Method: Dif-In-Dif and Two-way Fixed Effects Panel Regression.

Data:  Electricity consumption data at the feeder-hour level and 5 million SMS data. 

Codes: GitHub

Can Behavioral interventions shift the electrical load from on-peak to off-peak hours? In the summer of 2019, Tehran Electricity Distribution Company sent about 5 million SMS messages to some residential and commercial subscribers to warn them about a power outage and its possible damages. In the literature of behavioral interventions, this intervention is considered a combination of information feedback and cultivation. We intend to study the effectiveness of this intervention on electrical peak load, using electricity consumption data at the feeder-hour level and SMS data. The causal identification method of the paper is panel regression with two-factor fixed effects. Also, to test the robustness of the results, we use matching and difference-in-difference methods. We find that the reduction of per capita electricity consumption of subscribers due to the intervention is not robust. We come to this conclusion because the results of the main research method confirm the significance of the effect only in certain conditions, and also, all of the robustness checks show the insignificance of the intervention. The secondary results of the research also indicate that there was no long-term effect of the intervention and no neighborhood spillover effect.