U.S. Energy Information Administration logo
Skip to sub-navigation

Analysis & Projections

Full report

Previous edition:


Price Elasticity for Energy Use in Buildings in the United States

Release date: January 14, 2021


Within the U.S. Energy Information Administration’s (EIA) National Energy Modeling System (NEMS), the Residential Demand Module (RDM) and Commercial Demand Module (CDM) are two separate modules that are used to project energy consumption in the residential and commercial sectors. Despite similarities between these models, they differ in terms of equipment technology choice and internal accounting.1

This report explains the responses in the Annual Energy Outlook 2020 (AEO2020) version of the NEMS RDM and CDM to changes in delivered energy prices. The economic concept is known as price elasticity of demand, or the percentage change in energy consumption relative to the percentage change in prices, all other factors being equal. In general, an increase in a fuel price causes consumers to use less of that fuel or switch to a different fuel. The extent to which each of these changes takes place is useful to stakeholders in the energy sector and, particularly, in energy policy design.

The NEMS residential and commercial models can provide both short-run and long-run responses to energy price changes.2 Short-run responses to price changes are more behavioral and temporary in nature and may affect energy consumption of energy-using appliances or equipment. An example of a short-run response would be a building occupant adjusting the thermostat during cold or hot weather in response to high energy bills, leading to lower energy demand for heating or cooling. Long-run elasticities are driven by stock turnover and by technology choices for major fuels and major end uses, and they usually take place over a longer time period. For example, installing a more efficient space conditioning system or upgrading windows in a building.

To estimate responses to energy price changes, this analysis considers a series of simulations performed by adjusting the energy price paths from the AEO2020 Reference case. Only the buildings3 modules are used in these simulations; integrated effects with the rest of NEMS are not included.4 The elasticities computed from the NEMS projections reflect the results of statistical analyses of historical energy price and consumption data, along with expert judgment. The three main energy sources for the commercial and residential sectors are electricity, natural gas, and distillate. To compute the price elasticities, for each simulation, the price of one of these energy sources is doubled.


1For detailed information on both models, see U.S. Energy Information Administration (EIA), Model Documentation Report: Commercial Sector Demand Module of the National Energy Modeling System, DOE/EIA-M066; and Model Documentation Report: Residential Sector Demand Module of the National Energy Modeling System, DOE/EIA-M067. For modeling assumptions and general techniques, see EIA, Assumptions to AEO2020

2 In economics, the short run is generally defined as a period over which capital stock remains fixed. Because the typical service lifetime of installed capital can vary among economic sectors, energy end uses, and equipment types, no single definition differentiates between short run and long run.

3 Throughout the paper “buildings” or “buildings sector” refer to the residential and commercial demand sectors.

4 NEMS is an integrated, modular system where fuel prices and energy demand interact until an equilibrium is met for the entire system for each model year. In the buildings modules of NEMS, energy demand equilibrates to fuel prices from other modules. Although year-to-year changes in fuel price have the biggest effect on energy demand during the year in which they occur, the full effect of fuel price change is also spread over the following two years to prevent unrealistic demand responses to price spikes of one year or other short-term durations. Table 1 and Table 2 present elasticity results for Years 1–3, where Year 1 in the model is when the simulated price change from the Reference case was programmed to take place and Year 2 and Year 3 are the two following years (Year 1 corresponds to 2020 in the model).

See complete report.