Fix Modeling and Cost-Benefit Analysis

The federal government’s core economic models—used to forecast growth, estimate budget impacts, and evaluate policy—systematically mishandle one of the most powerful forces reshaping the economy: climate change and the transition to clean energy. While these models are often outdated and underfunded, the root of the problem is that policymakers and the modelers they rely on face strong incentives to produce and use analyses that appear comprehensive and defensible, regardless of whether the underlying frameworks actually support that appearance.

This institutional bias leads to two predictable failures. The first is a failure of omission. The macroeconomic frameworks developed in the early- to mid-20th century by economists such as Simon Kuznets, Richard Stone, and James Meade were built to track short-term fluctuations and manage the business cycle. Adding deeply uncertain and potentially transformational economic effects risks visibly contaminating the credibility of those models. So those effects get left out entirely.

In other contexts, the failure runs in the opposite direction: The pressure to appear comprehensive leads to falsely precise claims that appear to settle contested questions rather than usefully inform debate. The effects of climate change and the energy transition are of a scale, complexity, and novelty that frequently elude meaningful quantification, yet the standard cost-benefit framework is routinely applied to them as if they do not.

Correcting these failures requires three complementary reforms. The first is to replace a culture of false comprehensiveness with greater transparency about model limitations, deep uncertainties, and the value-laden judgments often embedded invisibly in technical assumptions. The second reform is to improve the underlying models in situations where better quantification is genuinely achievable, narrowing the range of effects that must be handled qualitatively. The third reform is to incorporate those improved models into federal macroeconomic frameworks so that climate risk and energy transition dynamics are reflected in the baseline economic picture that informs federal decision-making.

The Omission Failure

Failing to incorporate climate change and the transition to clean energy into macroeconomic projections distorts the overall picture of the economy. Policymakers relying on these projections are therefore making decisions based on incomplete and potentially misleading information.

Two types of climate-related economic effects are important to incorporate into federal modeling exercises. First are the physical effects of climate change. These include the destruction of infrastructure from hurricanes and other extreme weather events, as well as more slow-moving implications for labor productivity, agricultural yields, and capital stock from hotter temperatures and more volatile weather.¹ While it is difficult to parse the causes of any given event, in 2024 alone the United States experienced 27 weather and climate disasters causing more than $1 billion in damages each.²

Second are the economic effects associated with the transition to a lower-carbon economy. The transition will require reconfiguring how energy is generated, how goods are manufactured, and how people and products move through the economy. This will have direct consequences for investment, employment, trade patterns, supply chains, and geoeconomic power.³ While some of these effects may be modest in isolation, their interaction and simultaneity creates the risk of compounding into unanticipated shocks that are larger than the sum of their parts.

Macroeconomic tools often miss these shocks. For example, the rapid transition away from coal-fired power generation has devastated local Appalachian economies. The collapse of coal employment has led to not only job losses but also losses in tax revenues that fund schools, infrastructure, and other public services. Yet most models lack the geographic granularity to capture these localized economic shocks.⁴

Simultaneously, climate risk is already reshaping the insurance sector, with implications for household wealth and financial stability. As insurers withdraw from high-risk markets or raise premiums sharply, homeowners in affected areas face a stark choice between absorbing higher costs or going uninsured. Either outcome depresses property values, eroding the primary store of wealth for many American families. Falling property values in turn shrink local tax bases, threatening the public services and fiscal stability of communities that are often already economically vulnerable.⁵

How to Incorporate Climate into Federal Macroeconomic Frameworks

The Council of Economic Advisers, Office of Management and Budget, and Treasury jointly produce macroeconomic assumptions that underpin the President’s Budget. These include projections of GDP growth, interest rates, and employment that policymakers rely on to understand the trajectory of household wealth, financial stability, and local public finance. These projections provide a common baseline for estimating revenues and expenditures, shaping decisions across the federal government.

