Working Paper: NBER ID: w21671
Authors: Hunt Allcott; Judd B. Kessler
Abstract: “Nudge”-style interventions are often deemed “successful” if they cause large behavior change, but they are rarely subjected to full social welfare evaluations. We combine a field experiment with a simple theoretical framework to evaluate the welfare effects of one especially policy-relevant intervention, home energy social comparison reports. In our sample, the reports increase social welfare, although traditional evaluation approaches overstate welfare gains by a factor of 3.7. Overall, the welfare gains from home energy reports might be overstated by $620 million. We develop a prediction algorithm for optimal targeting; this would double the welfare gains.
Keywords: behavioral interventions; energy efficiency; machine learning; program evaluation; randomized field experiments; smart defaults; social comparisons; welfare analysis
JEL Codes: C44; C53; D12; L94; Q41; Q48
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| energy consumption (Q41) | retail expenditures (E20) |
| home energy reports (HERS) (Q48) | discomfort from social comparisons (D63) |
| discomfort from social comparisons (D63) | social welfare (I38) |
| home energy reports (HERS) (Q48) | social welfare (I38) |
| home energy reports (HERS) (Q48) | energy consumption (Q41) |
| home energy reports (HERS) (Q48) | willingness to pay (WTP) (Q26) |