(Photo by sasirin pamai on Shutterstock)
In A Nutshell
- Heat pumps showed nearly four times lower heating energy use intensity than gas furnaces.
- Winter thermostat settings strongly influenced household energy demand.
- More appliances and frequent use raised energy use in many states.
- Apartments used less energy per square foot than detached single-family homes.
- Only 3.5% of U.S. homes reported solar power use, showing adoption remains rare.
UNIVERSITY PARK, Pa. — Switching from gas to electric heating could be one of the most effective steps American homeowners can take to cut energy consumption, according to a national analysis of nearly 18,500 homes across all 50 states.
Researchers at Penn State University found that households using electricity for space and water heating consumed far less energy than those relying on natural gas. Electric heat pumps stood out in particular, with modeled heating energy use intensity nearly four times lower than that of traditional gas furnaces. For the average U.S. home, this difference indicates significant savings potential on utility bills.
Published in the journal Energy Policy, the study applied machine learning to the 2020 Residential Energy Consumption Survey, the most detailed dataset available on household energy use. The findings go beyond heating systems, identifying other practical choices that homeowners and renters can adjust to cut costs.
Why Heat Pumps Outperform Gas Furnaces
Electric heat pumps emerged as the clear winner for heating efficiency. Homes equipped with central heat pumps showed an average modeled heating energy use of 0.0237 megawatt-hours per square meter, compared to 0.0918 for homes with gas-operated central furnaces.
Heat pumps transfer heat rather than generating it, which explains their superior efficiency. According to the U.S. Department of Energy, this approach can reduce electricity use for heating by around 50 percent compared to electric resistance heating like furnaces and baseboard heaters.
Gas heating systems lose energy through combustion inefficiencies, with a portion of fuel energy escaping as waste heat or through incomplete burning. Electric systems avoid these combustion losses on-site. Homes using electric water heaters also consumed less energy than those using propane, gas, or oil systems, with oil water heaters showing roughly double the energy intensity of electric models.
With natural gas still the most common heating fuel nationwide, and electricity second, the findings suggest millions of households could reduce their energy use by switching to heat pumps.
Turn Down the Thermostat to See Real Savings
Occupant behavior matters more than many realize. Winter thermostat settings ranked among the top five energy factors in states including Colorado, Connecticut, Maine, Massachusetts, New York, and South Dakota. Higher temperature settings during winter consistently linked to increased energy consumption.
Lowering the thermostat setpoint during the heating season and raising it during cooling season directly influences energy consumption. Summer cooling showed less impact on overall energy use than winter heating, reflecting the dominance of heating demands in American homes.
This finding carries particular weight because older adults, who tend to prefer warmer indoor temperatures due to lower metabolic rates and reduced ability to retain heat, are driving up energy use as the U.S. population ages.
Upgrading Appliances Pays Off
The study also identified everyday appliances as a major factor. More refrigerators and televisions, paired with frequent clothes dryer and dishwasher use, raised household energy use in 39 percent of states.
While these patterns reflect lifestyle choices, appliance standards and energy labels make it easier to pick efficient models. Replacing older equipment with efficient options when it wears out offers a simple way to reduce demand without changing routines.
Apartment Living Cuts Energy Costs
Housing type also played a role. Detached single-family homes averaged 0.169 megawatt-hours per square meter of energy use, compared with 0.126 for apartments in buildings with five or more units.
Apartments benefit from shared walls, floors, and ceilings, which insulate against both heat loss in winter and heat gain in summer. Detached homes expose more surface area to the outdoors, requiring more energy to maintain comfortable temperatures year-round.
Owned homes consumed an average of 25.95 megawatt-hours annually, while rented homes used 14.84. However, when adjusted for home size, the gap narrowed considerably. Owned homes showed energy use intensity of 0.15 megawatt-hours per square meter versus 0.14 for rentals, suggesting homeowners’ higher energy use stems largely from bigger houses rather than wasteful behavior.
Solar Panels and Other Cost Cutters
Homes generating electricity on-site with solar panels drew less power from the grid. Still, only 3.48 percent of U.S. residences reported solar generation, showing adoption remains rare.
