Carbon Shelters: Carbon Accounting as Tax Law

Article II vests the executive power in the President. Yet Congress routinely empowers private plaintiffs, not just the President, to enforce public regulatory laws. Because of this, in almost every area of law—from environmental and antitrust law to civil rights and securities law—the bulk of enforcement occurs through private civil suits rather than government-initiated litigation. Scholars and originalist judges are, for the first time, seizing on this apparent contradiction to argue that this mode of “private enforcement” may be an unconstitutional delegation of executive power. In TransUnion v. Ramirez, the Supreme Court observed that a “regime where Congress could freely authorize unharmed plaintiffs to sue defendants” would “infringe on the Executive Branch’s Article II authority.”[1] This, along with invitations by Justices Alito and Thomas, ignited a series of lower court cases and historically grounded critiques of private enforcement. In an emblematic case, Circuit Judge Kevin Newsom drew on TransUnion and other cases to argue that “Article II’s vesting of the ‘executive Power’ in the President . . . prevents Congress from empowering private plaintiffs to sue for wrongs done to society.”[2]

This emerging “Article II Challenge” relies, at its core, on a particular reading of Founding Era history and the original meaning of the Vesting Clause. Challengers argue that the Take Care Clause’s command that the President “shall take care that the laws be faithfully executed,” combined with Founding Era history, establishes that public rights claims—those seeking civil penalties or enforcement of regulatory violations against society—must be enforced exclusively by the executive branch. If the Supreme Court embraces this argument, hundreds of statutory schemes and thousands of claims will crumble.

This Article provides a historical corrective and refutes the emerging Article II Challenge. Our original historical investigation of “penal statutes”—a category of Founding-Era regulatory legislation that anticipated modern private rights of action—uncovers the deep constitutional foundation of this tradition of private enforcement. We offer three key historical contributions. First, we demonstrate that Founding Era lawmakers passed reams of statutes empowering private plaintiffs to sue for wrongs done to society, at both the state and federal levels. Private enforcement of regulatory law was ubiquitous in England, colonial America, and the Early Republic, a broader legal history that originalist proponents of the Article II Challenge have overlooked or misunderstood. Second, we show that post-ratification contemporaries saw no conflict between executive power and private enforcement, even when these suits vindicated public rights. Third, we reveal that the Crown, governors, and presidents could not control private penal statute claims through the writ of nolle prosequi or other mechanisms. We demonstrate this by surfacing a previously neglected 1791 case in which President Washington and Secretary Hamilton concluded that the President could not control civil litigants seeking statutory penalties, even when they proceeded nominally on behalf of the United States. This historical evidence weakens both the Article II Challenge and the “unitary executive” theory because it shows that Congress was free to empower private plaintiffs to execute the law. We then supplement those historical insights with an analysis of contemporary legal doctrines governing executive control over private enforcement. We conclude that private enforcement does not violate Article II, except under extremely narrow conditions. Our proposed constitutional test would leave most private enforcement schemes untouched. Ultimately, this Article provides a comprehensive historical and doctrinal defense of private enforcement writ large.

Table of Contents Show

Introduction

How do you transition an economy from fossil fuels to zero-carbon energy? At a conceptual level, there are four major approaches reflecting two axes of difference. Decarbonization policy can operate through carrots (incentives) or through sticks (penalties), and it can be universal or selective. The distinction between carrots and sticks is relatively familiar: Any decarbonization policy faces a choice between penalizing the production of excessive carbon emissions or rewarding clean alternatives.[1] The distinction between universal and selective policy is less obvious, but equally important. Carrots or sticks can both be selective, meaning that they apply only to energy generation or usage in specific industries, or in specific firms singled out for inclusion in the policy. Alternatively, carrots or sticks can both be universal, meaning that they apply to all sources of carbon emissions and/or clean energy. Universal policies are said to have the appeal of being “technology neutral,” meaning that they are indifferent between any two technologies that reduce carbon emissions by the same degree.[2]

Consider examples of each of the four approaches. Starting with sticks, the selective sort are those policies that penalize or restrict some (but not all) carbon emissions, like the Environmental Protection Agency (EPA)’s rules on power plant emissions, promulgated pursuant to the Clean Air Act.[3] Those rules are selective because they pertain only to emissions in a particular industry (the power sector). An example of a universal stick would be a carbon tax, insofar as it would be imposed on all carbon emissions.[4] Selective carrots, on the other hand, include grants offered on a discretionary basis to renewable energy companies, as well as traditional technology-specific tax credits such as a tax credit for solar energy production. Finally, an example of a universal carrot would be a technology-neutral tax credit—a subsidy offered to any recipient that can displace carbon emissions with clean energy.

This Article explores the appeal—and unintended consequences—of embracing universal carrots, the option Congress has largely chosen in recent years. Universal carrots are theoretically well suited to encouraging innovation in clean energy technology because they are technology neutral. But they carry a distinctive risk: that in their universality they will overflow beyond their targeted recipients and instead end up in the wrong hands, enriching the very carbon emitters they were meant to exclude.[5]

In order to deal with that concern, a policy of universal carrots needs rules and/or standards that define the universe within which carrots can be handed out freely. In the case of decarbonization policy, that is the task of carbon accounting. Carbon accounting is the discipline of measuring the greenhouse gas (GHG) emissions that result from an industrial process.[6] Over the last few years, carbon accounting has emerged as a critical measurement paradigm within environmental law.[7] In an influential article, Shelley Welton identified conceptual flaws in carbon accounting as one of several critical risks to the “net-zero” paradigm of reducing carbon emissions.[8] More recently, Leehi Yona explained how GHG undercounting has the potential to undermine various bodies of regulatory law.[9] In general, scholars who have addressed emissions accounting have taken for granted that if it is to become part of our law, it will be part of environmental law.[10] But in fact, Congress has assigned this baroque task of scientific measurement and accounting not to any of the science-focused agencies, but to the Department of Treasury (Treasury) and the Internal Revenue Service (IRS), because those agencies administer the universal carrots. For better or worse, real-world carbon accounting will be a product of tax law.

Is tax law an appropriate home for gatekeeping access to the universal carrots of clean energy policy? David Weisbach and Jacob Nussim reason that it makes sense to channel spending through tax law only when “the coordination benefits between the tax system and the other program are high and the specialization benefits of separate programs are low.”[11] So, for example, welfare programs are a good candidate for tax integration because the tax system and welfare system make determinations along similar dimensions, like applicants’ incomes.[12] But integrating carbon accounting into tax law does not only mean that the IRS will become responsible for administering it; it also means that tax experts at the Treasury will come up with rules, tax accounting firms will recommend avoidance strategies under those rules, tax lawyers will litigate disputes, and tax academics will propose principles for law reform. In this sense, integrating a policy domain into tax law means integrating it with a professional community eager to apply its own internally legitimate principles and standards.[13] Whether this integration is a good thing should depend on whether the other domain (here, carbon accounting) has much to learn from tax law.

This Article argues that the principles, rules, doctrines, and administrative strategies of tax law can usefully inform how the government conducts carbon accounting. Fundamentally, income tax accounting and carbon accounting are both projects involving tracking down a measure of interest (income, emissions) and deciding how to allocate it among responsible parties. At the level of theory and goal setting, concepts from tax theory have already influenced the prevailing conception of what carbon accounting is meant to achieve. These concepts include a comprehensive tax base, technology neutrality, and additionality, the principle that one should only be able to take credit for effects that would not have occurred without one’s actions. At the level of rules and doctrine, tax law’s experience in determining when to split income among taxpayers and when to blend disparate tax rates together can usefully inform questions of splitting and blending within emissions accounting. And at the level of administration, tax law’s familiarity with third-party verification—including verification from other administrative agencies—suggests a path for overcoming the technical challenges of carbon accounting that the IRS could not manage alone. In sum, tax law is surprisingly well positioned to house the law and administration of carbon accounting.

Yet even tax law cannot easily dispatch with the risks of universal carrots. Universal carrots are supposed to be available to anyone who produces energy without GHG emissions. That standard, however, gives taxpayers complete discretion in how they achieve zero emissions. Like other performance standards, it opens up boundless degrees of freedom for clever pseudo-compliance that violates the purpose of the law.[14] This is a familiar conundrum in tax law, where taxpayers invent “tax shelters” that technically comply with the rules but substantively avoid tax. This Article argues that energy firms will likewise develop “carbon shelters,” transactions that qualify for green subsidies while making use of high-emission energy.

In fact, the current set of green-subsidy rules actually create carbon shelters by granting taxpayers “safe harbors,” fact patterns that are treated as compliant with a zero-emissions standard even though they likely fall short of it.[15] Safe harbors are administratively simpler than pure performance-based regulation, but in practice, many of the safe harbors designed by the Treasury and the IRS achieve that simplification by giving taxpayers a surefire defense for sheltering emissions.[16] This Article diagnoses logical flaws in many of the safe harbors proposed in green-subsidy rulemaking and cautions against relying on safe harbors as the solution to the rigors and uncertainty of performance-based regulation.[17]

Instead, I argue that tax law’s green-subsidy regulations should evolve in the opposite direction: Rather than granting safe harbors for potentially high emission practices, the law should deny subsidies under fact patterns where positive emissions are likely but hard to measure. In other words, when administrative complexity makes it impossible to stick with a pure performance standard, the law should default to unsafe harbors (or prohibitions) rather than safe harbors.

This cautious approach is the appropriate response to the asymmetric downside risk posed by carbon shelters. The history of carbon accounting has been a continual story of underestimating the full scope of emissions before updating the standard methodology to account for emissions previously ignored.[18] Especially when new energy technologies become eligible for subsidy—the very point of technology neutrality—regulators will probably underestimate their consequential emissions in the first instance. And because most green tax credits are uncapped, there is no limit to the amount of money the government might become obligated to spend on carbon shelters—the very opposite of the statutory intent to reward green energy.

To counteract this risk, energy tax law should adopt broad anti-abuse rules that rule out subsidy eligibility for certain classes of transactions that, while not necessarily abusive in every instance, are particularly likely to serve as shelters. This approach borrows its logic from a prominent strategy used to combat tax shelters. There, Congress has written broad rules targeting risky fact patterns. The passive loss rules reflect this approach: While the concept of a passive loss is entirely coherent within income tax accounting, experience teaches that taxpayers opportunistically generate passive losses to offset nonpassive income.[19] Similarly, carbon offsets and negative emissions technology are theoretically valid constructs within carbon accounting, but the risks of taxpayers exploiting those concepts to shelter emissions are immense. For that reason, this Article proposes a broad prohibition on including negative emissions in green-subsidy carbon accounting.[20]

This Article proceeds as follows. Part I recounts the intellectual and policy history by which selective carrots and universal sticks gradually transformed into the near-universal carrots that define federal energy policy today. Part II examines the challenge of making carrots universal by comparing all potential subsidy recipients on the basis of their carbon emissions. This Part introduces the conceptual challenge of measuring emissions as well as the administrative challenge of instantiating emissions accounting inside the IRS. Part III then argues that the federal government’s current approach to emissions accounting will spur the proliferation of carbon shelters, transactions that win green subsidies while increasing emissions. This Article proposes to head off carbon shelters by developing broad anti-abuse rules, modeled on prophylactic rules from tax law.

I. The Rise of Universal Carrots: A Brief History

This Part presents the intellectual and policy history of how carbon accounting has become part of tax law. I argue that carbon accounting is a necessary ingredient for any regime of technology-neutral, universal energy incentives. Therefore, understanding the emergence of carbon accounting requires understanding the move toward these universal carrots. This Part begins by depicting a longtime consensus in energy policy around addressing climate change through sticks, including the universal stick of a carbon tax. It then explains why Congress instead deployed selective carrots for several decades. Finally, it narrates the recent shift from selective to (mostly) universal carrots—carrots completely reliant on carbon accounting for their basic functioning.

A. From Sticks to Carrots

For decades, the conventional wisdom was that any effort to tackle climate change through the tax code would take the form of a carbon tax.[21] The intellectual architecture undergirding the idea of carbon taxation can be traced back to the foundations of environmental economics.[22] A.C. Pigou, the Cambridge economist, identified what are today known as externalities: the effects of an economic transaction on third parties.[23] Today, taxes aimed at discouraging the imposition of harms onto others are commonly called “Pigouvian.”[24] In his famous 1977 paper, William Nordhaus advocated for the uniform Pigouvian taxation of carbon emissions across the economy.[25] Nordhaus reasoned that taxes would be superior to regulating the allowable quantity of emissions because the cost of abating a ton of carbon varies from sector to sector, and so the approach of regulating quantities would have to be customized for every application, whereas a single tax rate per ton of carbon could be applied uniformly to the entire economy.[26] This principle of uniformity is seen as one of the main virtues of carbon pricing: It does not force anyone in particular to reduce their emissions but allows the market to decide where it is most cost-effective to decarbonize and where continuing to emit is worth the high cost.

The carbon tax idea, brought to maturity by Nordhaus, soon became one of the most popular ideas within academic economics. The “Largest Public Statement of Economists in History” is an open letter endorsing a carbon tax and dividend, originally published in The Wall Street Journal in 2019 and now signed by over 3,500 U.S. economists.[27] Former Treasury Secretary Larry Summers called it “one of the few ideas of economic policy that commands broad, bipartisan support.”[28] The academic popularity of carbon taxation has brought modest real-world success. Finland, Sweden, Norway, Denmark, and Poland adopted carbon taxes in the early 1990s.[29] Switzerland, New Zealand, Australia, the United Kingdom, and France have since adopted versions of carbon taxes, although France suspended its tax after one year of operation in 2018.[30] Cap-and-trade markets, an alternative approach that sets a pollution quota and allows participants to sell emissions permits, have been more politically successful, emerging in the European Union, the northeastern United States, and California.[31]

Despite its immense academic popularity, carbon taxation has its own shortcomings.[32] Its most significant weakness is one of political economy: Carbon taxes are difficult to achieve in a democracy because they are experienced as a cost to all but a concentrated benefit to none.[33] From a political durability perspective, it is easier to pursue clean energy policies that “create sticky interests by incentivizing multidecade investments in clean energy infrastructure.”[34] Beyond the political challenge, a concern about efficacy is that a carbon price threatens emitting firms with higher operating expenses but does not directly subsidize the large capital expenses that will be necessary to substantially decarbonize the economy.[35] In principle, the promise of lower operating expenses can convince firms to make capital investments since, over a long enough time, the savings will recoup the initial investment. But firms will only do this once their current assets exhaust their useful life, which will be several decades hence for many gas and coal plants.[36] Furthermore, due to status quo bias and risk aversion, it often takes an extreme incentive to encourage firms to make a risky change.[37] In light of these critiques, some scholars have concluded that decarbonization policies should offer concentrated incentives (i.e., subsidies) rather than the diffuse incentives of a carbon price.[38]

Indeed, irrespective of the decades of academic discourse on carbon taxation, subsidies for low- and zero-emissions energy technology have long found more success at the federal level. Subsidies can be justified on the same Pigouvian grounds as taxes: If there are positive social externalities to some behavior, a subsidy allows anyone to “fully internalize the benefits to society of his or her behavior,” which leads to an optimal supply of the subsidized behavior by reducing its price.[39]

Starting in the 1960s, Congress enacted a string of tax subsidies for green-energy technologies. The first step was the initial investment tax credit (ITC), enacted in 1962.[40] The investment credit permitted taxpayers to claim a credit for 7 percent of the amount invested in new capital equipment, with no limitation by type of equipment or industry.[41] In 1978, motivated by reducing dependence on Middle Eastern oil, Congress moved to supplement the investment credit with a more targeted version exclusively for investments made in alternative energy equipment.[42] Congress defined six categories of credit-worthy technology: alternative energy property, solar or wind energy property, specially defined energy property, recycling equipment, shale oil equipment, and equipment for producing natural gas from geopressured brine.[43] Each category reflected a distinct subsector of the burgeoning renewable energy industry.

In 1992, Congress created the newest form of energy subsidy: a production tax credit (PTC) for the production of electricity from renewable sources.[44] Whereas the ITC was distributed in proportion to money invested in renewable energy equipment, the PTC would be distributed in proportion to kilowatt-hours of electricity generated and sold.[45] In this way, the logic of the PTC was more closely tied to the ultimate outcome of interest: the quantity of renewable energy consumed and its displacement of fossil alternatives. The PTC was also somewhat more targeted than the ITC had been in terms of technological scope: It initially only applied to electricity produced from wind or closed-loop biomass.[46] Additional methods of electricity generation were gradually included; most notably, Congress added solar in 2004.[47]

Solar and wind have been the two most widespread sources of renewable energy generation since the beginning of the renewable industry in the 1970s. Although frequently grouped together, solar and wind have different use cases, market structures, and cost constraints.[48] The original design of the PTC was much more favorable to wind than to solar: The subsidy level was enough to make wind cost competitive with fossil generation but not enough to do the same for solar.[49] The ITC was equally one-sized, if better suited to solar than to wind. Paired together, they were adequately targeted to an infant renewables sector composed almost entirely of those two technologies.[50] But as new green energy technologies came onto the scene, it became increasingly difficult to design subsidies customized to the needs of each one. In that context, this Article next turns to the development of universal carrots, subsidies meant to apply equally across different technologies.

B. From Selective to Universal

The most influential principle in the evolution of modern green-subsidy policy has been the conviction that subsidies should be neutral among technologies. According to this view, subsidies should not be made available only for specific technologies (e.g., solar and wind energy) that policymakers may favor but should be open to any technology that can achieve a certain goal (e.g., electricity generation or fuel combustion) with low or zero emissions. This conviction grew directly out of the academic consensus behind carbon taxes, which have a virtue of making all sources of carbon emissions more expensive and therefore equally benefiting all technologies that might mitigate emissions. As David Schizer explains, “covering all sources of harm (with taxes or permits) is easier than reaching all options for abating this harm (with subsidies).”[51] In other words, it is easier to identify the relevant harm—carbon emissions—than to identify all possible solutions for that harm.

