, respectively. The primary outcome was service quantity, analyzed using physician and patient-profile fixed-effects models. We examined bonus-level and severity interactions, adjusted for multiple testing, and assessed experimentally defined patient benefit and gross provider payment descriptively. Adding performance pay increased service provision by 0.298 units (SE = 0.080; p < 0.01) in the primary fixed-effects model. The interaction between performance pay and high illness severity was positive (β = 0.350; unadjusted p = 0.025), but did not remain statistically significant after Holm adjustment (adjusted p = 0.098). The 20% bonus produced no statistically detectable additional response relative to 5%. In exploratory change-score analyses, no substantive coefficient remained statistically significant after correction for multiple comparisons. Average gross provider payment increased by 1.73 and 3.13 experimental currency units in the low and high bonus conditions, respectively, while patient benefit increased by 0.58 and 0.44 units. The resulting incremental gross-payment-to-patient-benefit ratios were 2.98 and 7.11, respectively; these descriptive ratios do not represent formal cost-effectiveness estimates. Adding performance pay to capitation increased service provision among rural Chinese primary care physicians. The estimated severity-related heterogeneity was suggestive but not confirmed after multiplicity adjustment, and nonsignificant differences between bonus conditions should not be interpreted as equivalence. These findings complement German experimental evidence but require validation in routine care using clinical outcomes and cost data.