Measuring the Welfare Effects of Residential Energy Efficiency Programs with Self-Selection into Program Participation

We introduce a framework to evaluate the welfare effects of residential energy efficiency programs and estimate key parameters using a 100,000-household field experiment. Results generally contradict conventional wisdom: there is no evidence of informational or behavioral market failures, efficiency investments entail large non-monetary costs and benefits, and realized energy savings are just 58% of engineering predictions. The programs we study reduce social welfare by $0.18 per subsidy dollar, because investment subsidies are poorly targeted to externality damages and marginal program participants are unlikely to make externality-reducing investments. Such self-selection may undermine socially desirable program expansion in this and other domains.