Transient and persistent energy efficiency in the US residential sector: evidence from household-level data

In this paper, we measure the energy efficiency implicit in residential energy consumption using a panel dataset comprised of 40,246 observations from US households observed over 1997-2009. We fit a stochastic frontier model of the minimum input of energy needed to meet the level of energy services demanded by the household. This benchmarking exercise produces a transient and a persistent efficiency index for each household and each time period. We estimate that the US residential sector could save approximately 10% of its total energy consumption if it reduced persistent inefficiencies and 17% if it were possible to eliminate transient inefficiencies. These figures are in line with recent economy-wide assessments for the USA. Our results suggest that savings in energy use and associated emissions of greenhouse gases may benefit from both policy measures that attain short-run behavioral changes (e.g., nudges, social norms, display of real-time information about usage, and real-time pricing) as well measures aimed at the long run, such as energy-efficiency regulations, incentives on the purchase of high-efficiency equipment, and incentives towards a change of habits in the use of the equipment.