This paper builds an environmental belief predictor using the households’ sociodemographics to explore the relationship of environmental belief and residential energy expenditure using GSS and AHS data. The analysis starts by showing the stable and different environment perceptions across households’ characteristics which may suggest different sociodemographics contribute to the various environmental beliefs. Then I use a logit modeling strategy to select the predictors and match the availability in AHS data to predict the corresponding environmental belief of those households and using the predicted probabilities to explore the cross-individual and cross-time variations in residential energy expenditure by using quantile regression and fixed effects model respectively. In order to cope with the discrepancies between two different datasets, the Lewbel IV approach is implemented to mitigate the measurement error. The results from the analysis suggest that environmental belief affects the residential energy expenditure both in cross-individual and cross-time analyses, which implicate that people who are more environmentally concerned reduce their own energy consumption to protect the environment.