If it were not possible to have a carbon tax, Schizer and likeminded peers argue that any subsidy-based alternative should at least be broadened to include any low-emissions technology rather than attempting to “pick winners.”[52] Zach Liscow and Quentin Karpilow summarize the “conventional view” among economists and legal scholars that the “government should not attempt to directly influence the direction of technological development” but should instead “tax pollution, provide technology-neutral innovation support, and then let the cards lay where they fall, trusting the private sector to respond in the cheapest, most efficient way.”[53]

And yet, some scholars who cheer technology neutrality in the case of a carbon tax have warned that it could not be replicated in a world of green subsidies. David Weisbach explains that “we cannot design a technology-neutral subsidy” in the sense of being neutral between fossil energy sources and the green technologies that might replace them.[54] This is because the emissions impact of a subsidy depends on which dirty technology it helps replace, and so a subsidy for wind cannot simultaneously be neutral between wind and coal and between wind and gas.[55] If the subsidy value for wind is equivalent to an efficient carbon tax on gas (i.e., one that accounts for the negative externalities of an equivalent amount of gas generation), then it must be too low in relation to coal, because coal produces more emissions than gas.[56] If the subsidy for wind is efficient in relation to coal, then it must be too high in relation to gas. The only way around this problem would be to customize subsidies to account for which fossil energy source the subsidized green energy is likely to replace—a horrendously complicated undertaking.[57]

A second limitation to the idea of technology neutrality is that, from an innovation policy perspective, the goal of spreading government support among all equally performing technologies may miss the power of “innovation snowballing.”[58] In certain cases, there are increasing returns to concentrating investment within a select few technological paradigms. Encouraging investment in technologies related to solar energy (for one example) will reduce the costs of solar and attract additional entrepreneurs to the solar sector, potentially leading to an innovation snowball that would be delayed if the initial investment were spread among many technologies.[59] The innovation snowballing perspective is the best justification for investments that Congress has directed the Department of Energy (DOE) to coordinate in selected clean energy paradigms like hydrogen and carbon capture.[60]

Nonetheless, the ideal of technology neutrality and the closely associated method of performance-based regulation have been central to the evolution of green subsidies over the past fifteen years. Senator Max Baucus encapsulated the appeal of technology neutrality in the opening remarks of a 2009 hearing: “[T]he government might just set a performance standard, regardless of the technology employed. We could encourage things like reductions in greenhouse gas emissions, improvement in efficiency, or increased energy content. And then we would leave the job of picking the best technology to the competition of the market.” [61] The first notable proposal came in 2013, when a draft bill overhauling the energy tax credit landscape would have reduced forty-two separate credits into two technology-neutral credits, one for all possible means of generating electricity and the other for all possible transportation fuels.[62] That particular bill did not make it onto the President’s desk, but the basic principles of that discussion draft reappeared in the Inflation Reduction Act (IRA), which finally enshrined technology neutrality in law.[63]

C. Universal Carrots Require Carbon Accounting

In 2021 and 2022, Congress enacted sweeping energy-related tax credits. The Congressional Budget Office estimated that the new credits would cost $400 billion, while private sector analysts concluded that the figure might actually double to $800 billion.[64] The uncertainty between the two sums arose because Congress made most of the credits available to as many firms as qualified, in as large a volume as those firms wished to pursue the incentivized activities.[65] What distinguished these green subsidies from prior ones was not just their size but the centrality of GHG emissions accounting principles found within their rules. Post-IRA, six major tax credits relied heavily on emissions accounting.[66]

The practice of GHG emissions accounting first emerged in response to the United Nations’ Kyoto Protocol, which asked participating nations to compile national emissions inventories.[67] The Kyoto Protocol and its successor, the Paris Agreement, are concerned not just with measuring emissions but with assigning responsibility to specific parties—in that case, nations.[68] For non-state actors, the emissions accounting protocol of choice has been the GHG Protocol, the source of the now-famous tripartite distinction between scopes of emission accounting.[69] Under this approach, Scope 1 emissions include an actor’s “direct” GHG emissions, Scope 2 emissions include indirect emissions resulting from the actor’s electricity use, and Scope 3 emissions include all other indirect emissions within the actor’s supply chain.[70]

At the federal level, emissions accounting first appeared in a 2007 amendment to the Clean Air Act authorizing the EPA’s Greenhouse Gas Reporting Program, which requires emissions reports from facilities that emit at least 25,000 megatons of CO2 annually.[71] The EPA estimates that this program covers 85 to 90 percent of total U.S. greenhouse gas emissions from over eight thousand facilities.[72]

Emissions accounting is also critical to the EPA’s Renewable Fuel Standard program, where the EPA must approve putatively renewable fuels before refiners and importers of gasoline and diesel fuel can claim credit for incorporating those fuels into their mix. When companies petition for a new “fuel pathway” (a type of fuel made from a specific feedstock, e.g., jet fuel made from soybean oil), the EPA conducts an analysis of the total “life-cycle” emissions of the fuel pathway.[73] This form of evaluation, known as life-cycle assessment (LCA), is the core exercise of emissions accounting. As of 2013, the EPA had approved ten of the forty-two petitions it had received for new fuel pathways.[74] These fuel pathway evaluations have served as the cutting edge of government-supervised emissions accounting.[75] Emissions accounting has also made its way into federal securities law under an SEC rule that requires firms to include emissions-related disclosures in their annual reports.[76]

The first appearance of emissions accounting in the tax code came in 2018. Prior to 2018, there was already a tax credit for capturing and disposing of carbon oxides. When carbon oxide is captured and disposed of in secure geologic storage, there is no need for complicated emissions accounting: Every ton of carbon stored is one fewer ton in the atmosphere, simple as that.[77] But in 2018, Congress amended the preexisting tax credit for carbon oxide sequestration to add a new subcredit for the utilization (i.e., rather than just sequestration and storage) of carbon oxides.[78] Carbon oxide can be “utilized” by being sold for use in oil wells, concrete production, chemical manufacturing, or carbonated beverage bottling. In those scenarios, the amount of carbon emissions actually displaced is highly fact specific.[79] Congress therefore made the carbon capture and utilization credit dependent on the amount of carbon oxide “permanently isolated” or “displaced from being emitted” into the atmosphere.[80] Furthermore, Congress directed that those terms be measured based on “an analysis of lifecycle greenhouse gas emissions”—an instruction to engage in emissions accounting.[81]

The remaining instances of emissions accounting in the tax code all appeared in August 2022 in the IRA.[82] The credit for sustainable aviation fuel paid at a rate indexed to the “lifecycle greenhouse gas emissions reduction percentage,” or the extent to which the fuel’s lifecycle greenhouse gas emissions are lower than that of petroleum-based jet fuel.[83] The credit for clean fuel production was set to replace the sustainable aviation fuel credit and all other fuel-specific credits in 2025, and it too subsidized fuels based on their emissions rates.[84] The IRA offered credit for clean hydrogen at four tiers, where the more valuable tiers depended on the hydrogen being produced through a process that results in a lifecycle greenhouse gas emissions rate below certain thresholds.[85] Congress created the “Clean Electricity Credits” to replace the longstanding electricity Production Tax Credit and Investment Tax Credit in 2025. The primary difference between these new credits and their predecessors was that, while the old ones were made available for specifically enumerated electricity generation technologies (as discussed above), the new ones were available for any electricity produced at a facility with a “greenhouse gas emissions rate not greater than zero.”[86] In just a few years, “lifecycle greenhouse gas emissions” and “greenhouse gas emissions rate” had become critical terms to the operation of the new era of universal carrots.

D. Universal Carrots Get Less Universal

In 2025, Congress passed and President Trump signed the One Big Beautiful Bill Act (OBBBA), a partial rebuke to the suite of green subsidies enacted under President Biden. The most important change was the acceleration of the termination date for solar and wind ITCs and PTCs, and hydrogen PTCs: Credits will no longer be available for solar, wind, and hydrogen facilities placed in service after 2027.[87] Removing solar, wind, and hydrogen from the scope of green energy credits is obviously a step away from universality. But it is not quite a return to the world of selective carrots that predated the IRA. Prior to the IRA, Congress had a whitelist of approved energy technologies. Post-OBBBA, Congress has a blacklist of disallowed technologies but still relies on an open-ended carbon accounting standard to determine which non-blacklisted technologies can qualify for subsidy.

The impending disappearance of certain green technologies from credit eligibility moderately softens some of the concerns raised in this Article. On the one hand, solar and wind never would have posed the carbon accounting challenges discussed throughout this Article. There is no debate that they are zero-emissions technologies, and the Treasury’s regulations have provided an expedited process for confirming that they are entitled to full credit under the technology-neutral ITC and PTC.[88] The hydrogen credit, on the other hand, would have been a major locus of accounting contestation and will remain so through 2027 when its credit expires. I discuss issues with implementing the hydrogen credit at various points herein, both because tax credit expiration dates are rarely set in stone and because of general lessons exemplified by the hydrogen credit and its regulations.

Moreover, other provisions of the OBBBA added to the concerns about measurement and carbon shelters raised below. As discussed in the next Part, the IRA wrote a maximally ambitious version of carbon accounting into statute and left the Treasury with the job of making pragmatic compromises in regulations. The OBBBA took a different approach, adding various statutory rules about what factors regulators may and may not consider in carbon accounting. And so, while the OBBBA validates the continued centrality of carbon accounting to energy tax law, it also expands opportunities for carbon sheltering, which I delve into in future sections.[89]

II. Administering Universal Carrots

In this Part, I explore the challenge of making carrots universal, which is fundamentally the challenge of distinguishing low-carbon recipients from high-carbon alternatives. The former are meant to be included, and the latter kept out. This Part breaks down the task of administering universal carrots into three components: what emissions get counted, who does the counting, and how to count.

The first challenge of administering universal carrots is the question of which emissions count. I begin by articulating the principles that carbon accounting is meant to uphold, according to Congress’s mandate. Congress has instructed the tax agencies to count allemissions “related to” subsidized projects.[90] I argue that this directive requires conducting both attributional and consequential carbon accounting. The former is concerned with tracking down emissions in a product’s supply chain and the latter with modeling indirect emissions caused by a production process, even when they occur at a distance elsewhere in the economy. I show that the central normative principles behind both attributional and consequential accounting have either traveled over from, or have informative parallels in, the world of tax law.

The second challenge is the question of which government officials oversee taxpayers’ carbon accounting. The IRS will rely on peer agencies like the EPA and DOE, both to develop standardized carbon accounting models, and to evaluate more bespoke models submitted by taxpayers. But even with the greater expertise of those agencies, the difficulty of evaluating carbon accounting is that, according to statute, each taxpayer’s submission is subject to a broad standard—whether the taxpayer achieves zero emissions—rather than a clear-cut set of rules. Regulators face the prospect of having to double-check each taxpayer’s modeling choices, which threatens to be administratively unworkable, no matter which agency is given the job.

This leads to the third challenge: If an open-ended performance standard is not administratively viable, what can replace it? The Treasury and IRS have answered this question by creating safe harbors, rules that simplify adjudication by treating certain fact patterns as presumptively zero-emission, even if they might actually not be. In effect, the Treasury and IRS are sacrificing accuracy for administrability. But sacrificing too much accuracy risks producing self-defeating outcomes and undermining the task of reducing GHG emissions.

A. What Gets Counted

Once Congress has committed to technology neutrality, and once it has determined that the test subsidy seekers must pass is achieving a low GHG emissions rate, the next question must be: Which emissions count?

In tax law, the first step in designing any new tax is to define the base to which it applies, i.e., to identify the set of economic activities subject to tax. For example, the federal income tax has a base of “all income from whatever source derived,”[91] while the excise tax on coal has a base of “[the value of] coal from mines located in the United States sold by the producer.”[92] Perhaps the most consequential step in defining the scope of green subsidies is to name which emissions are inside and outside the base.

In tax theory, there is a normative ideal of a “comprehensive tax base.”[93] The idea is that the base should be as broad as possible to avoid distorting economic decision-making. If the base is less than fully broad, actors will have an incentive to move their activities outside the base. The comprehensive base can also be justified on fairness or neutrality grounds: It is neutral among all activities because all are included within the base.[94] The comprehensive base concept can be applied to an income tax, a consumption tax, or even a carbon tax: In each instance, the idea is that all cases of the target phenomenon (income, consumption, carbon emissions) should be included.

But there is something inherently more difficult about applying the comprehensive base ideal in the context of a subsidy compared to a tax. Assuming that every person and entity who might potentially conduct the taxable activity is subject to a tax, nothing should escape the base. If everyone who might potentially earn income is subject to an income tax, the base has a chance to be comprehensive (assuming no exemptions, etc.). But consider a subsidy that, by design, will only flow to a select cohort of recipients. If the logic of the subsidy is “pay everyone who produces energy without carbon emissions,” but not everyone involved in producing energy is examined, it is possible that the industry will arrange itself such that subsidy seekers will not generate any emissions directly but instead outsource their emissions-generating activities to other firms that do not apply for the subsidy. The firms that cannot hide their connection to carbon emissions will not apply for the subsidy but will nonetheless benefit from it by continuing to do business with the subsidy winners and possibly bargaining for some share of the subsidy spoils.

To guard against the risk of mistakenly subsidizing firms with hidden emissions, Congress has adopted an expansive understanding of which emissions should count. As described in Part I, each of the major green-subsidy statutes calls for an analysis of “lifecycle greenhouse gas emissions,” and in each statute that term is given the meaning as described in subparagraph (H) of section 211(o)(1) of the Clean Air Act (42 U.S.C. § 7545(o)(1)).[95] Congress had added this definition of lifecycle emissions for the purpose of the EPA’s Renewable Fuel Standard program.[96] The definition is:

The aggregate quantity of greenhouse gas emissions (including direct emissions and significant indirect emissions such as significant emissions from land use changes), as determined by the Administrator, related to the full fuel lifecycle, including all stages of fuel and feedstock production and distribution, from feedstock generation or extraction through the distribution and delivery and use of the finished fuel to the ultimate consumer, where the mass values for all greenhouse gases are adjusted to account for their relative global warming potential.[97]

The definition draws attention to two distinct ways carbon accounting could go wrong. First, one might fail to count emissions at every relevant stage in the production process, or “the full fuel lifecycle.”[98] As explained below, this is the challenge of attributional accounting: the task of attributing all relevant emissions, and only the relevant emissions, in a multistage, multiproduct supply chain to the product at hand. Second, one might fail to count “significant indirect emissions.” As explained below, this is the challenge of consequential accounting: the task of determining which emissions are causally attributable to a production process, even if they do not occur anywhere in the direct supply chain. Attributional and consequential emissions are both necessary to count in order to fulfill Congress’s mandate, but they implicate completely different justifications and require different accounting strategies as the rest of this Section details.

1.   Attributional Accounting

The task of attributional carbon accounting is to identify which emissions a given product should be held responsible for.[99] At a high level, this process has two steps: first, identify all the emissions connected to the product under evaluation (i.e., the emissions in its supply chain), and then determine which ones should be assigned to the product. The first step is about making sure to include all emissions, encompassing those upstream and downstream of the product. The second step is about deciding whether any of those emissions should be allocated to other products to avoid double counting. The need to count all emissions both upstream and downstream of the product under analysis follows directly from Congress’s directive to count emissions at “all stages of fuel and feedstock production and distribution, from feedstock generation or extraction through the distribution and delivery and use of the finished fuel to the ultimate consumer.”[100] Analysts (and Treasury regulations) refer to that expansive scope of analysis as “cradle to grave.”[101]

After one has identified all emissions from cradle to grave, the next step of attributional accounting is to decide whether any of those emissions should nonetheless be assigned to some other product rather than the one under evaluation. Depending on the purpose at hand, carbon accounting can be applied in either a mutually exclusive or an overlapping manner. In an overlapping approach, emissions are assigned to every actor with the capability of reducing those emissions (e.g., assign the same supply chain emissions to a producer, transporter, and customer).[102] The overlapping approach allows double counting, which can be appropriate if the goal is to assign responsibility to all actors who might be able to reduce emissions.[103]

If double counting is unavailable, the subsidy applicant will be tempted to split emissions among multiple products so as to reduce the emissions associated with the product under investigation. Often, a single industrial process yields multiple products. From a subsidy claimant’s perspective, the incentive is to attribute some of the emissions associated with the product covered by the law to other products not covered by the law. For example, the production of hydrogen often generates steam as a byproduct, and a taxpayer might like to allocate any emissions from this process to the steam rather than to the hydrogen (because the steam is not subject to any tax or subsidy). Depending on how emissions are allocated among products—e.g., by volume or sales value—the producer might try to increase the prominence of secondary products to deflate the emissions share assigned to the primary product. Perverse incentives for generating waste have already plagued carbon offset markets; here, there would be an incentive to produce what was previously classified as waste and sell it (no matter how cheaply) as a byproduct to soak up the primary product’s emissions.[104]

The prospect of producers strategically allocating emissions resembles a familiar problem in tax law of taxpayers allocating income to low-tax related parties. In general, the tax law does not allow taxpayers to offload income to related parties while continuing to benefit from it. Parents might wish to gift income-generating assets to their lower-bracket children but federal law taxes children’s investment income above a threshold at the parents’ marginal rate.[105] U.S. companies might wish to license intellectual property or sell services to their controlled foreign affiliates at below-market rates so as to allocate relatively more taxable income to low-tax jurisdictions and relatively less to the United States, but section 482 (the transfer pricing rule) allows the IRS to reallocate tax items between the parties in line with an arm’s-length standard, i.e., what an unrelated third party would have paid.[106] And a high-tax partner might wish to contribute an appreciated asset to a partnership and then distribute the asset to a low-tax partner, but the “mixing bowl” rules force the contributing partner to recognize the taxable gain.[107]

Among these tax law analogies, the transfer pricing rule is the best analogy for emissions splitting. Just as the transfer pricing rule necessitates a complex, fact-dependent inquiry into the proper arm’s-length price between related parties, so too does emission splitting require a complex, fact-dependent inquiry into the share of emissions properly attributable to the main product versus byproducts. Instead of looking to arm’s-length prices as the relevant baseline, here the IRS might limit eligible byproducts to those the taxpayer produced before a certain date, such as the date the IRA was enacted.[108] Similarly, the IRS could limit the allowable quantity of byproducts based on prevailing technical efficiency standards.[109] Relying on these heuristics is one way to protect against the abuse of allocating excessive emissions to byproducts.

