Introduction

Emissions from household tasks, such as driving and heating, account for approximately one third of U.S. emissions. While federal and state climate policies, including the U.S. Inflation Reduction Act (IRA), have budgeted billions in subsidies for clean technologies, macroeconomic literature has traditionally focused on firm production and innovation. This paper develops a quantitative model of household energy use to study the determinants of emissions and the design of household climate policy. The research focuses on home-production tasks where households combine equipment and energy to produce services for their own use, such as heating, cooling, and transport.

Household emissions and clean equipment: Stylised facts

The research establishes five key facts based on existing data and a new nationally representative survey of 3,362 households. Fact 1 confirms that household emissions account for roughly one third of aggregate emissions. Fact 2 notes that high-income households are more likely to use clean equipment and residential solar, particularly electric vehicles (EVs) and solar panels. Fact 3 reveals that reducing emissions is a significant motivation for many households when adopting clean technology. Fact 4 demonstrates that “environmental preferences”—the desire to reduce one’s own carbon emissions—are uncorrelated with income. Fact 5 shows that households with stronger environmental preferences are substantially more likely to adopt clean equipment, with “more-concerned” households being more than twice as likely to purchase an EV.

Model

The researchers developed a static model where households choose how much of each home service to produce, whether to use clean or dirty technologies, and whether to install solar panels. Households differ in income, environmental preferences (modelled as a distaste for their own emissions), and roof capacity for solar. A key implication of the model is that emissions reduction is a “normal good”; even though preferences are uncorrelated with income, wealthier households are better able to afford clean technologies. The model covers four primary home-services tasks: driving, heating and cooling, water heating, and cooking and cleaning. Together, these account for 89 per cent of total household energy use.

Model calibration and validation

The model was calibrated using survey data alongside the Residential Energy Consumption Survey (RECS) and the Consumer Expenditure Survey (CEX). The calibration targets included clean-technology adoption rates, equipment expenditures, and the long-run price elasticity of energy demand. The model successfully matched untargeted moments, such as the distribution of emissions across tasks and the variations in solar adoption by income. For instance, the model predicted that rooftop solar generates 5 per cent of household electricity, closely matching the 4 per cent observed in the data. The model also reproduced the positive income gradient in clean equipment adoption for all five technologies studied.

Quantitative results

The study used the model to solve for a cost-minimising subsidy policy aimed at achieving a 10 per cent reduction in household emissions. Three major findings emerged. First, residential solar is central to cost-effective abatement. It receives the highest subsidy rate (75 per cent) and accounts for 38 per cent of the total emissions reduction. Solar is highly effective because it reduces emissions across all electricity-using activities and complements technologies like EVs. Removing solar from the subsidised set would raise the cost of achieving a 10 per cent abatement target by approximately 50 per cent.

Second, the research found that restricting subsidy eligibility through income caps yields only modest cost savings. While income caps reduce payments to inframarginal high-income households, they also restrict the population that can respond to the policy. Achieving the same target with a smaller group requires higher subsidy rates, which largely offsets the savings. Comparing policies with and without caps showed that income caps only reduce the total policy cost by about 5 per cent.

Third, existing environmental preferences already account for significant reductions. Setting these preferences to zero for all households would raise emissions by 15.6 per cent. However, these preferences also increase the cost of achieving additional abatement by approximately 22 per cent, as the households with the lowest utility costs for adoption have already switched without subsidies.

Conclusion

The analysis highlights that the decision to subsidise residential solar is far more impactful on policy cost than the decision to target subsidies by income. Solar’s role is amplified by its complementarity with electricity-intensive clean equipment. Furthermore, as public concern for climate change continues to rise, the model predicts a further 1.7 per cent decline in emissions by 2033 even without new policies. This projected reduction is equivalent to the effect of a $10.60 per ton carbon tax. Ultimately, the study suggests that household preferences for the environment are a critical lever in climate economics, independent of firm-level production or market consumption choices.