Government intervention and carbon welfare performance: Nonlinear effects and spatial spillovers from China

A central challenge in green development is not only how to reduce carbon emissions, but how to do so without weakening improvements in human well-being. This paper addresses that question through the concept of carbon welfare performance (CWP), defined as the welfare generated per unit of carbon emissions. We construct a welfare-oriented measure of CWP based on an improved Human Development Index and examine how government intervention affects it using panel data for 30 Chinese provinces from 2000 to 2023. Our empirical strategy is designed to match the economic question. Because carbon welfare performance is shaped by regional interdependence, policy diffusion, and factor mobility, and because the effect of government intervention may be nonlinear, we estimate a Spatial Durbin Model with a quadratic specification. This approach allows us to identify both local and spillover effects while testing whether the role of government changes with intervention intensity. The results reveal a robust U-shaped relationship between government intervention and CWP. At lower levels, intervention is associated with lower carbon welfare performance, consistent with resource misallocation, administrative distortion, and weak spending efficiency. Beyond a threshold, however, intervention improves CWP by correcting environmental externalities, supporting public goods provision, and facilitating low-carbon structural transformation. Mechanism analysis shows that environmental expenditure is an important transmission channel, while the level of marketization significantly conditions the strength and timing of the relationship. We also find substantial spatial spillovers, indicating that local policy choices affect neighboring regions through competition, imitation, and factor mobility. The broader implication is that the welfare consequences of decarbonization depend not simply on how much governments intervene, but on whether intervention is institutionally effective, spatially coordinated, and compatible with market adjustment.