2. Consequential Accounting

The purpose of consequential carbon accounting is to measure indirect emissions. Whereas attributional accounting starts with a known total universe of emissions and works to allocate them among products and responsible parties, consequential accounting starts with an industrial process and seeks to estimate the emissions it will cause elsewhere in the economy. The motivation for consequential accounting is the understanding that focusing only on the direct emissions of a given project “omits partial and general equilibrium responses.”[110] Much of the scientific literature GHG accounting relies on has been produced based on small-scale, localized studies. The concern is that these studies might underestimate the aggregate and systemic effect of many actors pursuing the same incentive at once.[111] So-called “indirect” emissions refer to emissions caused elsewhere in the economy by the industrial process under evaluation, most commonly via supply and demand effects. Indirect emissions are substantial: Based on evidence from existing carbon taxes in Europe, indirect carbon leakage through electricity and fuel markets is quantitatively larger than direct carbon leakage through supply chains.[112]

IRS’s legal directive to conduct consequential accounting can be found in the same clause of the Clean Air Act discussed above and incorporated into every green-subsidy statute.[113] The key language is “including direct emissions and significant indirect emissions such as significant emissions from land use changes . . . related to the full fuel lifecycle.”[114] “Indirect emissions” require consequential modeling to estimate. The example of indirect emissions offered by the statute—“such as significant emissions from land use changes”—provides the most useful guidance as to what emissions should be considered.[115]

The reference to land use changes invokes the category of indirect emissions that has received the closest study to date, even as it is just one example of a broader set of indirect effects that consequential carbon accounting should theoretically track. The requirement to measure indirect land-use emissions was first created under the EPA’s Renewable Fuel Standard rule.[116] There, the concern was that by turning crops into biofuels, the global economy would need to cultivate new agricultural land to replace the lost food supply, and that the resulting deforestation would cause substantial emissions.[117] Indeed, the Department of Energy has acknowledged that diverting a large fraction of global corn production for ethanol has caused enough land use changes (including nitrate leaching, phosphorus runoff, and soil erosion) to undermine the fuel emissions benefits.[118] In response, Congress instructed the EPA to account for land use changes, and the EPA interpreted this mandate to include such changes wherever in the world they might occur.[119] In the OBBBA, however, Congress reversed course and explicitly excluded land use effects from the calculation of emissions for the Clean Fuel Credit.[120] Excluding consideration of indirect land-use emissions goes much further than merely failing to require them; it is an example of an emissions-sheltering safe harbor, discussed below in Part III.

But land use changes are just one example of a broader category of what some analysts call “market-mediated emissions.”[121] Market-mediated emissions arise when producers use some scarce input of production (e.g., land, or solar energy), which reduces supply of that input for the rest of the economy, which causes other people or firms to substitute for the scarce input with some higher-emissions replacement. For example, consider a hydrogen production plant that runs by electrolysis, the process of splitting water into hydrogen and oxygen. Suppose the plant powers its electrolyzers with electricity purchased from a solar farm. Solar energy is zero-emissions in its own right, but suppose that a residential community had previously been buying the very same solar energy, and now that the hydrogen plant has stepped in, the community will have to buy electricity from a gas power plant. The resulting emissions are market-mediated: The hydrogen plant is indirectly responsible for the gas-fired emissions because it used the scarce solar resource without bringing any new solar (or other renewable energy) into existence. To extrapolate out from this example, whenever the subsidized activity relies on some input that other buyers might otherwise wish to procure, an indirect emissions analysis requires asking about the consequence of those buyers needing to find a replacement.

3. Additionality in Emissions Accounting

The requirement to count indirect and market-mediated emissions, discussed above, implicates the principle of additionality. Additionality is the principle that one should only be able to take credit for effects that occur on the margin, i.e., that would not have taken place absent the policy or action under evaluation. In carbon accounting, whenever someone claims that a production process uses only green energy, it is important to ask whether that green energy is additional to the overall energy system or whether it has been displaced from some preexisting green energy application. If the green energy would have existed anyway, it does not make sense to reward a subsidy claimant for using it. The additionality principle should also prevent taking credit for actions that were otherwise already required by law.[122]

Commentators have long recognized the additionality problem as a potentially severe flaw in tax subsidy programs.[123] When Congress subsidizes some behavior, there is often a possibility that most subsidy recipients would have conducted the behavior in the absence of a subsidy, and so the payments they receive are arguably wasteful.[124] For example, models produced by the National Renewable Energy Laboratory suggest that solar energy deployment would continue to expand and remain cost competitive even if tax credits were to expire, suggesting that those credits do not achieve additionality.[125] Tax policy analysts have often noted that it would be desirable to distinguish between activities that would have been conducted anyway and those that were triggered by a subsidy, and only pay for the latter group.[126] One relatively simple way to implement such a rule might be to subsidize only activities that exceed the taxpayer’s historical baseline. Congress designed the Research and Development Credit (R&D Credit) to work in just that fashion, indexed to increases in research expenditure over a firm-specific historical baseline.[127] Measuring increases over a firm-specific baseline remains a key operation in calculating the credit.[128] Multiple studies have found that each dollar of R&D Credit stimulates about a dollar in research investment. In other words, the credit achieves additionality.[129]

By making additionality an implicit requirement of consequential accounting, Congress, the Treasury, and the IRS are attempting to make firms responsible for indirect emissions that few previous climate policies have managed to internalize. The regulations under the Clean Electricity Credits interpret “significant indirect emissions” to include “market-mediated changes in related commodity markets.”[130] Similarly, the regulations under the Clean Hydrogen Production Tax Credit aim to incorporate “significant indirect emissions associated with electricity,” i.e., the market-mediated effect of creating new demand for electricity.[131] The regulations make clear that the Treasury interpreted Congress’s instruction as encompassing an additionality requirement, specifically that hydrogen plants must be powered by incremental, new clean energy sources, not preexisting clean energy.[132]

It is striking that the additionality-related safeguards in the Clean Hydrogen Credit and Clean Electricity Credits are much more administratively complex than the R&D Credit or any other additionality-related rules Congress or the IRS have attempted in the tax code to date. James Salzman and David Weisbach notice this asymmetry and ask why there is a “double standard” in the extent to which commentators worry about additionality in the world of carbon accounting and offsetting compared to the world of tax policy more generally.[133] They convincingly dispatch any distinction grounded in non-additional carbon offsetters being bad actors, in the difference between offsets preventing a bad versus subsidies achieving a good, or in offsets being quantity-based versus subsidies being price-based.[134] But Salzman and Weisbach do not address one important reason for taking additionality more seriously in the context of carbon accounting: the possibility that the intervention will not just fail to be additional but will actually subtract from the intended outcome.[135] Subsidies for green manufacturing (e.g., hydrogen production) often carry this risk: Not just that they will displace existing green energy sources but that dirty energy will fill the gap left by the displaced green energy, leading to higher net carbon emissions than in the absence of a subsidy. Green subsidies don’t merely risk wasting money; they risk perverse, self-defeating outcomes.

B. Who Does the Counting

For observers skeptical that tax law can accommodate carbon accounting, one major reason for doubt is the issue of administrative competence. How will the IRS, an agency with limited experience in environmental science or energy systems, enforce highly technical carbon accounting rules? Rather than administer domain-specific programs by itself, the IRS increasingly functions as a hub for gathering information and assessments from external authorities both within the executive branch and outside government. By relying on outside experts to make specific technical determinations, the IRS need not be limited by in-house capacity. Furthermore, just as it does in tax administration, the IRS relies wherever possible on information provided by independent third parties to cross-reference subsidy seekers’ claims.

Of course, climate policy is not alone in turning to tax law for its implementation. The tax code is also home to the federal government’s most significant undertakings in low-income housing policy, anti-poverty policy, and technological research and development policy. Congress relies so heavily on tax law for multiple reasons. For one, taxing and spending legislation enjoys a procedural advantage in overcoming the filibuster in a tightly divided Senate.[136] Second, as courts have become more skeptical of novel regulatory uses of the Commerce Clause, Congress has increasingly sought to wrap its regulatory efforts in the Spending Clause, where courts have remained relatively accepting.[137] Third, when it comes to disbursing money, it is superficially simpler for Congress to write an eligibility formula into the tax code rather than create a new program office and hire a workforce to deliver discretionary grants. We can observe the administrative difference between tax subsidies and nontax grants in the CHIPS and Science Act, which created parallel tax credits and grants for semiconductor manufacturing. The tax credit included no appropriation for additional program staff, whereas the Commerce Department received funds to hire approximately 140 people to administer semiconductor-related grants.[138] Of course, imposing an unfunded mandate on the IRS is not necessarily administratively superior in the long run. Doing so forces the IRS to rely on its peer agencies for technical support, as detailed in the rest of this Section.

Administrative law scholars have long recognized the theoretical appeal and empirical reality of interagency coordination.[139] Congress often directs agencies to coordinate with one another to share expertise. In tax rulemaking, Congress sometimes directs the Treasury to follow some other agency’s lead in defining a statutory term. For example, the semiconductor manufacturing tax credit (§ 48D) defines the critical term “legacy semiconductors” by reference to a simultaneously enacted provision that includes an equivalent concept, and Commerce Department regulations thereunder.[140] Likewise in adjudications, Congress sometimes directs the IRS to cross-reference data collected by other agencies as part of determining eligibility for a tax expenditure. For example, in administering the health insurance premium tax credit, the IRS coordinates with the Centers for Medicare & Medicaid Services in combining eligibility criteria from each agency’s respective databases.[141] And in guidance for prevailing wage requirements applicable to various tax credits, Congress directed the IRS to follow prevailing wage determinations published by the Department of Labor.[142] The widespread reality of interagency coordination challenges the notion that the IRS can only administer traditional tax laws.[143]

The IRS’s reliance on peer agencies is essential for adjudicating carbon accounting. The agencies most relied upon are the Department of Agriculture, the EPA, and the DOE, as detailed in the following. To start, in guidance under the Sustainable Aviation Fuel credit, the IRS announced that biofuel producers who use crops certified under the Department of Agriculture’s Climate Smart Agriculture pilot program will receive a reduction from the lifecycle emissions value otherwise calculated under the regulations.[144] The Climate Smart Agriculture program does not audit or monitor crops for any preexisting regulatory purpose, so an independent third-party verifier, rather than the Department of Agriculture, must certify compliance with the program.[145]

Relying on peer agencies is even more useful when the peer agency already collects relevant data under a preexisting program. For example, the IRS relies on the EPA for verifying the sequestration of carbon under the Carbon Sequestration Credit. In the course of developing its own Underground Injection Control program, the EPA had established a monitoring, reporting, and verification program to track carbon dioxide sequestered in underground wells.[146] The newest Treasury regulations require taxpayers to comply with either the EPA’s monitoring standard or a different, third-party monitoring standard.[147] This approach involving the EPA was created in response to a scandal where the Treasury Inspector General found that of over $1 billion in carbon sequestration credits claimed between 2010 and 2019, 87 percent of credits were claimed by taxpayers that did not have an approved EPA monitoring plan in place at the time the credit was claimed.[148] By setting eligibility for the tax credit based on demonstrated compliance with the EPA monitoring program, the IRS aims to avoid future discrepancies of this sort.

But among all the agencies, reliance on the Department of Energy is most critical to administering carbon accounting. The DOE has the technical expertise to build and evaluate life-cycle assessment (LCA) models, which play a key role in taxpayers’ claims for technology-neutral credits. For several of the green energy credits, the DOE has released a customized LCA model that satisfies the statutory requirement to measure lifecycle emissions. For example, the Sustainable Aviation Fuels Credit requires that LCA modeling be in accordance with either the Carbon Offsetting and Reduction Scheme for International Aviation, a model adopted by the International Civil Aviation Organization, or “any similar methodology” that satisfies the Clean Air Act definition of lifecycle emissions.[149] In response, the DOE developed an LCA model customized for modeling aviation fuel emissions.[150] This model modified the general-purpose GREET model, the DOE’s longtime standard-bearer emissions model. Likewise, the Clean Hydrogen Credit requires that lifecycle emissions be measured according to GREET “or a successor model (as determined by the Secretary).”[151] The DOE responded by developing a hydrogen-specific GREET model, 45VH2-GREET.[152] Treasury regulations provide that if the taxpayer’s feedstock and production pathway are included in the latest GREET, the taxpayer must use that model to determine its emissions rate.[153] The DOE has not yet developed a GREET model for electricity generation—likely the most daunting of these undertakings because of the vast range of possible technologies. But Treasury regulations provide a process for the Secretary to designate a suitable LCA model and for taxpayers to rely on emissions rates determined using such a model, suggesting that 45Y-GREET may be forthcoming.[154]

The DOE’s most significant role comes in adjudicating taxpayers’ petitions for emissions rates in circumstances not covered by the pre-approved GREET models. Both the Clean Hydrogen Credit and Clean Electricity Credits provide a process to petition the DOE for a “provisional emissions rate,” i.e., an emissions rate determined by the taxpayer’s own bespoke emissions modeling.[155] The petition process follows a template established by the Carbon Utilization Credit, which requires submitting an LCA report to the DOE for technical review.[156]

The DOE’s technical review of LCA reports is essentially the last line of defense in the distribution of green subsidies. Of course, the IRS can reject a credit application even after DOE approval, but it is hard to expect it would do so on technical, LCA-related grounds.[157] The DOE’s review is also the most opaque stage of the administrative process.[158] Taxpayers have been dissatisfied with the lack of visibility into the DOE’s review and have requested more information on emissions factors for comparison technologies (necessary for establishing baselines in LCA), materiality thresholds for possible deviation between taxpayer-conducted LCA and DOE models, and examples of approved LCA submissions.[159]

C.   How to Count: Performance Standards and Safe Harbors

The ambiguity and discretion that taxpayers decry in the DOE’s technical review process is no accident: It is the only way to adjudicate a scientifically contested performance standard like a GHG emissions rate. Performance-based regulatory standards work by defining a required outcome but leaving “the means of achieving that outcome to the discretion of the regulated entity.”[160] Put simply, performance standards are standards that can be achieved by unlimited potential means.

Performance standards differ from standards that specify exactly how a regulated entity must act—known as means standards, prescriptive standards, or design standards.[161] For example, the EPA’s regulation of diesel engine emissions obligated manufactures to pass an emissions test “while leaving it to each company’s discretion to determine how to attain that required outcome.”[162] Performance standards are praised for giving regulated parties flexibility to choose the least costly means of achieving the stated outcome.[163] In theory, this open-endedness should allow firms to innovate and develop new means of compliance over time without waiting for the law to change and accommodate them. Cary Coglianese refers to this feature of performance standards as allowing “longitudinal flexibility” (i.e., evolving methods of compliance over time), in addition to “cross-sectional flexibility” (i.e., enabling different firms to comply in different ways).[164] Performance-based regulation has been endorsed by Presidents of both parties and enshrined within the guiding documents of regulatory review.[165] At the 2009 Senate hearing that laid the intellectual groundwork for technology-neutral energy tax subsidies, an environmental economist testified that “[b]ecause of our limited ability to foresee technological solutions that are possible but do not yet exist, it is almost always more effective and economical to specify the energy or environmental objective rather than a specific means of achieving it.”[166]

Despite its justifiable appeal, performance-based regulation has distinct downsides, as often noted by administrative law scholars. For several reasons, performance-based standards are more difficult to administer than design standards.[167] To start, performance standards put pressure on legislators to choose the right measure of performance, because emphasizing performance on one metric will tend to undermine the importance of performance on other, not-explicitly relevant metrics. Next, performance standards invite conflict over adjudicating performance, whereas design standards tend to rely on cut-and-dry rules that are more easily adjudicated.[168] As a result, performance standards offer regulated parties less certainty and predictability about whether they will be considered compliant.[169] Performance standards will generally require monitoring, especially to the extent the standard invites parties to choose their own method of compliance.[170] Finally, performance standards may invite gaming or evasive behavior intended to narrowly satisfy the measurable standard while departing from regulators’ underlying goals. This is sometimes referred to as “teaching to the test.”[171] The end result is to increase the cumbersomeness and complexity of regulation.[172]

A GHG emissions rate is a particularly challenging performance standard to implement because there is no universally applicable formula for determining the emissions rate of an industrial process. There are agreed-upon modeling frameworks in the form of ISO standards and GREET models, but operating a model still requires countless judgment calls, customizations, and ad-hoc modifications. Modeling is relatively foreign to the tax law and to the IRS, which is why the key modeling responsibilities within emissions tax law are outsourced to the DOE and to modeling tools it publishes. Whoever conducts LCA must make decisions about the upstream and downstream boundaries of the system, allocating emissions among coproducts, establishing a counterfactual for the emissions that would have been caused by input products, and so on.[173] If the IRS and DOE are committed to treating GHG emissions rate as a performance standard, then they must assess each of these choices on a fully individualized basis. And unlike in attributional life cycle assessments, where third-party information sources can help verify taxpayer claims, there is no authoritative third-party source on a project’s consequential emissions.

