A central assumption in economics is that consumers distinguish marginal cost from fixed cost. This paper empirically tests this assumption by using a quasi-experiment in heating price reform in China. The introduction of a two-part tariff created policy-induced variation in the marginal cost and the fixed cost, and staggered policy implementation by building allows us to develop an event-study research design. Using administrative records on daily household- level heating usage over ten years, we find strong evidence that consumers distinguish marginal cost from fixed cost in a way that is consistent with standard economic theory. Consequently, the policy provided intended social welfare gains from allocative efficiency and environmental externalities. Our findings provide important implications for energy policy because a growing number of developing countries including China are in the process of implementing consumption- based energy billing in lieu of pre-existing inefficient fixed-charge billing.