By using customer‐level residential billing data from 2008 to 2010 of a major utility company in Phoenix metropolitan area, this study adopts a matching approach and a difference‐in‐differences method to estimate empirically the impact of a prepaid electricity plan on residential electricity consumption, after correcting for selection bias. Results show that the prepaid program is associated with a 12% reduction in electricity usage, customers with lower level of wealth or those with higher amount of arrearage prior to switching to the prepaid program tend to save more electricity after switching, and prepaid customers save more electricity in the summer than winter. (JEL L94, Q41)