Homes generating electricity on-site through solar panels consumed less energy from the grid. Yet only 3.48 percent of U.S. residences reported using solar electricity, showing adoption remains rare. Wider use of rooftop solar could help households manage energy costs while reducing reliance on centralized grids.
Climate remained one of the strongest predictors of energy consumption. Heating degree days, a measure of how cold a location gets during winter, along with total energy-consuming square footage, number of rooms, and household size all increased consumption as expected.
The researchers highlight several practical steps:
- Lower thermostats in winter and raise them in summer for immediate savings.
- Choose efficient appliances when replacing older ones.
- Install heat pumps in new homes or major retrofits for the largest reductions.
State and federal policies could accelerate adoption through incentives for heat pump installation, support for on-site solar generation, and stricter efficiency standards for appliances. The study analyzed data representing roughly 123.5 million American homes, making its conclusions broadly applicable across diverse climates and housing types.
As electricity infrastructure improves and more households shift toward electric heating, grid capacity will need strengthening to handle increased demand. But electrification emerged as the strongest driver of reduced household energy use in this national analysis.
Disclaimer: This article and guide is for general information only and not a substitute for professional energy or financial advice.
Paper Summary
Methodology
Researchers analyzed the 2020 Residential Energy Consumption Survey microdata from the U.S. Energy Information Administration, which included responses from 18,496 households representing approximately 123.5 million American homes. The team initially identified 23 influential features from previous studies, then expanded to 75 potential features using expert judgment and the enhanced 2020 dataset. Through an iterative process of machine learning modeling and sensitivity analysis, they refined the feature set to 41 variables that maximized predictive performance. Data underwent preprocessing including one-hot encoding for categorical variables, standardization of numerical features, and removal of outliers using K-means clustering. Four tree-based machine learning algorithms were tested, with CatBoost performing best. The team incorporated statistical weights to ensure household-level data accurately represented broader population trends at state and national levels.
Results
At the national level, the CatBoost model achieved an R-squared value of 0.71, explaining 71 percent of variance in residential energy consumption. The five most influential features were electricity for space heating (2.35 MWh impact), electricity for water heating (1.67 MWh), heating degree days (1.63 MWh), energy-consuming square footage (1.58 MWh), and total number of rooms (1.27 MWh), compared to a baseline of 23.173 MWh. Using electricity for heating significantly reduced consumption, while gas heating increased it. State-level models showed strong to moderate fit in 87 percent of states. Key determinants at the state level included energy-consuming area (98 percent of states), heating fuel type (90 percent), number of rooms (65 percent), housing type (41 percent), appliance number and usage (39 percent), household size (35 percent), thermostat settings (12 percent), and on-site solar generation (10 percent). Electric heat pumps demonstrated heating energy use intensity nearly four times lower than gas furnaces.
Limitations
Measurements reflected site energy consumption rather than source energy, which does not account for generation, transmission, and distribution losses. Weak model performance in 13 percent of states suggests additional features beyond those in the RECS dataset may be needed to fully capture energy consumption patterns in those regions. The research used a single year of data rather than analyzing temporal variations across multiple RECS datasets. The analysis did not examine potential challenges of widespread electrification such as new peak loads on the power grid, nor did it assess carbon emissions reductions or air quality improvements from transitioning to electric systems. The researchers acknowledged that achieving meaningful energy savings requires addressing socioeconomic disparities and preventing rebound effects where efficiency improvements lead to increased consumption.
Funding and Disclosures
The researchers acknowledged support from Penn State’s Hamer Center for Community Design and its Resource and Energy Efficiency Lab, the Stuckeman Center for Design Computing, and the Institute of Energy and the Environment. The authors declared no known competing financial interests or personal relationships that could have influenced the work. During manuscript preparation, the authors used ChatGPT to improve readability and language, with all content reviewed and edited by the authors who take full responsibility for the publication.
Citation
Korsavi, S.S., Azari, R., Iulo, L.D., & Mahdavi, M. “Determinants of U.S. residential energy consumption at national and state levels: Policy implications,” was published in Energy Policy, Volume 202, July 2025.