In light of the uncertainty and intense regulatory workload of the performance standard approach, it is easy to see why both regulators and regulated parties might prefer to see relatively more rule-based adjudication and relatively less of the nebulous lifecycle emissions standard. To that end, safe harbors and unsafe harbors can provide a zone of rule-like certainty.[174] Safe harbors are common in tax law. For example, the test for whether tax law should treat a partnership as publicly traded is a facts-and-circumstances test that asks whether partners are able to buy, sell, or exchange their partnership interests in a manner that is economically comparable to trading on an established securities market.[175] But within that standard are two zones of relative certainty: a safe harbor based on easily verifiable measures of lack of trading (in which case the partnership is deemed not publicly traded),[176] and an unsafe harbor if partnership interests trade on an established securities market (in which case it is deemed publicly traded).[177] Likewise, inside the tax rule that excludes a portion of gain from the sale of a home when sold due to a change in one’s place of employment, there is a safe harbor that guarantees a sale will be deemed to be by reason of a change in place of employment if the “individual’s new place of employment is at least 50 miles farther from the residence sold or exchanged than was the former place of employment.”[178] Within 50 miles, a facts-and-circumstances test would apply.

As Susan Morse observes, the main effect of a safe harbor is to encourage behavior to converge on the boundary line just inside the safe harbor.[179] Those whose behavior would be noncompliant move into the safe harbor to become compliant and avoid regulatory penalties, while those who were already compliant move to the outer edge of the safe harbor (assuming that being “too” compliant is costlier than just barely complying).[180] This dynamic raises the costs of regulators defining an overly permissive safe harbor: Not only will firms that shouldn’t have received regulatory protection benefit from it, but firms that otherwise would have performed better than the safe harbor will worsen their behavior up to its boundary.[181] In the next Part, I examine how the performance standards created to ease the administration of carbon accounting risk encouraging firms into carbon shelters.

III. The Problem of Carbon Shelters

What could go wrong with universal carrots? The most likely answer is that they will become too universal and enrich high-carbon firms. At best, that would mean wasting some of the money appropriated for the green transition on a counterproductive use. At worst, it would mean subsidizing emissions that outweigh the emissions reductions purchased with other carrots. The worst-case scenario would elevate the critique of universal carrots from futility to perversity.[182] In this Part, I describe the carbon shelters that lawmakers should aim to avoid: transactions that win green-subsidy money despite generating higher emissions than the law intends to allow.

Carbon shelters arise due to the tradeoff between accuracy and administrability in carbon accounting. As explained in the previous Part, carbon accounting is susceptible to gaming because it is a form of performance-based regulation. In the first instance, carbon shelters form because it is difficult for regulators to estimate the attributional and consequential emissions of each taxpayer’s bespoke submission. Regulators respond by crafting safe harbor rules to ease the administrative challenge. However, these rules often create additional, legally authorized carbon shelters.

I argue that regulators should err on the side of caution in their pursuit of administrability. This means disqualifying applicants who potentially have high emissions, even though some members of this category will turn out to have acceptably low emissions. Greater risks of subsidizing net-positive emissions justify greater caution. Because it is generally cheaper for firms to produce energy with higher emissions than lower, the availability of a high-emissions way to qualify for carrots threatens to swamp the attractiveness of the low-emissions options. And because most green tax credits are uncapped, there is no limit to the number of high-emission projects that could be inadvertently subsidized.

A. From Tax Shelters to Carbon Shelters

The bogeyman of tax law is the reviled tax shelter. A suitable working definition of a tax shelter is a tax-motivated transaction that produces a tax loss in excess of any economic loss, in a manner inconsistent with legislative intent or purpose.[183] A tax shelter is an exploitable loophole—a way for a taxpayer to have their cake (reducing tax liability) and eat it too (without reducing economic income). In the realm of carbon accounting, a comparable problem is likely to emerge: carbon shelters. I define a carbon shelter as a transaction where a taxpayer receives a green subsidy despite producing carbon emissions in excess of that supposedly allowed by the subsidy.

To illustrate, consider one of the largest possible carbon shelters lurking inside Congress’s green subsidies: the risk of the Clean Hydrogen Credit subsidizing hydrogen electrolysis powered by a fossil-dominated grid.[184] One study estimates that without proper additionality requirements, the credit could fund tens to hundreds of millions of tons of new GHG emissions annually at a cost of $30 billion annually to the federal government.[185] From the taxpayer’s perspective, a carbon shelter like this is simply a shortcut to getting paid by the government without having to do the hard work of reducing emissions. That is, a carbon shelter is desirable for the taxpayer on the assumption that it is cheaper to produce higher emissions rather than lower. From the government’s perspective, a carbon shelter is an own goal: paying taxpayers to do something socially undesirable.

At a high level, taxpayers can form carbon shelters by violating any of the carbon accounting principles introduced in Part II. Subsidy seekers can shelter emissions by outsourcing indirect emissions to parties outside the scope of the subsidy (undermining attributional accounting), attributing emissions to byproducts and waste products (same), or by offsetting emissions with low-carbon inputs that would have been produced anyway (undermining consequential accounting). In theory, attributional and consequential lifecycle assessment are meant to catch these failure modes. But lifecycle assessment is imperfect, and some emissions fall through the cracks. Moreover, when regulators erect overinclusive safe harbors, they give taxpayers explicit permission to produce emissions that violate the green subsidy principles. Every safe harbor provides a corresponding carbon shelter.

It is important to recognize some key differences between the concepts of tax shelters and carbon shelters. First, while there is a widespread view that taxpayer motive or purpose should be one of the primary indicia of a tax shelter, my use of “carbon shelter” is not limited to transactions entered exclusively for the purpose of obtaining subsidies.[186] Tax shelters generally involve claiming tax losses, and the steps required to generate tax losses tend to have little appeal outside that context. Conversely, carbon shelters generally involve producing energy in a high-emission manner, which is a common business practice with or without the allure of subsidies. Therefore, unlike with tax shelters, it is not useful to analyze the taxpayer’s purpose or intent in determining whether a transaction is a carbon shelter.[187] Taxpayers might rationally pursue the same high-emission practices with or without the possibility of green-subsidy incentives, even though green subsidies can make those practices more profitable. Most carbon shelters reflect a policy failure, not any rule-bending fault on the part of the taxpayers who benefit.

The second important difference is that, unlike tax shelters, carbon shelters are, in general, not the result of ambiguous and exploitable statutory language. Tax scholars have long analyzed tax shelters as problems caused by formalist statutory interpretation, where taxpayers seize on ambiguous language to achieve outcomes that Congress did not intend.[188] When the problem is understood that way, the solution is to invent canons of pragmatic interpretation to recover Congress’s intent.[189] For example, one of the most important weapons in Congress and the IRS’s fight against tax shelters has been the “economic substance” doctrine, an influential anti-abuse principle that allows the IRS to disallow transactions that do not change the taxpayer’s economic position or lack a substantial nontax purpose.[190] To be sure, Congress has created carbon shelters through statutory ambiguity as well. In the “black liquor” scandal, paper firms claimed the alternative fuels mixture tax credit, which provided a subsidy for mixing in biofuels to petroleum fuels by adding petroleum fuel to black liquor, a wood byproduct they had historically used for fuel.[191] That is, the paper firms seized on the ambiguity of the term “mixture” and got paid to make their existing biomass fuel more carbon-intensive—the exact opposite of the type of mixture Congress meant to encourage.[192]

But that scandal is not the sort we should expect in the new era of a zero-emissions performance standard. The paper firms’ gambit could not plausibly meet a standard like “greenhouse gas emissions rate not greater than zero” or the similar language found in all technology-neutral green subsidies—to the credit of the performance-based approach. Congress has now provided language that justifies the most rigorous possible carbon accounting, with the exception of certain carveouts like the exclusion of indirect land use emissions under the Clean Fuel Credit.[193] Instead, for the most part, the current generation of carbon shelters will be made possible by overly permissive administration (either in granting safe harbors or in failing to apply the underlying performance standard).

Even if the new generation of carbon shelters do not arise from ambiguous statutory language, Congress might still have an important role to play in closing them. In tax law, one of the most powerful approaches to attacking tax shelters has been the enactment of substantive default rules ruling out broad categories of behavior that, while not necessarily abusive in every instance, are likely to attract shelters.[194] The prototypical example would be Section 469, the “passive loss rules,” enacted under the Tax Reform Act of 1986.[195] Those rules stop taxpayers from claiming losses from business activities in which they do not materially participate, a strategy that was a dominant feature of tax shelters in the 1970s and 1980s.[196] In a subsequent era, Marvin Chirelstein and Lawrence Zelenak proposed a “silver bullet” rule disallowing deductions in excess of a measurable reduction in the taxpayer’s net worth or that involve the allocation of noneconomic income to a tax-indifferent party.[197] Whether or not that was the right rule for its day, the structure of the proposal is notable. Chirelstein and Zelenak described the contours of tax shelter transactions and proposed a deliberately broad, overinclusive rule to exclude such transactions.

In the next Section, I will examine the subset of carbon shelters that arise due to overinclusive safe harbor rules. Then, in Part III.C, I will draw on the Chirelstein-Zelenak example to develop a broad anti-abuse rule for carbon shelters.

B. Shelters in Safe Harbors

As discussed at the end of Part II, some carbon shelters are explicitly granted to taxpayers in the form of safe harbor rules. These rules give permission to not conduct a thorough and rigorous carbon accounting. The more a green-subsidy law relies on safe harbors, the greater the risk of a divergence between true carbon emissions and the emissions counted by law. If the safe harbors are overinclusive, they will allow activities that produce higher-than-desired emissions. The implications of being underinclusive are less damaging because activities not covered by the safe harbor will still be evaluated according to the underlying performance standard.

1.   Technology Safe Harbors

The Treasury and IRS often create safe harbors meant to provide a regulatory fast track for energy production technologies considered to be presumptively low emission. An important production safe harbor appears in the Clean Electricity Credits, where a set of electricity generation modalities are labeled as “Non-Combustion & Gasification” (Non-C&G) and treated as categorically having a GHG emissions rate of zero.[198] These include wind, solar, hydropower, geothermal, and nuclear fission. It is easy to understand why the Treasury and IRS would not want to spend resources measuring modest differences in the emissions intensity of these technologies, which are generally understood to be very low emission compared to most combustion and gasification alternatives.

But there are potentially significant differences in the emissions intensity of these technologies. In the case of nuclear fission, large amounts of electricity may be used to enrich uranium. The carbon emissions associated with that input fuel should be counted just as the emissions associated with converting water into hydrogen are counted under the Clean Hydrogen Credit.[199]

Likewise, the operation of a hydroelectric dam can yield “reservoir emissions,” which occur when the flooding of a dam area submerges organic matter. The matter decomposes, releases methane, and emits “degassing emissions” when methane-rich deep water is pulled through the turbine.[200] The Clean Electricity Credits preamble acknowledges but disclaims responsibility for these emissions on the grounds that they “are not created by the fundamental transformation of potential energy in flowing water into electricity.”[201] This argument seems especially weak with respect to degassing emissions, which occur when flowing water powers a turbine. The argument could be more reasonable if the creation and maintenance of a reservoir is treated as separate from the operation of the attached power plant. But as advocates point out, taxpayers may claim the ITC on the value of the entire hydroelectric complex, including the reservoir, and therefore should have to count reservoir-related emissions.[202]

In both the nuclear and hydroelectric cases, it seems that a desire for simplicity undercounts relevant emissions (neither of which are unusually difficult to measure). There may be good reason to treat emissions below some de minimis threshold as equivalent to zero emissions for subsidy eligibility purposes, but one cannot apply a de minimis rule without measuring emissions in the first place. To avoid creating a carbon shelter, the regulations should require nuclear and hydroelectric plants to measure their emissions and submit LCA documentation just as combustion and gasification plants must.

Even less justifiable is Congress’s decision to exempt one additional technology, fuel cells, from the carbon accounting procedures of the Clean Electricity Credits entirely. A fuel cell is a machine that converts the chemical energy of a fuel source into electricity. In the OBBBA, Congress specified that fuel cells’ eligibility should be determined “without regard to” a greenhouse gas emissions-rate analysis.[203] Most fuel cells are powered by hydrogen, which raises the same questions about the emissions intensity of upstream hydrogen production discussed throughout this Article.[204] For all the issues with measurement of emissions induced by hydrogen production discussed in this Article, at least the 45V hydrogen regulations try to measure emissions. The credit for fuel cells effectively provides a backdoor hydrogen subsidy that circumvents the Clean Hydrogen Credit’s entire enforcement apparatus.

2. Temporal Safe Harbors

The Treasury and IRS also create safe harbors to simplify their responsibilities in overseeing subsidized projects over long periods of time. The tax authorities sometimes provide that after a certain duration has passed, applicants are no longer responsible for maintaining the low GHG emissions rates which initially qualified them for subsidies.[205] A temporal safe harbor is particularly relevant in case of the ITC, which is awarded when property is placed in service but has not yet begun operating. Before the start of operations, there is no observed GHG rate for regulators to measure; there is only an “anticipated GHG emissions rate.”[206]

There are three main conceptual issues in holding applicants to a performance standard based on future behavior. First, what indicia observable at time t=0 are sufficient to establish that the anticipated emissions rate will be not greater than zero? Second, for how long must the facility’s emissions rate be anticipated to remain at or below zero? Third, assuming that the tax authorities will continue monitoring the facility’s emissions in the future, for how long will the facility be subject to credit recapture for violating its anticipated GHG rate?

The Clean Electricity Credits Rule answers the first question with an open-ended performance standard and the second and third questions with safe harbors. The performance standard allows the IRS to consider all facts and circumstances to determine an anticipated emissions rate. These include physical characteristics of the facility and the existence of long-term contracts to purchase low-carbon fuels or to capture and dispose of carbon.[207] But the performance standard is subject to a safe harbor.

For facilities that use feedstock fuels or carbon capture, the facts and circumstances must indicate that the facility will operate with net zero emissions for ten years.[208] Anticipated increases in emissions after ten years do not count against the anticipated emissions rate. While the rule seems to cover ten years of emissions, the provision for recapturing credits when recipients exceed the allowable emissions rate only operates for five years after the facility is placed in service.[209] The five-year recapture window allows an applicant to qualify for the ITC on the basis of a ten-year contract to purchase low-carbon fuels but then cancel the contract after five years and suffer no penalty.

Awarding subsidies based on an anticipated emissions rate is necessarily an uncertain business, and it would not be reasonable to ask the government to predict emissions much further than ten years into the future. In that light, the first safe harbor discussed above is appropriate. But a safe harbor based on anticipated behavior over ten years is not worth the name if the subsidy recipient is free to change plans after five years. The solution is a recapture period that matches the duration over which the subsidy is meant to regulate behavior: ten years. No safe harbor (in the form of freedom from recapture) should be granted for behavior between years five and ten.

3.   Model Safe Harbors

Model safe harbors are created by the erroneous background assumptions of government-provided emissions models. For most green tax subsidies, the IRS and the Treasury have directed taxpayers to use specialized LCA models created by the DOE; in other words, the regulations incorporate the rules embedded in the models.[210] These models are customized versions of the GREET model, the original lifecycle model developed by the DOE to measure the emissions impact of transportation fuels. In response to the Inflation Reduction Act, the DOE developed 40BSAF-GREET, a specialized model for aviation fuel, and 45VH2-GREET, a specialized model for clean hydrogen. The process of developing an official DOE model for the electricity sector is still underway (this will be a significant departure from prior GREET models, all of which dealt with fuels).

These models work by combining user-inputted data with background assumptions and formulas provided by the DOE. The user inputs characteristics of its production process such as the type of feedstock it is using to produce hydrogen, the quantity of power inputs used in the production process, and the quantity of output yielded (this takes place in an Excel spreadsheet).[211] Using DOE formulas, the model combines these parameters with background assumptions about the carbon intensity of specific production processes in order to output Scope 1, Scope 2, and Scope 3 emissions values. The results depend enormously on the background assumptions.

In particular, the GREET models face a tension between allowing users to input parameters specific to their own operations and fixing parameters in line with what the DOE believes to be nationally representative or, more conservatively, in line with a worst-case scenario. For example, the hydrogen GREET model includes a parameter for the upstream methane leak rate associated with processing and transporting natural gas, because natural gas can be used to power the electrolyzers that produce hydrogen. The higher the leak rate, the higher the carbon intensity of hydrogen production. The current version of 45VH2-GREET uses a fixed nationwide leak rate of 0.9 percent despite evidence that the leak rate varies dramatically by region and producer.[212]

Using the fixed nationwide leak rate means that a project with an abnormally high rate could qualify for the same subsidy as a project with a very low one.[213] To combat this, industry representatives have generally advocated for using site-specific emissions intensity data out of a belief that the nationwide averages tend to lag behind improvements producers make on the ground.[214] Allowing taxpayers to submit their own measurements on a variety of technical parameters, however, would mean that the IRS and DOE would have to audit and verify these numbers—a difficult administrative challenge. In the preamble to the final Clean Hydrogen Credit Rule, the Treasury and the IRS indicated that a future version of 45VH2-GREET would require taxpayers to use their own methane leak rate.[215]

For the sake of administrative simplicity, the Treasury and IRS may find it necessary to keep using fixed parameters in LCA models. In that case, they should nonetheless question whether the nationwide average of a given measure is the right value to fix. If one is risk-averse, the parameters should instead be set at the worst-case option among plausible values. This would set up the model so that worst-case users receive no safe harbor in qualifying for the subsidy.

