Working Paper: NBER ID: w23386
Authors: Hunt Allcott; Michael Greenstone
Abstract: We use a randomized experiment and a structural model to evaluate two home energy retrofit programs, which subsidized energy efficiency investments such as new insulation and heating systems. Two empirical findings drive the welfare analysis. First, the average energy savings were only 68 percent of the predictions provided to participants. Second, the programs’ subsidies were not closely aligned with environmental externality reductions. In our model, the inflated savings predictions and misaligned subsidies mean that the programs reduced total surplus, but with correct savings predictions, a program that aligns subsidies with externality reductions would generate positive social returns. Energy efficiency programs deliver much smaller gains than Pigouvian energy taxes, both because few households participate in the programs and because program participants still consume too much energy when energy prices are below social marginal cost.
Keywords: energy efficiency; program evaluation; randomized control trials; welfare analysis
JEL Codes: D12; L94; Q41; Q48
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| audit subsidy (M42) | audit takeup (M42) |
| informational or behavioral failures (D91) | program participation (H53) |
| unobserved benefits and costs (D61) | energy efficiency investments (G31) |
| self-selection effects (D91) | social welfare gains (I38) |
| realized energy savings (Q41) | predictions (F17) |
| perfectly calibrated subsidies (D41) | welfare (I38) |