Understanding how consumers respond to electricity prices is essential for predicting the effects of climate change policy and other policies that target electricity markets. To date, studies of long-run electricity demand have relied on price changes that are transient or endogenous, and none have utilized experimental or quasi-experimental variation. We study the dynamics of residential electricity demand by exploiting price variation arising from a natural experiment: an Illinois community aggregation policy that enabled many communities to select new electricity suppliers on behalf of their residents. Participating communities experienced average price decreases in excess of 10 percent in the three years following adoption of a new supplier. Employing a flexible matching approach, we estimate a one-year average price elasticity of -0.14 and three-year elasticity of -0.29. We also present evidence that consumers increased usage in anticipation of the price changes, which is consistent with a dynamic model of demand. Our results demonstrate the importance of accounting for long-run adjustments when projecting policy effects.