What Are the Most Cost-Effective Policies for Reducing Greenhouse Gas Emissions?

Hahn, Hendren, Metcalfe, and Sprung-Keyser (2025) ask which climate policies generate the largest social welfare gains per dollar of government spending. They analyze evidence from 96 U.S. climate-related tax and spending policies evaluated using experimental or quasi-experimental methods over the past 25 years. Using the marginal value of public funds (MVPF) framework, they estimate each policy’s benefits relative to its fiscal cost. They find subsidies for clean electricity production, especially wind production tax credits, produce the largest welfare gains, with MVPFs often above 2 and sometimes near 6 with learning-by-doing effects. Electric-vehicle subsidies generate smaller gains (about 1.4), while fuel taxes raise revenue efficiently with MVPFs typically below 0.7.

Why This Article Was Selected for The Policy Scientist

Climate policy design has become a central policy challenge as governments attempt to reduce greenhouse gas emissions while allocating limited public resources efficiently. Hahn, Hendren, Metcalfe, and Sprung-Keyser (2025) address this broad question by comparing the welfare effects of a wide range of climate policies using a consistent framework. The authors have contributed extensively to related literatures on policy evaluation and the marginal value of public funds, making this synthesis particularly timely as climate spending expands in the United States and abroad. The study compiles evidence from 96 policies evaluated with experimental or quasi-experimental methods, providing relatively high-quality causal estimates. The empirical approach strengthens confidence in the results, though generalizability across countries will depend on institutional and energy-system differences.

Full Citation and Link to Article

Hahn, R. W., Hendren, N., Metcalfe, R. D., & Sprung-Keyser, B. (forthcoming). A welfare analysis of policies impacting climate change. American Economic Review. https://doi.org/10.3386/w32728

Central Research Question

Hahn, Hendren, Metcalfe, and Sprung-Keyser examine a fundamental policy question: which climate policies generate the greatest social welfare gains per dollar of government spending. Much of the existing policy debate evaluates climate interventions using “cost per ton of carbon abated.” The authors argue that this framework often obscures important trade-offs because it focuses narrowly on emissions reductions rather than broader welfare effects. Their study instead asks how a wide range of climate policies—subsidies, behavioral nudges, and revenue-raising instruments—compare when evaluated using a consistent welfare framework. Specifically, the paper evaluates the marginal value of public funds (MVPF) associated with 96 policies affecting greenhouse gas emissions. The central objective is to determine which policies deliver the largest net benefits to individuals relative to their fiscal cost to the government and how those rankings differ from conventional cost-per-ton metrics commonly used in environmental policy analysis.

Previous Literature

The article builds on two distinct strands of research. The first is the large environmental economics literature that evaluates climate policies using cost-effectiveness metrics such as resource cost per ton of carbon dioxide abated. This approach has been widely used in policy analyses and in influential cost-curve frameworks. However, prior research has noted that different studies often rely on inconsistent definitions of “cost,” including resource costs, government program costs, and broader social costs. These differences make comparisons across studies difficult and can lead to conflicting policy conclusions.

The second strand of literature concerns the marginal value of public funds framework developed in recent public economics research. This approach evaluates policies by comparing individuals’ willingness to pay for a policy with the net fiscal cost to the government. Previous work applying the MVPF framework has evaluated a range of social policies, including education, tax credits, and labor market interventions. The present study extends this approach to environmental policy by incorporating environmental externalities into the welfare analysis.

The paper also connects to a growing body of empirical research estimating the causal effects of environmental policies using quasi-experimental and experimental methods. Many of these studies estimate the behavioral effects of policies such as clean energy subsidies, fuel taxes, or conservation programs. The authors synthesize these estimates and translate them into welfare impacts using a consistent valuation of environmental externalities. In doing so, the paper attempts to unify a large empirical literature under a common analytical framework.

Data

The authors construct a dataset of 96 climate-related policy interventions that have been rigorously evaluated in the academic literature. The policies examined were implemented primarily in the United States and span approximately the last twenty-five years. The sample was assembled by reviewing articles published in major economics journals and supplementing these with additional studies cited in the literature. Only studies using experimental or quasi-experimental designs were included, allowing the authors to rely on credible causal estimates of policy effects.

The policies fall into three main categories. The first includes subsidies designed to encourage low-carbon technologies or behaviors. Examples include wind production tax credits, residential solar subsidies, electric vehicle subsidies, hybrid vehicle subsidies, energy efficiency rebates, and vehicle retirement programs. The second category consists of behavioral interventions such as conservation nudges that attempt to reduce energy consumption through information or social comparison mechanisms. The third category includes revenue-raising policies such as gasoline taxes, diesel taxes, jet fuel taxes, and cap-and-trade permit auctions.

