HETEROGENEOUS TREATMENT EFFECTS OF NUDGE AND REBATE: CAUSAL MACHINE LEARNING IN A FIELD EXPERIMENT ON ELECTRICITY CONSERVATION

Abstract

AbstractThis study investigates the different impacts of monetary and nonmonetary incentives on energy‐saving behaviors using a field experiment conducted in Japan. We find that the average reduction in electricity consumption from the rebate is 4%, whereas that from the nudge is not significantly different from zero. Applying a novel machine learning method for causal inference (causal forest) to estimate heterogeneous treatment effects at the household level, we demonstrate that the nudge intervention's treatment effects generate greater heterogeneity among households. These findings suggest that selective targeting for treatment increases the policy efficiency of monetary and nonmonetary interventions.