4. Safe Harbors from Emissions Induced Elsewhere in the Economy

A final category of safe harbor exists in the rules meant to ensure that subsidy recipients do not induce emissions elsewhere in the economy. As introduced in Part II.A.2, the additionality principle requires ensuring that subsidies generate new clean energy rather than moving around clean energy sources that would have been used anyway. The Treasury and IRS’s attempted solution has been to establish proxies for additionality; however, these proxies function as safe harbors because they do not actually guarantee additionality.

In the case of the Clean Hydrogen Credit, the Treasury created a safe harbor that can be accessed by low, if not necessarily zero, market-mediated emissions.[216] The most likely reason for this choice is that attempting to model the marginal grid emissions associated with running each electrolyzer would be difficult to administer.[217] Instead of modeling, the Treasury’s regulations require hydrogen producers prove that their electricity supply satisfies “three pillars.” The electricity supply should be (1) procured from clean power plants in the same region, (2) procured from clean power plants built approximately contemporaneously or after the hydrogen plant, and (3) procured during the specific hours those clean plants are operating.[218] Together, the pillars are meant to ensure that hydrogen electrolysis uses clean electricity that other users did not previously rely on.

Still, despite the three pillars, hydrogen electrolysis can still induce additional grid emissions.[219] Buying power from a new, renewable power plant does not mean that no one else might have wanted to purchase that same clean power at the same moment. If renewables were scarce at that moment, the hydrogen plant using those resources may have pushed others to use fossil resources.[220]

Moreover, the Treasury and IRS offered additional safe harbors in the final clean hydrogen rule. Electricity will count as “incremental” (i.e., the second pillar above) if procured in a state where state law makes induced-grid emissions unlikely, as reflected in a legally binding state GHG emissions cap, or if procured from a nuclear reactor at risk of retirement.[221] These exceptions to the general incrementality rule are meant to reflect circumstances where induced emissions are relatively unlikely but do not guarantee that result. For example, electricity demand is projected to grow over the next decade. A nuclear reactor that was financially struggling in the period 2017–2021 (the period referenced by the rule) might nonetheless have multiple new sources of demand such that selling nuclear-generated electricity to a subsidized hydrogen project might indeed induce emissions elsewhere on the grid because other would-be consumers of the nuclear plant’s energy would need to use fossil fuels.[222]

When it comes to the Clean Electricity Credits, the concern is that biofuel feedstocks that are combusted to generate electricity may induce emissions elsewhere in the economy. Because biofuels and fossil fuels are substitutes, subsidies that increase demand for and reduce the price of biofuels will cause the price of fossil fuels to decline, which could cause more overall energy consumption (including more fossil energy consumption). If the prices of fossil fuels drop far enough, those fuels could displace consumption of renewable energy.[223] Likewise, if biogas is diverted from existing economic uses to produce electricity, other users would need to produce more biogas to replace it.

The Treasury and IRS’s current approach to emissions induced by biogas creates another safe harbor. In the Proposed Clean Electricity Credit rules, the Treasury and IRS contemplated requiring that biogas used to generate electricity originate from the “first productive use” of its biogas source.[224] Of course, there is a limit to this logic: If demand for biogas is increasing economy-wide, the fact that a source of biogas is new, with no prior productive use, does not at all prove that it would not have been used for some other purpose.[225]

In the final rule, the Treasury and IRS dropped the “first productive use” idea and instead specified acceptable “alternative fates” to assign to the biogas, depending on its source (e.g., methane from landfills vs. wastewater plants vs. animal waste).[226] “Alternative fates” is the Treasury’s term for counterfactual scenarios, specifically the emissions that the biogas source would have caused if it had not been used for electricity. Specifying alternative fates at the industry level—rather than requiring each facility to use the actual alternative fate of its biogas—is yet another safe harbor. Any facility whose actual practice is less carbon intensive than the industry norm (e.g., flaring rather than venting, or selling rather than flaring) gets to claim that they have saved the methane from the more carbon-intensive alternative fate and thereby get more credit for using the biogas for electricity.

The only area where Congress has weighed in on additionality analysis is the Clean Fuel Credit, where, in the OBBBA, Congress forbade the inclusion of emissions caused by land use changes.[227] This rule is a boon for producers of ethanol, palm oil, and other fuels, who will no longer be penalized for the deforestation-related emissions caused by pulling new land into feedstock production.

Rather than rely on proxies for induced emissions, Congress, the Treasury, and the IRS should mandate the use of general equilibrium economic models to compute the market-mediated effects of using clean fuels and clean electricity in credit-eligible activities. These models represent the best technique available for estimating the effect of consuming one resource (e.g., biofuel) on substitutable products.[228] There is precedent for using these models: Existing government models include the EPA’s SAGE model, the Energy Information Administration’s National Energy Modeling System, and the Department of Agriculture’s food system general equilibrium model.[229] If possible, the DOE should incorporate such models into its GREET models to provide an interface for calculating market-mediated emissions effects (currently, GREET only addresses only indirect land use changes and not other indirect emissions). Requiring general equilibrium economic modeling would ensure stringent adherence to the performance standard.

C.   Shelters in Carbon Offsets

Closing overly permissive safe harbors is just the first step in policing carbon shelters. It should be the easier half of the battle, since at least safe harbors are simple to identify, having been explicitly granted by regulators. The harder half of the battle is discovering and foreclosing the carbon shelters that taxpayers invent through creative carbon accounting.

The challenge here—which this Section confronts—is for the IRS to avoid being continually one step behind. For every carbon shelter discovered, a new one will be invented. As we have learned in the tax shelter context, a whack-a-mole strategy of outlawing one specific fact pattern at a time is not viable.[230] The alternative is to come up with a broad anti-abuse rule that can apply to varied fact patterns. The goal should be to strike a middle ground between specifying a laundry list of per se problematic fact patterns (“whack-a-mole”) and a rule so broad that judicial discretion is needed to apply it. In the tax shelter context, Chirelstein and Zelenak’s proposal for a rule against noneconomic deductions was intended to be broader and more prospectively useful than a rule focused on the varied Code sections exploited by thematically similar tax shelters (e.g., installment sales in ACM Partnership and tax-free incorporations in Black & Decker[231]) but narrower and more predictably applied than the “economic substance” doctrine.[232] In this Section, I propose that the most vital anti-abuse rule to protect against carbon shelters is as follows: Firms should not be able to use carbon offsets (as defined below) to reduce their GHG rates.

Just as taxpayers “game” the income tax by generating noneconomic losses (i.e., negative income), firms similarly game carbon accounting by generating misleading negative emissions. Negative emissions, or carbon offsets, serve to reduce a taxpayer’s emission rate, thereby allowing them to use some quantity of positive-emission techniques or inputs and still qualify for a net GHG rate of zero.

There are three main ways that taxpayers might seek to offset emissions. First, they might purchase carbon offset credits (i.e., a legal record representing emissions reductions entirely unrelated to the taxpayer’s subsidy-eligible project). The Clean Electricity Credits Rule disallows the use of offsets unrelated to the production of electricity or of any input fuel. Thus, this first option is currently off the board.[233] Second, taxpayers might aim to take credit for using input fuels with supposedly negative emission rates. The idea behind a negative emission rate is that if some input source had not been used for electricity, it would have generated emissions. Thus, the resulting lack of emissions should be treated as even better than zero emissions, but negative relative to the counterfactual. The use of negative-emission fuels to offset other emissions is endorsed by the Rule, at least for certain sources of biogas like landfills, wastewater, and farm animal waste.[234] Third, taxpayers might conduct negative emission activities, like carbon capture, at their own facility. Carbon capture is endorsed by the Rule as a legitimate way of reducing the GHG rate.[235]

At a high level, there are two potential objections to giving subsidies for offsets. First, that giving subsidies for offsets risks violating the additionality principle. Second, that even if it satisfies additionality, technology neutrality is meant to reward innovation, and it violates the spirit of green subsidies to reward emissions reductions that don’t come from more efficient energy generation technology.

1.   Offsets Can Violate Additionality

First, consider the additionality issue. Vast literatures in environmental science and law have documented the pervasive problem of carbon offsets overpromising and underdelivering true additional emissions reductions.[236] The exact reasons differ for each type of offset.

The use of biogas as a fuel source for electricity generation risks becoming the next important example of overpromising and underdelivering. Biogas is produced by the decomposition of organic matter from landfills, livestock operations, and wastewater treatment.[237] Each of these processes has the potential to leak methane (a greenhouse gas) into the atmosphere. Nonetheless, biogas is often assigned negative emission rates based on the assumption that the methane would have escaped into the atmosphere if it had not been used as fuel.[238] There are at least two potential flaws in this logic: that the assumption that the methane would have escaped is unfounded and the negative emission rate therefore overstated;[239] and that because methane is delivered via natural gas pipelines, it is hard to know how much methane (as opposed to gas) a taxpayer is actually combusting.[240]

Since the point of claiming credit for negative emissions is to offset one’s positive emissions, letting taxpayers claim emissions below zero will allow those taxpayers to produce positive emissions elsewhere in their production process and still receive a tax credit for balancing out to zero emissions on net. For example, analysts have calculated that using just a 25 percent share of methane alongside 75 percent fossil gas could be treated as net zero, due to the extreme negative emissions value attributed to some methane.[241] If the logic that justified the negative emissions turns out to be faulty, taxpayers would get a subsidy for producing energy using (mostly) natural gas.

We can now reevaluate the first two distinct methods of offsetting emissions in light of the additionality issue. Both unrelated offsets and offsets pertaining to input fuels like biomethane are susceptible to failure because of the risk of faulty additionality logic. Even worse, because offsets justify the inclusion of fossil fuels in a purportedly net-zero energy system, even a small error in additionality can lead to subsidizing net-positive emissions. The upshot of this realization is that blending negative emission fuels to a facility’s tested emissions rate should be forbidden. Stated differently, the minimum emissions rate that should be assigned to a putatively negative emissions feedstock like biomethane should be zero (never negative). This would mean that biomethane could not be blended with any positive-emissions fuel and still achieve net zero.[242] Indeed, the most valuable carbon accounting provision of the One Big Beautiful Bill Act was an update to the Clean Fuel Credit such that “the emissions rate for a transportation fuel may not be less than zero.”[243] The same change was not made for the Clean Electricity Credits.[244]

2.   Offsets Aren’t Innovation Policy

But even supposing for the sake of argument that offsets are truly additional, there is a further question of whether it is appropriate to credit them against a subsidy seeker’s emissions rate. This is not a climate science question but a policy one: To what extent does it achieve the goals of green-subsidy laws to generate emissions reductions disconnected from the subsidized activity?

Here, there is an analogous issue within tax law. When tax outcomes depend on whether taxpayers fall above or below a minimum effective tax rate, like under U.S. law’s Global Intangible Low-Taxed Income (GILTI) or the OECD’s proposed Pillar 2, there is a question of whether firms should be allowed to combine income earned and taxes paid across multiple countries to calculate the effective tax rate.[245] Under current federal law, firms are able to aggregate income from multiple countries, effectively using excess foreign tax credits from high-tax countries to soak up potential tax liability from low-tax countries.[246] If the GILTI tax were applied on a country-by-country basis, i.e., without aggregating, income earned in a low-tax country would be subject to tax, which is arguably the purpose of a minimum-tax policy like GILTI.[247] The policy purpose of GILTI is for firms to face an incentive to leave low-taxed jurisdictions. Instead, the current rule means that firms can satisfy GILTI by aggregating themselves into a corporate structure with activities in high-tax jurisdictions.[248]

Similarly, if the purpose of a green subsidy is to improve the emissions efficiency of individual power plants, electrolyzers, and fuel producers, linking together those industrial processes with unrelated emissions reductions does not advance the policy goal. This is clearest if one views the green subsidies as innovation policy, with a purpose of catalyzing improvement in power and fuel production.[249] That is a different goal than merely paying for net-zero energy, and indeed there are other policies—including the Carbon Sequestration Credit—directly targeted at achieving negative emissions. Notably, the credit pays less per ton of sequestered carbon than the Clean Hydrogen Credit for an equivalent emissions reduction. This suggests that Congress views offsets as less valuable than energy production efficiencies.[250]

We can now reevaluate carbon capture as a variant of the negative emissions problem. On the one hand, if a firm captures and sequesters carbon generated at their own facility, there is less of a concern about the process failing to satisfy net zero and thus implicating the additionality concern.[251] From an innovation policy perspective, however, giving out green subsidies based on carbon capture blunts the primary purpose of those subsidies.

Lastly, and fundamentally, this is an argument for preserving distinct policy instruments, not an argument against carbon capture on the merits. Preserving distinct policy instruments is important because it is uncertain which technological course (better carbon capture vs. more efficient renewable generation) will prove more valuable in the long run. Offsetting collapses such distinctions. It also turns targeted policies into more diffuse ones. In that light, a broad anti-abuse rule could state that carbon captured in the process of electricity generation should not be allowed to offset that facility’s GHG emissions rate for the purpose of the Clean Electricity Credits, nor should carbon captured while producing input fuels that feed electricity generation count against the emissions rate.[252]

D.   Asymmetric Risk Justifies Anti-Abuse Rules

The previous Section illustrated the logic of developing an anti-abuse rule to prevent carbon shelters. A defining feature of that logic was caution: Because it is impossible to anticipate each carbon shelter fact pattern in advance, the rule must extend across a broad range of fact patterns, not all of which will turn out to be shelters.

The downside of such an anti-abuse rule is that it will be overinclusive, punishing activities that should have been allowed under careful application of the underlying performance standard.[253] In the tax context, that would mean imposing tax on transactions that Congress meant to grant as losses, even if noneconomic.[254] Similarly, Congress requires taxpayers to disclose their participation in “reportable transactions”: transactions that it considers likely to be abusive but are not necessarily abusive in every case.[255]

In the carbon accounting context, using overbroad prohibitions would mean denying green subsidies to applicants whose production process might, on close inspection, turn out to be zero emission.[256] That result contradicts the aspiration to technology neutrality. If, in order to win a subsidy, the taxpayer must not only achieve zero emissions but also avoid guilt by association with a questionable practice like carbon offsetting, they no longer enjoy the flexibility of technology neutrality. The question is whether it is appropriate to sacrifice the absolutist version of technology neutrality in order to prevent carbon shelters.

The precise tradeoff is between the harm of denying subsidies to deserving recipients against the benefit of preventing subsidies from going to undeserving recipients. Framed this way, the question is whether false positive or “type I” errors (a good project is not subsidized) are likely to be more consequential than false negative or “type II” errors (a bad project is subsidized).[257] If one were agnostic about the relative prevalence and magnitude of type I and type II errors—meaning that the IRS would err an equal amount in either direction—then one would proceed to apply the performance standard with no need for special caution.[258] But the nature of carbon accounting suggests that the distribution of errors would probably not be symmetrical.

The brief history of carbon accounting suggests an ongoing pattern where modelers progressively learn of new causal pathways between energy systems and their consequential emissions.[259] As recounted in Part II.A, a previous era of carbon accounting did not recognize emissions resulting from indirect land use changes, but contemporary practice does (even when Congress, for unknown reasons, refuses to count those emissions). Presumably there will be more revelations in the years ahead.

This pattern is especially likely to repeat when a new energy source enters the equation, as is intended to happen under technology neutrality. Based on experience, the first attempt to model a new energy source rarely uncovers all consequential emissions.[260] Of course, it is theoretically possible that advances in carbon accounting will instead reveal that consequential emissions of energy systems are actually lower than modelers currently think. But this is not very plausible as applied to new energy sources, which will be chosen on the basis of seeming to have low or zero emissions and will be treated as such until proven otherwise. The IRS will therefore tend to err on the side of underestimating emissions for each new energy technology before iteratively correcting such errors.

It is nonetheless possible that even if administrative errors are more common in the direction of underestimating emissions, it might still be desirable to be neutral between type I and type II errors. If the magnitude of the emissions reductions gained by approving more good subsidies (i.e., minimizing type I errors) is greater than the emissions improperly subsidized through type II errors, the government would still come out ahead. Which way the scales are likely to balance depends on the exact unsafe harbor rule under consideration.

But, in general, the magnitude of type II errors is likely to be greater for two reasons. First, high-emission firms will tend to have a cost advantage over low-emission firms, because reducing emissions tends to be costly.[261] Thus, if subsidies are made available for both, the high-emission firms will maintain their advantage in product markets and capital markets. For that reason, it should be the goal of green-subsidy policy to create advantages for the low-emission firm, not to give equal carrots to all.[262] Second, the case for overcoming bad subsidies with good ones is especially weak if we suppose that there is a limit on how much can be spent on green subsidies. If there is only (say) $100 billion to spend on green subsidies, one should not waste any of that budget on projects with positive (or higher than allowable) emissions.[263] Instead, one should try to spend the entire budget on projects that are all zero emission, assuming such projects are available. In other words, a budget constraint increases the opportunity cost of committing type II errors.

Choosing deliberately overbroad anti-abuse rules like the rule against offsets proposed above necessarily sacrifices the absolutist version of technology neutrality. For example, if a taxpayer is not allowed to claim negative emissions for using captured methane, a power plant fueled by a blend of biogas and natural gas will not be eligible for tax credits because the blended emissions rate will not reach zero.[264] This would be understandably frustrating to the developers of a blended biogas plant confident that their counterfactual emissions estimate is correct and that their plant truly achieves net zero. But a responsible version of technology neutrality does not require the view that error costs are symmetrical between false positives and false negatives. Instead, technology neutrality should be compatible with a high standard of evidence of the sort embodied in a cautious anti-abuse rule. An openness to innovation, rather than blind permissiveness to risky projects, is the core of technology neutrality that energy tax law should preserve.