To translate policy impacts into environmental externalities, the authors combine the causal estimates from these studies with additional environmental and economic data. These include emissions estimates for electricity generation and transportation fuels, projections of future grid composition, estimates of local pollution damages, and values for the social cost of carbon. The analysis also incorporates information about policy costs, subsidy rates, and government revenues. By harmonizing these inputs across policies, the authors construct comparable welfare estimates for each intervention.

Methods

The study applies the marginal value of public funds framework to climate policy evaluation. The MVPF measures the ratio of benefits to individuals relative to the net cost to the government. Benefits are measured using individuals’ willingness to pay for policy-induced changes in prices, environmental quality, and other outcomes. Government costs include both the direct fiscal cost of the policy and the fiscal effects generated by behavioral responses.

The authors extend the MVPF framework to incorporate environmental externalities such as greenhouse gas emissions and local air pollution. Reductions in emissions are valued using estimates of the social cost of carbon and other damage functions associated with local pollutants. The framework therefore converts policy-induced behavioral changes—such as increased adoption of electric vehicles or reduced electricity consumption—into monetary measures of environmental benefits.

A key methodological contribution of the paper is the incorporation of “learning-by-doing” spillovers. In several clean technology industries, production costs decline as cumulative production increases. Subsidies that expand production today may therefore lower future costs and accelerate future adoption. The authors develop a sufficient-statistics approach to quantify the welfare implications of these dynamic effects and incorporate them into the MVPF calculations.

In addition to the MVPF analysis, the study constructs three traditional cost-per-ton measures: resource cost per ton, government cost per ton, and social cost per ton. These metrics allow the authors to compare the policy rankings produced by the welfare framework with those generated by conventional cost-effectiveness measures. The analysis is conducted using both the historical context of each policy’s implementation and a harmonized baseline scenario representing a national policy implemented in the United States in 2020.

Findings/Size Effects

The analysis yields several notable findings regarding the relative performance of climate policies. First, subsidies for technologies that directly displace carbon-intensive electricity production generate the largest welfare gains. Wind production tax credits and residential solar subsidies frequently produce MVPF values above 2, implying that each dollar of government spending generates more than two dollars in benefits to individuals. When learning-by-doing effects are incorporated, these values increase substantially; the MVPF for wind production tax credits rises from approximately 3.85 to roughly 5.87.

Second, electric vehicle subsidies generate more modest welfare gains. The authors estimate MVPF values around 1.4 for EV subsidies under baseline assumptions. This implies that each dollar of government spending produces approximately $1.40 in benefits. When learning-by-doing effects in battery production are excluded, the welfare gains fall to roughly one dollar per dollar spent, similar to other consumer subsidies such as appliance rebates and weatherization programs.

Third, behavioral interventions aimed at reducing electricity consumption can produce very large welfare gains under certain conditions. Energy conservation nudges targeting regions with relatively carbon-intensive electricity grids produce MVPF values exceeding 5. These high values reflect the low fiscal cost of such programs combined with measurable reductions in electricity consumption and associated emissions. However, their effectiveness varies significantly across regions, declining in areas with cleaner electricity grids.

Fourth, taxes on polluting goods function as relatively efficient revenue-raising instruments. Gasoline, diesel, and jet fuel taxes generally produce MVPF values below 1, often around 0.7. This indicates that each dollar of revenue raised imposes approximately seventy cents in welfare costs on individuals, reflecting the corrective nature of Pigouvian taxation.

Finally, the authors compare the welfare rankings produced by the MVPF framework with those generated by cost-per-ton metrics. The results show that different cost definitions can produce substantially different policy rankings. In some cases, policies that appear expensive under government cost-per-ton measures generate substantial welfare benefits once broader effects are incorporated.

Conclusion

The study provides a comprehensive welfare analysis of climate policies by synthesizing causal evidence from a large body of empirical research. By applying the marginal value of public funds framework to environmental policy, the authors provide a unified method for comparing policies that differ widely in design, cost structure, and environmental impact.

The results indicate that policy effectiveness varies considerably across intervention types. Production subsidies for clean electricity technologies and low-cost behavioral interventions produce relatively large welfare gains, while some consumer subsidies produce more modest benefits. Revenue-raising policies such as fuel taxes impose relatively small welfare costs relative to the revenue generated.

More broadly, the paper demonstrates that evaluating climate policies solely through cost-per-ton metrics can obscure important welfare trade-offs. By incorporating behavioral responses, fiscal impacts, environmental damages, and dynamic technological effects into a single analytical framework, the study provides a more comprehensive approach to policy evaluation. The analysis also illustrates how differences in data sources, valuation assumptions, and policy contexts can influence conclusions about the relative performance of climate policies.