The energy-saving effects of home retrofit programmes in Canada

This study applies recent advances in treatment-effects estimation methods suitable for observational data without panel structure to assess the energy-saving impacts of home energy-efficiency retrofits. The existing literature on the energy-saving effects of retrofits primarily employs variations of the difference-in-differences approach with exact matching, controlling for treatment status with dummy variables. However, because self-selection in retrofit decisions is common, it is important to condition retrofit probabilities on various household and house characteristics to reduce bias in the estimated treatment effects. By combining retrofit status data from a Canadian national survey with actual household energy consumption data, this study first estimates the retrofit probability as a function of household and house attributes. We then use the predicted retrofit probability and apply treatment-effect estimation techniques to determine the energy-saving effects of residential retrofitting programs. Our results indicate that the average natural gas savings from various retrofit incentive programs in Canada are approximately 7%. Our estimates remain consistent across different estimation methods and model specifications tested in the sensitivity analysis.