Conclusion

The principal appeal of universal energy policies is that they are supposed to be “technology-neutral,” or indifferent between any two technologies that reduce carbon emissions by the same degree. The only way to achieve technology neutrality is to estimate the carbon intensity of every project potentially eligible for carrots. This is the role of carbon accounting, the discipline of assigning carbon emissions to responsible parties. While existing scholarship treats carbon accounting as a problem for environmental law, it is instead tax law where carbon accounting faces its highest-stakes challenge yet. This Article argues that carbon accounting is surprisingly well suited to tax law because the normative and analytical principles behind carbon accounting bear close resemblance to principles within tax law. Concepts from tax theory including a comprehensive tax base, additionality, liability shifting, and rate blending can usefully inform the development of parallel doctrines within emissions accounting.

And yet, just as the income tax is susceptible to “tax shelters” that technically comply with the rules but substantively avoid tax, so too will firms develop “carbon shelters” that qualify for green subsidies while covertly making use of high-emission energy. This Article identifies the features of emerging carbon accounting law that facilitate carbon shelters and proposes reforms to close them down. Because of the difficulty of anticipating every carbon shelter in advance, an antishelter strategy needs deliberately overbroad prophylactic rules, including some modeled on similar rules from tax law, like the passive loss rules. If policymakers are to avoid inadvertently subsidizing unlimited emissions, they must be prepared to sacrifice some of the technology neutrality that motivates universal carrots in the first place. Technology neutrality is a means of reducing emissions; it should not be treated as an end in itself when it is no longer serving that goal.

‍ ‍

‍Copyright © 2026 Jeff Gordon, Assistant Professor, Vanderbilt Law School. I am grateful to Alice Abreu, Anne Alstott, Johnny Buckles, David Carel, Brian Galle, Zach Liscow, Joshua Macey, Daniel Markovits, Alan Mitchell, Morgan Ricks, Jim Rossi, Peter Salib, Jeff Schoenblum, Amy Sinden, Michael Vandenbergh, Larry Zelenak, Jon Zytnick, and participants at the Georgetown Tax Law and Public Finance Workshop and Northwestern Tax Law Colloquium for helpful comments and conversations. I thank the talented editors of the California Law Review for their hard work which has greatly improved the Article.

[1].     See, e.g., Brian Galle, The Tragedy of the Carrots: Economics and Politics in the Choice of Price Instruments, 64 Stan. L. Rev. 797, 812 (2012); see also Gerrit De Geest & Giuseppe Dari-Mattiacci, The Rise of Carrots and the Decline of Sticks, 80 U. Chi. L. Rev. 341, 354–55 (2013) (on the distinction between carrots and sticks).

[2].     See Gilbert E. Metcalf, Tax Policies for Low-Carbon Technologies, 62 Nat’l Tax J. 519, 523 (2009) (“An efficient energy policy should not favor one energy source over another after taking into account any positive or negative externalities associated with its production or consumption.”).

[3].     See New Source Performance Standards for Greenhouse Gas Emissions from New, Modified, and Reconstructed Fossil Fuel-Fired Electric Generating Units, 89 Fed. Reg. 39798 (May 9, 2024) (to be codified at 40 C.F.R. pt. 60).

[4].     In practice, a carbon tax might only be imposed on upstream fossil fuel producers and refiners and still capture the 80 percent of U.S. emissions corresponding to fossil fuel combustion. Gilbert E. Metcalf & David Weisbach, The Design of a Carbon Tax, 33 Harv. Env’t L. Rev. 499, 501 (2009).

[5].     See Colton Poore, Without Guidance, Inflation Reduction Act Tax Credit May Do More Harm Than Good, Princeton Eng’g (Dec. 20, 2022), https://engineering.princeton.edu/news/2022/12/20/without-guidance-inflation-reduction-act-tax-credit-may-do-more-harm-good [https://perma.cc/X8CM-NPDM] (“[W]ithout careful implementation, the credit could backfire by inadvertently increasing nationwide carbon pollution.”).

[6].     See Arnaud Brohé, The Handbook of Carbon Accounting 25–26 (2016).

[7].     A number of articles discuss GHG emissions accounting in relation to specific categories of emissions. See generally Alexandra B. Klass & Andrew Heiring, Life Cycle Analysis and Transportation Energy, 82 Brook. L. Rev. 485 (2017) (discussing ethanol and transportation sector emissions accounting); Daniel A. Farber, Indirect Land Use Change, Uncertainty, and Biofuels Policy, 2011 U. Ill. L. Rev. 381 (discussing indirect land use effects of biofuels); Taotao Yue & Marjan Peeters, Better Greenhouse Gas Emissions Accounting for Biofuels: A Key to Biofuels Sustainability, 6 Climate L. 279 (2016) (discussing biofuels emissions accounting generally); Jonathan Rosenbloom, Outsourced Emissions: Why Local Governments Should Track and Measure Consumption-Based Greenhouse Gases, 92 U. Colo. L. Rev. 451 (2021) (arguing for municipal GHG accounting); Clint Wallace & Shelley Welton, Taxing Luxury Emissions, 109 Corn. L. Rev. 1153 (2024) (arguing for GHG accounting of consumption-based emissions); Lynn M. LoPucki, Corporate Greenhouse Gas Disclosures, 56 U.C. Davis L. Rev. 405 (2022) (cataloguing emissions accounting protocols to which corporations are subject); Madison Condon, What’s Scope 3 Good For?, 56 U.C. Davis L. Rev. 1921 (2023) (discussing the significance of “Scope 3” or supply chain emissions to the SEC’s climate disclosure rule).

[8].     See Shelley Welton, Neutralizing the Atmosphere, 132 Yale L.J. 171, 195–98, 199, 202–07 (2022) (discussing accounting risks that might undermine claims of net-zero). See generally Albert C. Lin, Fixing Net Zero Leakage, 58 Wake Forest L. Rev. 119 (2023) (proposing strategies to combat emissions leakage undermining corporate net-zero pledges, including greater accuracy in carbon accounting); Daniel C. Esty & Nathan de Arriba-Sellier, Zeroing in on Net-Zero: From Soft Law to Hard Law in Corporate Climate Change Pledges, 94 U. Colo. L. Rev. 635 (2023) (proposing principles for evaluating and disciplining corporate net-zero pledges).

[9].     Leehi Yona, Emissions Omissions: Greenhouse Gas Accounting Gaps,49Harv. Env’t L. Rev. 597, 638–39 (2025).

[10].     See, e.g., id. at 647–54 (describing implications for environmental law). But see Condon, supra note 7, at 1928–39 (showing how emissions accounting has become a critical element of securities law).

[11].     David A. Weisbach & Jacob Nussim, The Integration of Tax and Spending Programs, 113 Yale L.J. 955, 996 (2004).

[12].     But see Anne L. Alstott, The Earned Income Tax Credit and the Limitations of Tax-Based Welfare Reform,108 Harv. L. Rev. 533, 535–36 (1995).

[13].     Tax law is a social field in the same way that law more broadly is a social field. See Pierre Bourdieu, The Force of Law: Toward a Sociology of the Juridical Field, 38 Hastings L.J. 814, 828–31 (1987) (describing the socio-professional world of lawyers, or the “juridical space,” as a self-contained site of contestation within distinct internal norms and modes of evaluation).

[14].     See Cary Coglianese, The Limits of Performance-Based Regulation, 50 U. Mich. J.L. Reform 525, 558–61 (2017).

[15].     See infra Part II.C.

[16].     See infra Part III.B.

[17].     See infra Part III.B.

[18].     See infra Part III.D.

[19].     I.R.C. § 469. See generally Joseph Bankman, The Case Against Passive Investments: A Critical Appraisal of the Passive Loss Restrictions, 42 Stan. L. Rev. 15 (1989).

[20].     See infra Part III.C.

[21].     See, e.g., Economists’ Statement on Carbon Dividends, Climate Leadership Council, https://clcouncil.org/economists-statement/ [https://perma.cc/9PE5-JD6S] (reflecting thousands of economists endorsing a carbon tax); see also Gilbert E. Metcalf & David Weisbach, The Design of a Carbon Tax, 33 Harv. Env’t L. Rev. 499, 500 (2009) (acknowledging the scholarly consensus behind carbon taxation and proceeding to technical design issues).

[22].     See generally H. Spencer Banzhaf, A History of Pricing Pollution (Or, Why Pigouvian Taxes Are Not Necessarily Pigouvian) (Nat’l Bureau of Econ. Rsch., Working Paper No. 27683, 2020).

[23].     See generally A.C. Pigou, The Economics of Welfare (1920);Francis M. Bator, The Simple Analytics of Welfare Maximization, 47 Am. Econ. Rev. 22 (1957) (proposing roughly this definition of an externality).

[24].     For a classical exposition of Pigouvian taxation and subsidy, see generally William J. Baumol, On Taxation and the Control of Externalities, 62 Am. Econ. Rev. 307 (1972); Louis Kaplow & Steven Shavell, On the Superiority of Corrective Taxes to Quantity Regulation, 4 Am. L. Econ. Rev. 1, 2 (2002); Kyle D. Logue & Joel Slemrod, Of Coase, Calabresi, and Optimal Tax Liability, 63 Tax L. Rev. 797, 829 (2010) (explaining Pigouvian subsidies); Jonathan Masur & Eric Posner, Toward a Pigouvian State, 164 U. Pa. L. Rev. 93, 96, 104–08 (2015) (“Pigouvian taxes are constantly advocated by economists who seek to influence public policy.”). Some authors spell the word “Pigovian.” The modern theory of pollution charges did not emerge until the 1960s, when economist Allen Kneese studied water pollution and proposed “a system of charges and bounties” in order to set the marginal costs of abatement equal to the marginal damages of foregoing abatement. Allen V. Kneese & Kenneth C. Nobe, The Role of Economic Evaluation in Planning for Water Resource Development, 2 Nat. Res. J. 445, 459 n.16 (1962).

[25].     See William D. Nordhaus, Economic Growth and Climate: The Carbon Dioxide Problem, 67 Am. Econ. Rev. 341, 341 (1977).

[26].     Id. at344.

[27].     Economists’ Statement on Carbon Dividends, supra note 21.

[28].     Jennifer A. Dlouhy, From Greenspan to Yellen, Economic Brain Trust Backs Carbon Tax, Bloomberg (Jan. 16, 2019), https://www.bloomberg.com/news/articles/2019-01-17/from-greenspan-to-yellen-economic-brain-trust-backs-carbon-tax [https://perma.cc/29DA-8MMG]. Nordhaus, who had since won the Nobel Prize for Economics for his pioneering work in climate economics, did not sign the statement. See Economists’ Statement on Carbon Dividends, supra note 21.

[29].     Satyajit Bose, Allison L. Bridges & Kelsie DeFrancia, Columbia Univ., Earth Inst., Carbon Pricing as a Policy Instrument to Decarbonize Economies 14 (2019).

[30].     Id. at 14–15.

[31].     See generally Danny Cullenward & David G. Victor, Making Climate Policy Work (2020) (detailing the successes and shortcomings of cap-and-trade markets).

[32].     See Leah C. Stokes & Matto Mildenberger, The Trouble with Carbon Pricing, Bos. Rev. (Sep. 24, 2020), https://www.bostonreview.net/articles/leah-c-stokes-matto-mildenberger-tk/ [https://perma.cc/97TQ-BMVW] (discussing the political weakness of carbon taxation).

[33].     See Eric Biber, Nina Kelsey & Jonas Meckling, The Political Economy of Decarbonization: A Research Agenda, 82 Brook. L. Rev. 605, 616 (2017) (“Our interpretation of these results is that first, carbon pricing instruments tend to be politically difficult to enact as first steps in jurisdictions dependent on fossil fuel power.”).

[34].     Jonas Meckling, A New Path for U.S. Climate Politics: Choosing Policies That Mobilize Business for Decarbonization, 685 Annals Am. Acad. Pol. & Soc. Sci. 82, 86 (2019).

[35].     See John Kemp, Who Will Pay for the Energy Transition, Reuters (Apr. 27, 2021) (emphasis added), https://www.reuters.com/business/energy/who-will-pay-energy-transition-kemp-2021-04-27/ [https://perma.cc/H876-QGJL] (on the high upfront costs of the energy transition).

[36].     On the problem of fossil fuel assets being locked in due to their long operating life, see Emily Hammond & Jim Rossi, Stranded Costs and Grid Decarbonization, 82 Brook. L. Rev. 645, 652–59 (2017).

[37].     See, e.g., Alan Jenn, Inês L. Azevedo & Pedro Ferreira, The Impact of Federal Incentives on the Adoption of Hybrid Electric Vehicles in the United States, 40 Energy Econ. 936, 941 (2013) (finding that “only sufficiently large incentive amounts” yielded an effect on electric vehicle sales). See generally Eric Lonergan & Corinne Sawers, Supercharge Me: Net Zero Faster 20–22 (2022) (advocating the use of “extreme positive incentives for change”).

[38].     See J.W. Mason, Climate Policy from a Keynesian Point of View, Heinrich Böll Stiftung (Apr. 21, 2022), https://eu.boell.org/en/climate-policy-keynesian [https://perma.cc/KJA2-5SRK] (discussing the choice between carbon pricing and investment-based approaches to decarbonization); Noah M. Sachs, Can We Regulate Our Way to Energy Efficiency? Product Standards as Climate Policy, 65 Vand. L. Rev. 1631, 1659 (2012) (arguing that demand “pull” for rapid technological transformation “is unlikely to be generated by economy-wide cap-and-trade systems or by carbon taxes, unless the carbon price is substantial”); Noah Kaufman, Should the Inflation Reduction Act End the Dream of a National Carbon Price?, Columbia | SIPA: Ctr. on Glob. Energy Pol’y (Jan. 26, 2023), https://www.energypolicy.columbia.edu/should-the-inflation-reduction-act-end-the-dream-of-a-national-carbon-price/ [https://perma.cc/FG72-XWNH]. Other scholars do continue to propose variants of carbon taxes, such as a tax on emissions associated with luxury consumption (e.g., private jets and second homes). Seegenerally Wallace & Welton, supra note 7.

[39].     Lily L. Batchelder, Fred T. Goldberg, Jr. & Peter R. Orszag, Efficiency and Tax Incentives: The Case for Refundable Tax Credits, 59 Stan. L. Rev. 23, 45 (2006) (showing that Pigouvian modification of prices results in the optimal supply of a good). A rare critique of the Pigouvian view comes from Victor Fleischer, who argues that Pigouvian taxes and subsidies fail to target costs and benefits toward those who deserve or need them most. Victor Fleischer, Curb Your Enthusiasm for Pigovian Taxes, 68 Vand. L. Rev. 1673, 1688, 1697 (2015).

[40].     Revenue Act of 1962, Pub. L. No. 87-834, § 2, 76 Stat. 960, 962–73 (1962).

[41].     Thomas W. Giegerich, The Monetization of Business Tax Credits, 12 Fla. Tax Rev. 709, 720–22 (2012).

[42].     Energy Tax Act of 1978, Pub. L. No. 95-618, § 301, 92 Stat. 3174, 3194–201 (1978). In 1986, Congress repealed the generic investment credit but retained the energy version. See S. Rep. No. 99-313, at 275–76 (1986) (noting that even though the oil crisis had abated and fossil fuel prices had steeply declined, giving up on renewable energy at this moment would waste innovation progress from the past decade).

[43].     I.R.C. § 48(l)(2) (1978) (amended 2025). “Alternative energy property” referred to any equipment operated with fuel sources other than oil and natural gas. Id. § 48(l)(3) (1978); see also Energy Tax Act of 1978, Pub. L. No. 95-618, § 301(b), 92 Stat. 3174, 3194–96.“Specially defined energy property” was the most narrowly targeted category, listing eleven specific types of products that could be used to reduce the amount of energy consumed in an industrial or commercial facility, including recuperators, heat pipes, and turbulators. I.R.C. § 48(l)(5) (1978); see also Energy Tax Act of 1978 § 301(b).

[44].     Energy Policy Act of 1992, Pub. L. No. 102-486, § 1914(a), 106 Stat. 2776, 3020–23.

[45].     Id. § 45(a) (current version at I.R.C. § 45(a)).

[46].     Id. § 45(c)(1) (current version at I.R.C. § 45(c)(1)).

[47].     See American Jobs Creation Act of 2004, Pub. L. No. 108-357, § 710(a), 118 Stat. 1418, 1552.

[48].     See generally Austin Brown, Philipp Beiter, Donna Heimiller, Carolyn Davidson, Paul Denholm, Jennifer Melius, Anthony Lopez, Dylan Hettinger, David Mulcahy & Gian Porro, Nat’l Renewable Energy Lab’y, Estimating Renewable Energy Economic Potential in the United States: Methodology and Initial Results 59, 60–61 (2016) (discussing the parameters affecting the economic potential of solar and wind energy, respectively).

[49].     Tax Credits for Electricity Production from Renewable Sources: Hearing Before the Subcomm. on Select Revenue Measures of the H. Comm. on Ways & Means, 109th Cong. 47–48 (2005) (statement of Christopher O’Brien, Vice President, Strategy & Gov’t Rels., Solar Sys. Div., Sharp Elecs., Chairman of the Bd. of Dirs., Solar Energy Indus. Ass’n).

 [50].     The chairman of the solar industry lobbying group testified to Congress in 2005 that “we would willingly be removed from the section 45 credit [PTC] in favor of an expansion of the existing investment credit.” Id. at 48.

[51].     David M. Schizer, Energy Subsidies: Worthy Goals, Competing Priorities, and Flawed Institutional Design, 70 Tax L. Rev. 243, 298 (2017).

[52].     Id. at 280; Jonathan H. Adler, Eyes on a Climate Prize: Rewarding Energy Innovation to Achieve Climate Stabilization, 35 Harv. Env’t. L. Rev. 1, 29–30 (2011) (arguing that the government is ill-equipped to identify and efficiently subsidize promising new technologies); Adam B. Jaffe, Richard G. Newell & Robert N. Stavins, A Tale of Two Market Failures: Technology and Environmental Policy, 54 Ecological Econ. 164, 171 (2005) (advocating that technology policy should be “‘technology neutral,’ encouraging all efforts that achieve specified objectives without focusing on a particular approach”).