These macroeconomic tools were not designed to capture large-scale structural transformations. They generally treat the composition of the economy—its industries, technologies, and energy systems—as fixed in the near term, assuming that the economic relationships observed in the past will continue. This assumption is always limiting but is especially problematic in the face of major structural economic shifts, including climate change and the transition to a net-zero-emissions economy, both of which are already beginning to reshape patterns of investment, production, and trade.⁶

Consider the automobile industry: A shift from internal combustion engine vehicles to electric vehicles could fundamentally restructure the auto manufacturing industry and its extensive domestic supply chain, particularly if foreign manufacturers capture market share that domestic producers currently hold. The resulting losses in manufacturing employment, supplier revenues, and regional tax bases would not show up in baseline projections that assume the industry’s economic footprint evolves as it has historically—a miscalibration that compounds across a 10-year projection window into significant misdirection for budget and policy planning.

Redesigning these tools requires both technical upgrades and a more fundamental shift in how analysts and policymakers think about what models can and cannot do. A growing body of work has begun to do this. The International Monetary Fund, World Bank, and Network for Greening the Financial System have developed approaches to incorporate physical and transition risks into macroeconomic models. Several governments, including Denmark, France, and the United Kingdom, have begun integrating these risks into their official economic projections. In the United States, the Congressional Budget Office has taken initial steps to incorporate the effects of changing temperature, precipitation, and hurricane patterns into its long-term outlook.⁷

The White House should learn from these approaches rather than start from scratch. But it should also recognize that these newly developed models should be understood as early steps rather than mature methodologies. The underlying models remain works in progress, and they raise difficult questions about how to handle deep uncertainty and avoid false precision.

Recognizing the Limits of Quantification

The same institutional pressure that produces omission also produces its opposite: analyses that claim more precision than the underlying methods can support.

Quantification is most useful when the projected future resembles the historical record on which models are built, and when analysts have granular data on the places and sectors being studied. Climate-related analyses routinely violate both conditions. They involve long horizons over which technologies, economic structures, and human behavior will change in impossible-to-predict ways, and effects that cascade through every sector of the global economy.

Yet modelers and political principals alike face strong incentives to quantify as much as possible, since a more complete-looking set of numbers confers greater perceived analytical authority and makes regulatory justifications harder to challenge, regardless of whether the underlying estimates are reliable enough to bear that weight.

A prominent example is incorporating estimates of the benefits of avoided damages from reducing greenhouse gas emissions (referred to as the social cost of carbon dioxide emissions) into regulatory benefit-cost analysis. Quantitative estimates of global climate damages over the many centuries that greenhouse gas emissions remain in the atmosphere lie far beyond our analytical capabilities. These models omit many (or perhaps most) of the potential harms that changes in the climate will impose on society, and they unrealistically assume that human behavior will adjust to unprecedented changes in ways consistent with historical experience. The estimates also embed strong moral judgments that are rarely acknowledged as such: how much more we value consumption today versus in the future, how much weight to give the risk of highly uncertain catastrophic outcomes, and how much more to weigh harms falling on the most vulnerable people relative to those who are better off.⁸

The result is a profound tension between what rigorous analysis requires and what tractable models can deliver. Nevertheless, it has become common to integrate estimates of the social cost of emissions into benefit-cost analysis and present the results as justifications for policy decisions.⁹

The same dynamic plays out in assessments of green industrial strategies. Policymakers pursue these strategies for a broad range of reasons: to build domestic productive capacity, reduce exposure to geopolitical risk, support regional economic stability, accelerate innovation with broad spillovers, and sustain political coalitions for long-term policy commitments. These are legitimate policy goals, but they have real costs, including the distortions introduced by trade barriers and the inefficiencies and corruption that can accompany directing capital toward favored sectors.

Existing models are poorly suited to weighing these benefits and costs, tending to capture only a narrow slice of either side of the ledger. Some studies omit important potential benefits of green industrial strategies and project negative outcomes,¹⁰ while others ignore key drawbacks and find that these strategies boost growth and jobs.¹¹ Too commonly, these analyses draw overconfident conclusions about their net economic effects, presenting partial and uncertain estimates as guidance for policymakers who should be encouraged to both use models and rely on a richer array of evidence.¹²

Recommendations

Correcting these failures requires three mutually reinforcing reforms. The most fundamental is not to the models themselves but to the decision-making framework within which the models operate: replacing the culture of false comprehensiveness with increased transparency about model limitations.

The 2023 revision to the federal government’s guidance on regulatory benefit-cost analysis (Circular A-4) made genuine progress in providing clearer guidance on nonquantifiable benefits and costs.¹³ But in layering this new guidance on top of the older norms rather than replacing them, it created an awkward hybrid in which analysts were encouraged to acknowledge what they could not quantify while still facing institutional pressure to produce comprehensive-looking net-benefit numbers.

