Designing cost-effective climate policies through preference prediction: Evidence from Chinese households’ low-carbon alternatives

Designing cost-effective climate policies requires understanding not only who accepts low-carbon alternatives but also heterogeneous household preferences. Using a unique household survey from China, this study examines household willingness to accept (WTA) low-carbon alternatives and evaluates the potential efficiency gains from predicting these preferences. The results suggest that demographic characteristics such as location, gender, education, and income are associated with systematic differences in WTA. Knowledge of carbon neutrality and related policies is correlated with WTA in a non-linear manner. Households with basic awareness tend to report lower compensation requirements, while those with greater familiarity report higher stated WTA, possibly reflecting skepticism about the costs and perceived fairness of low-carbon transitions. To translate these insights into policy design, we compare econometric and machine-learning approaches to predicting household preferences. Compared with random assignments, predictive models substantially increase acceptance rates while reducing average compensation. However, the simulation results indicate that targeted investments in public education may generate long-term efficiency gains comparable to or even exceeding those from predictive precision alone