[53].     Zachary Liscow & Quentin Karpilow, Innovation Snowballing and Climate Law, 95 Wash. U. L. Rev. 387, 391 (2017).

[54].     David Weisbach, Designing Subsidies for Low-Carbon Energy, 20 J. Env’t. & Sustainability L. 1, 6 (2013).

[55].     Technology Neutrality in Energy Tax: Issues and Options: Hearing Before the S. Comm. on Fin., 111th Cong. 60 (2009) (Statement of Gilbert E. Metcalf, Professor of Econ., Tufts Univ.).

[56].     Weisbach, supra note 54, at 14.

[57].     One way to deal with this problem is to set the subsidy value for green energy technologies equivalent to the implied carbon tax necessary for the dirtiest possible alternative, usually coal. The effect of this approach would be to “over-subsidize” green technologies that compete with natural gas.

[58].     Liscow & Karpilow, supra note 53, at 392; see also Daron Acemoglu, Philippe Aghion, Leonardo Bursztyn & David Hemous, The Environment and Directed Technical Change, 102 Am. Econ. Rev. 131, 132–33 (2012) (arguing that optimal environmental regulation should pair a carbon tax with direct encouragement for the development of specific clean technologies).

[59].     See, e.g., Gregory F. Nemet, How Solar Energy Became Cheap 175–79 (2d ed. 2026) (documenting innovation snowballing in the solar industry).

[60].     Regional Clean Hydrogen Hubs,U.S. Dep’t of Energy, https://www.energy.gov/oced/regional-clean-hydrogen-hubs-0 [https://perma.cc/9TJB-UTFF]; Regional Direct Air Capture Hubs, U.S. Dep’t of Energy, https://www.energy.gov/oced/DACHubs [https://perma.cc/9QZS-JR2V].

[61].     Technology Neutrality in Energy Tax: Issues and Options:Hearing Before the S. Comm. on Fin., supra note 55, at 1 (statement of Sen. Max Baucus, Chairman, S. Comm. on Fin.).

 [62].     S. Comm. on Fin, 113th Cong., Staff Discussion Draft: Energy Tax Reform (2013); see also Staff of J. Comm. on Tax’n, 113th Cong., Technical Explanation of the Senate Committee on Finance’s Staff Discussion Draft to Reform Certain Energy Tax Provisions (2013) (providing explanations of the proposed technology-neutral credits).

[63].     See infra Part II.C.

[64].     Cong. Budget Off., Estimated Budgetary Effects of H.R. 5376, the Inflation Reduction Act of 2022 (2022); Credit Suisse, Treeprint: US Inflation Reduction Act – A Tipping Point in Climate Action 4 (2022).

[65].     See generally Jeff Gordon, Statutory Contracts, 42 Yale J. on Regul. 660 (2025) (explaining the open-ended obligation of these spending programs).

[66].     I.R.C. §§ 40B, 45Q, 45V, 45Y, 45Z, 48E.

[67].     Kyoto Protocol to the United Nations Framework Convention on Climate Change, Dec. 10, 1997, 2303 U.N.T.S. 162; 1 Nat’l Greenhouse Gas Inventories Programme, Intergovernmental Panel on Climate Change, 2006 IPCC Guidelines For National Greenhouse Gas Inventories: General Guidance and Reporting 1.4 (Simon Eggleston, Leandro Buendia, Kyoko Miwa, Todd Ngara & Kiyoto Tanabe eds., 2006).

[68].     United Nations Framework Convention on Climate Change, Draft Handbook For The Review Of National Greenhouse Gas Inventories 54 (2019) (providing for a national inventory report); see also Yona, supra note 9, at 619–20 (on the distinction between measuring, reporting, and accounting for emissions).

[69].     See generally World Bus. Council for Sustainable Dev. & World Res. Inst., The Greenhouse Gas Protocol: A Corporate Accounting And Reporting Standard (rev. ed. 2015).

[70].     Id. at 25; Yona, supra note 9, at 629.

[71].     Energy Independence and Security Act of 2007, Pub. L. No. 110-140, sec. 201, 121 Stat. 1492, 1519–21; Mandatory Reporting of Greenhouse Gases, 74 Fed. Reg. 56260 (Oct. 30, 2009) (to be codified at 40 C.F.R. pts. 86, 87, 89, 90, 94, 98, 1033, 1039, 1042, 1045, 1048, 1051, 1054, 1065).

[72].     U.S. EPA, Fact Sheet: Greenhouse Gas Reporting Program Implementation (2013), https://www.epa.gov/sites/default/files/2014-09/documents/ghgfactsheet.pdf [https://perma.cc/JP35-QREP].

[73].     Renewable Fuel Petition Review Process, U.S. EPA (Apr. 10, 2025), https://www.epa.gov/renewable-fuel-standard-program/renewable-fuel-petition-review-process [https://perma.cc/F4WD-MPPR].

[74].     James S.N. McCubbins & A. Bryan Endres, EPA Biofuel Pathways and Petitions: Failure to Launch?, 3 farmdoc daily 100, 100 (2013).

[75].     See, e.g., Renewable Fuel Standard Program: Grain Sorghum Oil Pathway, 83 Fed. Reg. 37735, 37736 (Aug. 2, 2018) (to be codified at 40 C.F.R. pt. 80) (adding indirect land use effects to the evaluation of a fuel pathway). This methodological issue is discussed further below. See infra Part II.A.2.

[76].     See The Enhancement and Standardization of Climate-Related Disclosures for Investors, 89 Fed. Reg. 21668, 21916 (Mar. 28, 2024) (to be codified at 17 C.F.R. pts. 210, 229, 230, 232, 239, and 249).

[77].     The accounting can become more complicated if one considers the possibility of carbon leaking out of storage.

[78].     Bipartisan Budget Act of 2018 § 41119, I.R.C. § 45Q(a)(2).

[79].     See Renée Cho, Capturing Carbon’s Potential: These Companies Are Turning CO2 into Profits, Columbia Climate Sch.: State of the Planet (May 29, 2019), https://news.climate.columbia.edu/2019/05/29/co2-utilization-profits/ [https://perma.cc/ESR2-WQEM] (illustrating the range of industrial applications for carbon oxides).

[80].     I.R.C. § 45Q(f)(5)(B)(i).

[81].     Id.

 [82].     Inflation Reduction Act of 2022, Pub. L. No. 117-169, 136 Stat. 1818.

[83].     I.R.C. § 40B(b), 40B(e).

[84].     Id. § 45Z(b).

[85].     Id. § 45V(b)(2).

[86].     Id. §§ 45Y(b)(1)(A), 48E(b)(3)(A). Furthermore, the rules under 45Y create an interaction with the carbon sequestration credit, in that the emissions rate for an electricity facility must be lowered to the extent carbon dioxide is captured and stored or utilized pursuant to 45Q. See 26 C.F.R § 1.45Y-5(e) (2025).

[87].     One Big Beautiful Bill Act, Pub. L. No. 119-21, § 70512, 139 Stat. 72, 252 (2025). There is an exception for projects that begin construction within one year of the passage of OBBBA, or up through July 3, 2026.

[88].     See Treas. Reg. § 1.45Y-5(c)(2) (2025).

[89].     See infra text accompanying notes 116–120 (on indirect land use changes), text accompanying note 203 (on fuel cells), and note 251 (on carbon utilization). But seeinfra text accompanying note 243 (on negative emissions treatment for fuels).

[90].     See 42 U.S.C. § 7545(o)(1)(H).

[91].     I.R.C. § 61(a).

[92].     Id. § 4121(a)(1).

[93].     See generally Boris I. Bittker, A “Comprehensive Tax Base” as a Goal of Income Tax Reform, 80 Harv. L. Rev. 925 (1967); Henry Aaron, What Is a Comprehensive Tax Base Anyway?, 22 Nat’l Tax J. 543 (1969).

[94].     See Weisbach & Nussim, supra note 11, at 967 (explaining that the comprehensive base concept is grounded in both efficiency and fairness).

 [95].     I.R.C. §§ 40B(e)(2), 45Q(f)(5)(B)(ii), 45V(c)(1)(A), 45Y(b)(2)(B).

[96].     See Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel Standard Program, 75 Fed. Reg. 14670, 14765 (Mar. 26, 2010) (to be codified at 40 C.F.R. pt. 80).

[97].     42 U.S.C. § 7545(o)(1)(H).

[98].     The Clean Air Act language refers to a “fuel lifecycle” because it pertains to the Renewable Fuel Standard program. When Congress incorporates that language for the purpose of tax credits that do not exclusively pertain to fuels, one should understand “fuel lifecycle” as substitutable for whatever product lifecycle is under evaluation.

[99].     For definitions of attributional and consequential carbon accounting, see Göran Finnveden, Michael Z. Hauschild, Tomas Ekvall, Jeroen Guinée, Reinout Heijungs, Stefanie Hellweg, Annette Koehler, David Pennington & Sangwon Suh, Recent Developments in Life Cycle Assessment, 91 J. Env’t Mgmt. 1, 3 (2009); Tomas Ekvall, Attributional and Consequential Life Cycle Assessment, in Sustainability Assessment at the 21st Century 41, 44 (María José Bastante-Ceca, Jose Luis Fuentes-Bargues, Levente Hufnagel, Florin-Constantin Mihai & Corneliu Iatu eds., 2020).

[100].     42 U.S.C. § 7545(o)(1)(H).

[101].     See Credit for Carbon Oxide Sequestration, 86 Fed. Reg. 4728, 4745 (Jan. 15, 2021) (to be codified at 26 C.F.R. pt. 1) (endorsing the use of a “cradle-to-grave boundary” for the lifecycle analysis under Section 45Q).

[102].     Yona, supra note 9, at 608–09.

[103].     Seeid. at 608; Lambert Schneider, Maosheng Duan, Robert Stavins, Kelley Kizzier, Derik

Broekhoff, Frank Jotzo, Harald Winkler, Michael Lazarus, Andrew Howard & Christina Hood, Double Counting and the Paris Agreement Rulebook, 366 Science 180, 181 (2019).

[104].     See Lambert Schneider & Anja Kollmuss, Perverse Effects of Carbon Markets On HFC-23 And SF6 Abatement Projects in Russia, 5 Nature Climate Change 1061, 1061 (2015) (describing the perverse incentives for waste generation in carbon offset markets).

[105].     I.R.C. § 1(g) (known as the “kiddie tax”).

[106].     Id. § 482.

[107].     Id. § 704(c)(1)(B).

[108].     This approach could draw lessons from a United Nations program meant to incentivize the destruction of HFC-23, a coolant manufacturing byproduct with deleterious climate impacts. In response to the Clean Development Mechanism’s incentives for destroying HFC-23, producers began to produce more of the gas in order to destroy it. The UN’s ultimate response was to refuse to award carbon credits to any new factories built after the announcement of the incentive. Elisabeth Rosenthal & Andrew W. Lehren, Profits on Carbon Credits Drive Output of a Harmful Gas, N.Y. Times (Aug. 8, 2012), https://www.nytimes.com/2012/08/09/world/asia/incentive-to-slow-climate-change-drives-output-of-harmful-gases.html [https://perma.cc/7GKU-MM9M] (“The United Nations, recognizing the temptation for companies to jump into the lucrative business, has refused since 2007 to award carbon credits to any new factories destroying the waste gas.”).

[109].     See U.S. Dep’t of Energy, Guidelines to Determine Well-to-Gate Greenhouse Gas (GHG) Emissions of Hydrogen Production Pathways Using 45VH2-GREET 11–12 (rev. 2024) [hereinafter GHG Guidelines] (restricting the amount of steam coproduct that steam methane reformers can claim based on the quantity of steam that an optimally designed reformer is expected to be capable of producing, in order to avoid incentivizing overproduction of hydrogen co-products).

[110].     Fran Moore, How Much Hot Air?, Briefing Book (Feb. 19, 2024), https://www.briefingbook.info/p/how-much-hot-air [https://perma.cc/Q3TS-PLJX].

[111].     Id.Cf. Welton, supra note 8, at 231 (making a similar point about the unpredictable equilibrium effect of encouraging mass adoption of carbon offset strategies, even if well-understood when undertaken at small scale).

[112].     Pol’y Dep’t for External Rels., Eur. Parliament, Four Briefings on Trade-Related Aspects of Carbon Border Adjustment Mechanisms, at 11 (Apr. 2020) (“[W]hile [carbon border adjustment] almost totally cancels sectoral leakages, its impact on overall leakage is limited, reducing it from 10.8 % to 8.2 % in the best case. Half of the overall leakage (5.9 p.p.) comes from the indirect channel: the emissions increase in the sectors that generate electricity, where emission intensity increases because of reduced fuel and other primary energy prices.”).

[113].     42 U.S.C. § 7545(o)(1)(H). See supra text accompanying notes 78–86 (detailing the tax credit statutes that reference the Clean Air Act definition of lifecycle emissions).

[114].42 U.S.C. § 7545(o)(1)(H).

[115].     Id.

[116].     Seegenerally Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel Standard Program, 75 Fed. Reg. 14670 (Mar. 26, 2010) (to be codified at 40 C.F.R. pt. 80) (also known as the “RFS2 rule”).

[117].     See Stephanie Searle, If We Use Livestock Feed for Biofuels, What Will the Cows Eat?, Int’l Council on Clean Transp. (Jan. 10, 2019), https://theicct.org/if-we-use-livestock-feed-for-biofuels-what-will-the-cows-eat/ [https://perma.cc/5HR8-7EV9] (discussing how if corn oil is used for biofuel, farmers then need to replace it with something else in livestock feed).

[118].     Tyler J. Lark, Nathan P. Hendricks, Aaron Smith, Nicholas Pates, Seth A. Spawn-Lee, Matthew Bougie, Eric G. Booth, Christopher J. Kucharik & Holly K. Gibbs, Environmental Outcomes of the US Renewable Fuel Standard, 119 Procs. Nat’l Acad. Scis. U.S. Am. 1, 2 (2022).

[119].     Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel Standard Program, 75 Fed. Reg. at 14766 (“The definition of lifecycle GHG emissions makes it clear that EPA is to determine the aggregate emissions related to the ‘full’ fuel lifecycle, including ‘all stages of fuel and feedstock production and distribution.’ Thus, EPA could not, as a legal matter, ignore those parts of a fuel lifecycle that occur overseas.”).

[120].     One Big Beautiful Bill Act, Pub. L. No. 119-21, § 70521, 139 Stat. 72, 276–79 (2025).

[121].     See Deepak Rajagopal & David Zilberman, On Market-Mediated Emissions and Regulations on Life Cycle Emissions, 90 Ecological Econ. 77, 77 (2013) (introducing the concept of market-mediated emissions).

[122].     For example, subsidy applicants in the biogas sector should not be able to claim credit for avoiding a counterfactual of venting methane into the atmosphere to the extent the Clean Air Act would have required them to collect or flare such gases. See Inst. for Pol’y Integrity, Comment Letter on Proposed Rule for Section 45Y Clean Electricity Production Credit and Section 48E Clean Electricity Investment Credit (Aug. 2, 2024), https://policyintegrity.org/documents/Policy_Integrity_Section_45Y_48E_Comments.pdf [https://perma.cc/MR9U-2KZ9]. On the concept of regulatory additionality in general, see Brian Joseph McFarland, Carbon Reduction Projects and the Concept of Additionality, 11 Sustainable Dev. L. & Pol’y 15 (2011).

[123].     See, e.g., Laurence H. Meyer, Joel L. Prakken & Chris P. Varvares, Designing an Effective Investment Tax Credit, 7 J. Econ. Persps. 189, 195 (1993); see Jane G. Gravelle, What Can Private Investment Incentives Accomplish? The Case of the Investment Tax Credit, 46 Nat’l Tax J. 275, 282–84 (1993).

[124].     See James Salzman & David Weisbach, The Additionality Double Standard, 48 Harv. Env’t L. Rev. 117, 129–33 (2024) (documenting the ubiquity of the additionality problem within tax subsidies).

[125].     A. Will Frazier, Cara Marcy & Wesley Cole, Wind and Solar PV Deployment after Tax Credits Expire: A View from the Standard Scenarios and the Annual Energy Outlook, 32 Elec. J. 1, 7 (2019).

[126].     This problem is referred to as inframarginality or “buying the base.” See Daniel E. Herz-Roiphe, Virtue’s Reward: A (Qualified) Defense of Infra-Marginal Tax Subsidies, 36 Va. Tax Rev. 1, 2, 5 n.11 (2017).

[127].     Economic Recovery Tax Act of 1981, Pub. L. No. 97-34, § 221, 95 Stat. 172, 241–47.

[128].     I.R.C. § 41(c)(3) (setting the base amount against which increases in research are measured in relation to historical spending on research).

[129].     Laura Tyson & Greg Linden, Ctr. for Am. Progress, The Corporate R&D Tax Credit and U.S. Innovation and Competitiveness 42–43 (2012); Cong. Budget Off., Federal Support for Research and Development 24 (2007).

[130].     Treas. Reg. § 1.45Y-5(e) (2025).

[131].     Credit for Production of Clean Hydrogen, 90 Fed. Reg. 2224, 2254 (Jan. 10, 2025) (to be codified at 26 C.F.R. pt. 1)

[132].     See infra Part III.B.4.

[133].     Salzman & Weisbach, supra note 124, at 119–20.

[134].     Id. at 133–38.