For example, the updated Circular A-4 allowed agencies to voluntarily move beyond the Kaldor-Hicks potential compensation principle that treats a policy as justified if those who gain could in theory compensate those who lose. However, within a culture that requires comprehensive-seeming estimates of net benefits, abandoning the Kaldor-Hicks principle would force federal agencies to make difficult judgments about the societal benefits of regulatory changes with important distributional consequences. In practice, agencies defaulted to the familiar approach, even in contexts like climate change, where the assumption that winners could compensate losers is clearly just a convenient fiction.¹⁴

The Trump administration’s abandonment of the A-4 update, while damaging in the short term, creates an opportunity for the next administration to try again with greater clarity and ambition. A revised Circular A-4 should be explicit that only under narrow conditions are quantified estimates of benefits and costs sufficiently reliable and comprehensive to warrant direct comparison. The circular should also require agencies to identify when those conditions are not met. Outside those conditions, agencies should be guided away from producing net-benefit estimates altogether, presenting instead a transparent account of what is known, what is uncertain, and what value judgments are required. In this way, quantitative analysis of costs and benefits would serve its proper role as one input among several that accountable decision-makers weigh in reaching a conclusion, rather than an illusory bottom line that forecloses the judgment that good governance requires.

The same principle applies to the macroeconomic modeling that underlies the federal budget. The current practice of presenting point estimates of GDP growth as the basis for budget projections implies a false precision that is difficult to defend even before accounting for climate and energy transition dynamics. At a minimum, the Council of Economic Advisers, Treasury, and Office of Management and Budget should replace point estimates with a scenario-based approach that presents a range of plausible economic trajectories. This would also create a natural framework within which climate risks and energy transition dynamics could be incorporated without the appearance of contaminating an otherwise precise forecast. Establishing honest analytical boundaries is not a retreat from quantification.

A second complementary reform is to improve the underlying models. The federal government has an important role to play in funding and supporting that work. The government should invest in better quantification where it is genuinely achievable. Improved models must come with explicit acknowledgment of their limitations so they do not become a new vehicle for false precision. There may be ways to deploy new data and/or new analytic capacities toward these goals—including AI—and the White House should explore those options.

Several areas are ripe for investment. Climate damage modeling has advanced considerably in academic work, moving away from the traditional top-down “damage function” approach toward more rigorous analysis of how climate change affects specific sectors and drivers of economic growth. Models underlying green industrial policy analysis should also be improved to better account for the motivations that drive these policies, including the desire to build domestic productive capacity and reduce exposure to geopolitical risk. Finally, regional economic modeling remains underdeveloped. Dedicated federal investment in modeling local labor markets, effects on local fiscal outcomes, and broader community impacts would generate both better analysis and the data infrastructure needed to evaluate whether place-based interventions are achieving their goals.

With analytical boundaries established and underlying models improved, the case for incorporating climate and energy transition effects into federal macroeconomic frameworks becomes both more compelling and more credible. It no longer rests on an overreach beyond what models can deliver, but on a genuine account of what improved models can and cannot tell us.

Conclusion

When economic analysis systematically overstates its own precision, it not only misleads policymakers but crowds out the transparent engagement with trade-offs and value judgments that sound decision-making requires. And when the numbers underlying policy decisions are eventually exposed as unreliable, the credibility loss damages not just the analysis, but also the policies built on top of it.

The reforms proposed here point toward a different model of what economic analysis is for. A future administration that replaces false precision with transparent accounts of what is known and unknown, who gains and who loses, and what value judgments underlie a policy decision would be better positioned to make the case for climate and industrial policies on a broader set of grounds. Many of the considerations that matter most for these policies—the long-run resilience of communities, the strategic imperatives of energy security, and the moral case for protecting those least able to bear the costs of inaction—resist meaningful quantification. Making them explicit, rather than burying them in technical assumptions or omitting them entirely, gives policymakers a more defensible and transparent basis for the decisions they are already making.

Better federal modeling is not just a technocratic project. It is a prerequisite for the kind of evidence-based, politically durable climate governance that the scale of the challenge demands. What is needed is the institutional will to stop pretending that we know more than we do, and to govern confidently from what we actually know.