[135].     Examples of “perverse incentives,” or incentives that encourage behavior undermining their intended purpose, abound. In anticipation of the implementation of the Endangered Species Act, landowners preemptively altered land to make it uninhabitable for protected species so they would not be covered by the law. John A. List, Michael Margolis & Daniel E. Osgood, Is the Endangered Species Act Endangering Species? 27(Nat’l Bureau of Econ. Rsch., Working Paper No. 12777, 2006). The CAFE standards increased fuel efficiency, which made driving less costly per mile, but incentivized more driving, which increased emissions relative to the counterfactual. Todd Litman, Efficient Vehicles Versus Efficient Transportation Comparing Transportation Energy Conservation Strategies, 12 Transp. Pol’y 121, 128 (2005).

[136].     Ellen P. Aprill & Daniel J. Hemel, The Tax Legislative Process: A Byrd’s Eye View, 81 Law & Contemp. Probs 99, 103 (2018); see Rebecca M. Kysar, Tax Law and the Eroding Budget Process, 81 Law & Contemp. Probs. 61, 68 (2018) (describing the expansion of the reconciliation process).

[137].     On the appeal of tax-driven regulation in light of judicial hostility to other modes of regulation, seeGillian E. Metzger, To Tax, To Spend, To Regulate, 126 Harv. L. Rev. 83, 109 (2012) (predicting that tax law would emerge as a favored mode of economic regulation after National Federation of Independent Business v. Sebelius, 567 U.S. 519, 587 (2012)); Linda Sugin, The Great and Mighty Tax Law: How the Roberts Court Has Reduced Constitutional Scrutiny of Taxes and Tax Expenditures, 78 Brook. L. Rev. 777, 779 (2013) (noting a potential for tax law to “swallow all government policy” following Sebelius).

[138].     See Mackenzie Hawkins, Commerce Department Is Reaping Hundreds of Chips Act Inquiries, Bloomberg News (Aug. 9, 2023), https://news.bloomberglaw.com/ip-law/commerce-department-is-reaping-hundreds-of-chips-act-inquiries [https://perma.cc/6HBG-USZL].

[139].     See Jody Freeman & Jim Rossi, Agency Coordination in Shared Regulatory Space, 125 Harv. L. Rev. 1131, 1142 (2012) (introducing the idea of interagency coordination); Jason Marisam, Interagency Administration,45 Ariz. St. L.J. 183, 190(2013) (arguing that a rational agency enlists an outside agency to contribute expertise it lacks); Bijal Shah, Uncovering Coordinated Interagency Adjudication, 128 Harv. L. Rev. 805, 822 (2015) (on the need for varied expertise in adjudication as reason to bring in multiple agencies); Blaine G. Saito, Tax Coordination, 38 Ga. St. U. L. Rev. 735, 752–57 (2021) (advocating for interagency coordination in the tax law context).

[140].     See I.R.C. § 50(a)(7)(D)(ii).

[141].     See id. § 6103(l)(21).

[142].     See, e.g., id. § 45(b)(7)(A)(ii). In cases where the Department of Labor has not published a prevailing wage determination for a particular occupation, Treasury guidance directs taxpayers to use procedures established by the Secretary of Labor to request a wage determination. This means that the IRS will play no direct role in making such determinations; instead, it will rely on determinations made by Labor. IRS, Notice 2022-61, Prevailing Wage and Apprenticeship Initial Guidance Under § 45(b)(6)(B)(ii) and Other Substantially Similar Provisions (2022).

[143].     To be sure, there remain some practical differences between how the IRS and other peer agencies administer their respective spending programs. For example, Eric Toder notes that the costs of administering direct expenditure programs are generally more transparent than the costs of administering tax expenditures, because the IRS budget does not include the incremental costs of enforcing individual tax provisions, while agency budgets typically break out costs by program. Eric Toder, Tax Cuts or Spending - Does it Make a Difference?, 53 Nat’l Tax J. 361, 365 (2000).

 [144].     IRS, Notice 2024-37, Sustainable Aviation Fuel Credit; Lifecycle Greenhouse Gas Emissions Reduction Percentage and Certification of Requirements Related to the Clean Air Act; Climate Smart Agriculture; Safe Harbors (2024).

[145].     SeeClimate-Smart Agriculture and Forestry, U.S. Dep’t of Agric., https://www.farmers.gov/conservation/climate-smart [https://perma.cc/V959-PBK4].

[146].     Mandatory Reporting of Greenhouse Gases: Injection and Geologic Sequestration of Carbon Dioxide, 75 Fed. Reg. 75060, 75061 (Dec. 1, 2010) (to be codified at 40 C.F.R. pts. 72, 78, 98).

[147].     Treas. Reg. § 1.45Q-3(b)(2)(ii) (2021).

[148].     Letter from J. Russell George, Inspector Gen., Tax Admin., to the Sen. Robert Menendez 2 (Apr. 15, 2020), https://legacy-assets.eenews.net/open_files/assets/2020/04/30/document_gw_07.pdf [https://perma.cc/MUT2-39ZG].

[149].     I.R.C. § 40B(e)(2).

[150].     See IRS, supra note 144, § 3.03.

[151].     I.R.C. § 45V(c)(1)(B).

[152].     See generally GHG Guidelines, supra note 109.

[153].     Treas. Reg. § 1.45V-4(a)–(c) (2025).

[154].     Seeid. § 1.45Y-5(h)(6) (2026).

[155].     Id. §§ 1.45V-4(c)(3), 1.45Y-5(h)(3).

[156].     Id. § 1.45Q-4(c)(6) (2026).

[157].     See id. (making clear that the IRS determines whether to approve an LCA after DOE review). The fact that the IRS retains ultimate adjudicatory authority may be important for protecting against APA challenges. The D.C. Circuit has held that statutory silence does not constitute an implicit ban on interagency consultation, so long as the agency delegated by statute is the one making the final decision. See Nat’l Mining Ass’n v. McCarthy, 758 F.3d 243, 249 (D.C. Cir. 2014). See generally Bijal Shah, Interagency Transfers of Adjudication Authority, 34 Yale J. on Regul. 279 (2017) (arguing that interagency power transfers are constitutional and beneficial).

[158].     Looking to procedures under the Carbon Utilization Credit as a representative guide, the DOE will choose between two modes of review: a “Conformance Review” if the LCA report has been verified by an independent third party, or a “Critical Review” if it has not. A Conformance Review checks to see that the LCA conforms to applicable standards and to LCA guidance documents published by the National Energy Technology Laboratory, while a Critical Review involves “a detailed technical assessment of the LCA model and supporting data.” IRS, Notice 2024-60, Required Procedures to Claim a Section 45Q Credit for Utilization of Carbon Oxide 11–12 (2024).

[159].     See, e.g., Celanese Corp., Comment Letter on IRS Notice 2022-57 on the Credit for Carbon Oxide Sequestration (Dec. 3, 2022), https://www.bakerbotts.com/~/media/Files/Thought-Leadership/Publications/2022/December/CelaneseCorporation-58.pdf [https://perma.cc/5UR2-UPQZ].

[160].     Coglianese, supra note 14, at 532.

[161].     For an example of a design standard, consider the renewable energy investment tax credit, which names eleven acceptable forms of energy property, including solar energy equipment, geothermal equipment, fuel cell power plants, microturbine power plants, and so on. For each form of energy property, the statute provides further detailed design requirements. I.R.C. § 48(a)(3)(A).

[162].     Cary Coglianese & Jennifer Nash, The Law of the Test: Performance-Based Regulation and Diesel Emissions Control, 34 Yale J. on Regul. 33, 37 (2017).

[163].     Id.

[164].     Coglianese, supra note 14, at 545.

[165].     See Exec. Order No. 12,866, 3 C.F.R. 638 (1994) (urging regulatory agencies to “specify performance objectives, rather than specifying the behavior or manner of compliance that regulated entities must adopt”); Off. of Mgmt. & Budget, Circular A-4, at 8 (2003) (performance standards are “generally superior to engineering or design standards because performance standards give the regulated parties the flexibility to achieve regulatory objectives in the most cost-effective way.”); Exec. Order No. 13,563, 3 C.F.R. 13563 (2012) (directing administration to prioritize flexible approaches including “performance objectives”); see also Cass R. Sunstein, Simpler: the Future of Government 11–12 (2013) (“[F]lexible ‘performance standards’ . . . reduce[] costs [and] promote[] freedom.”).

[166].     Technology Neutrality in Energy Tax: Issues and Options:Hearing Before the S. Comm. on Fin., supra note 55, at 33 (statement of David L. Greene, Visiting Scholar, Univ. of California at Davis, Corp. Fellow, Oak Ridge Nat’l Lab’y).

[167].     See, e.g., Wendy E. Wagner, The Triumph of Technology-Based Standards, 2000 U. Ill. L. Rev. 83, 94–103 (arguing that design standards are easier to implement because they only require the government to identify a set of technological approaches that already work); Coglianese, supra note14, at 553–62.

[168].     Coglianese & Nash, supra note 162, at 45 (discussing the methods of assessment needed to enforce a performance standard).

[169].     See Emily Cauble, Safe Harbors in Tax Law, 47 Conn. L. Rev. 1385, 1398–400 (2015). However, certainty is not necessarily desirable. Tax scholars have observed that providing regulatory certainty helps taxpayers exploit poorly written rules. See Noël B. Cunningham & James R. Repetti, Textualism and Tax Shelters, 24 Va. Tax Rev. 1, 33 (2004); David A. Weisbach, Formalism in Tax Law, 66 U. Chi. L. Rev. 860, 867–72 (1999).

[170].     Coglianese & Nash, supra note 162, at 86.

[171].     Id. at 81.

[172].     Id. at 83.

[173].     See, e.g., GHG Guidelines, supra note 109.

[174].     See Cauble, supra note 169, at 1390, 1405; Susan C. Morse, Safe Harbors, Sure Shipwrecks, 49 U.C. Davis L. Rev. 1385, 1389 (2016).

[175].     Treas. Reg. § 1.7704-1(c) (1995).

[176].     I.R.C. § 7704(b); Treas. Reg. § 1.7704-1(h), (j).

[177].     I.R.C. § 7704(b).

[178].     I.R.C. § 121(c)(2); Treas. Reg. § 1.121-3(c)(2)(ii) (as amended in 2004).

[179].     Morse, supra note 174, at 1389.

[180].     Id. at 1397–402.

[181].     Id. at 1419–24 (discussing the risks of over- and under-inclusion in safe harbors).

[182].     See generally Albert O. Hirschman, The Rhetoric of Reaction: Perversity, Futility, Jeopardy (1991) (classifying three standard objections to reform proposals).

[183].     Joseph Bankman, The Tax Shelter Problem, 57 Nat’l Tax J. 925, 925–26(2004).

[184].     As of the enactment of the OBBBA, the Clean Hydrogen Credit is set to terminate for facilities beginning construction after December 31, 2027. See One Big Beautiful Bill Act, Pub. L. No. 119-21, § 70511, 139 Stat. 72, 252 (2025).

[185].     Dan Esposito, Eric Gimon & Mike O’Boyle, Energy Innovation Pol’y & Tech. LLC, Smart Design of 45V Hydrogen Production Tax Credit Will Reduce Emissions and Grow the Industry 12 (2023), https://energyinnovation.org/wp-content/uploads/2023/04/Smart-Design-Of-45V-Hydrogen-Production-Tax-Credit-Will-Reduce-Emissions-And-Grow-The-Industry.pdf [https://perma.cc/C68U-D3HS].

[186].     See generally Peter C. Canellos, A Tax Practitioner’s Perspective on Substance, Form and Business Purpose in Structuring Business Transactions and in Tax Shelters, 54 SMU L. Rev. 47 (2001).

[187].     See generally Joseph Bankman, The Business Purpose Doctrine and the Sociology of Tax, 54 SMU L. Rev. 149 (2001) (discussing the business purpose doctrine, a predecessor to the economic substance doctrine).

[188].     See Brian Galle, Interpretative Theory and Tax Shelter Regulation, 26 Va. Tax Rev. 357, 380–86 (2006).

[189].     See, e.g., Jonathan H. Choi, Beyond Purposivism in Tax Law, 107 Iowa L. Rev. 1439, 1442 (2022); Jonathan H. Choi, The Substantive Canons of Tax Law, 72 Stan. L. Rev. 195, 209–16 (2020) [hereinafter Choi, Substantive Canons of Tax Law].

[190].     I.R.C. § 7701(o)(1); see also Choi, Substantive Canons of Tax Law, supra note 189, at 225–27 (describing the codification of the economic substance doctrine).

[191].     See Jad Mouawad & Clifford Krauss, Lawmakers May Limit Paper Mills’ Windfall, N.Y. Times (Apr. 17, 2009), https://www.nytimes.com/2009/04/18/business/energy-environment/18sludge.html [https://perma.cc/A9T5-3D7R].

[192].     Id.; see I.R.S. Gen. Couns. Mem. 200941011 (Oct. 9, 2009).

[193].     See supra Part II.B (on the statutory language that supports a comprehensive emissions base).

[194].     See Marvin A. Chirelstein & Lawrence A. Zelenak, Tax Shelters and the Search for a Silver Bullet, 105 Colum. L. Rev. 1939, 1951–52 (2005) (discussing the appeal of “silver bullet” solutions that apply to a wide variety of shelters).

[195].     Tax Reform Act of 1986, Pub. L. No. 99-514, § 501, 100 Stat. 2233.

[196].     See James S. Eustice, Abusive Corporate Tax Shelters: Old “Brine” in New Bottles, 55 Tax L. Rev. 135, 147 n.71 (2002) (stating that Section 469 “deservedly received most of the credit for ending the prior shelter activity”); Chirelstein & Zelenak, supra note 194, at 1951 (“The pre-1986 shelters were all very much of a kind—virtually all involved the creation of artificial (noneconomic) losses for passive investors through the combination of tax preferences (most commonly accelerated depreciation) and interest expense deductions.”).

[197].Chirelstein & Zelenak, supra note 194, at 1952–53.

[198].     Treas. Reg. § 1.45Y-5(c)(2) (2025).

[199].     NYU Tax L. Ctr., Comment Letter on Guidance for Sections 48E and 45Y Technology-Neutral Credits 4–5 (Aug. 1, 2024), https://www.regulations.gov/comment/IRS-2024-0026-1571 [https://perma.cc/VA68-P4TT].

[200].     John A. Harrison, Yves T. Prairie, Sara Mercier-Blais & Cynthia Soued, Year-2020 Global Distribution and Pathways of Reservoir Methane and Carbon Dioxide Emissions According to the Greenhouse Gas from Reservoirs (G-res) Model, 35 Global Biogeochemical Cycles 1, 2–3 (2021). See generally Laura Scherer & Stephan Pfister, Hydropower’s Biogenic Carbon Footprint, 11 PLOS One 1 (2016).

[201].     Section 45Y Clean Electricity Production Credit and Section 48E Clean Electricity Investment Credit, 89 Fed. Reg. 47792, 47802 (proposed June 3, 2024) (to be codified at 26 C.F.R. pt. 1).

[202].     See NYU Tax L. Ctr., supra note 199, at 23–24; see also Consumers Power Co. v. Comm’r, 89 T.C. 710, 726 (1987) (holding that all parts of a hydroelectric plant “must be viewed as one integrated unit” for depreciation and ITC purposes because they “operate simultaneously and as a unit in order to produce electrical power”).

[203].     I.R.C. § 48E(j).

[204].     Types of Fuel Cells, U.S. Dep’t of Energy, https://www.energy.gov/eere/fuelcells/types-fuel-cells [https://perma.cc/P9NR-7K9X] (“Most fuel cells are powered by hydrogen.”); see infra Part III.4.

[205].     For example, the most recent Carbon Sequestration Credit regulations reduced the recapture period from five years to three years for carbon oxide leaking out of geologic storage. Credit for Carbon Oxide Sequestration, 86 Fed. Reg. 4728, 4750 (Jan. 15, 2021) (to be codified at 26 C.F.R. pt. 1) (“The final regulations revise the recapture period from five years to three years.”).

[206].     I.R.C. § 48E(b)(3)(A)(iii).

[207].     Treas. Reg. § 1.48E-5(h)(1), (2) (2025).

[208].     Id. § 1.48E-5(h)(1).

[209].     Id. § 1.48E-4(f)(1).

[210].     See generally Emily S. Bremer, Incorporation by Reference in an Open-Government Age, 36 Harv. J.L. & Pub. Pol’y 131 (2013). Changes to the models could substantially change the operation of the regulations, and so commenters urge the use of notice and comment rulemaking for any changes to the models. See Clean Air Task Force, Comment Letter on Proposed Section 45V Credit for Production of Clean Hydrogen; Section 48(a)(15) Election to Treat Clean Hydrogen Production Facilities as Energy Property 42–44 (Feb. 26, 2024), https://www.regulations.gov/comment/IRS-2023-0066-29565 [https://perma.cc/MH7N-PDV3].

[211].     See GHG Guidelines, supra note 109, at 19–20.

[212].     See generally Luke Hellgren, Pye Russell & Sierra Fraioli, Benchmarking Methane and Other GHG Emissions of Oil & Natural Gas Production in the United States (2023) (finding dramatic variations between companies and basins). That national estimate might itself be too low. See, e.g., Ramón A. Alvarez et al., Assessment of Methane Emissions from the U.S. Oil and Gas Supply Chain, 361 Science 186, 186 (2018); Union of Concerned Scientists, Comment Letter on Proposed Section 45V Credit for Production of Clean Hydrogen; Section 48(a)(15) Election to Treat Clean Hydrogen Production Facilities as Energy Property 12 (Feb. 26, 2024) https://www.regulations.gov/comment/IRS-2023-0066-29522 [https://perma.cc/E2W7-92U3].

[213].     Clean Air Task Force, supra note 210, at 35.

[214].     See, e.g., Coal. for Renewable Nat. Gas, Comment Letter on Proposed Section 45Y Clean Electricity Production Credit and Section 48E Clean Electricity Investment Credit 18–20 (Aug. 2, 2024), https://www.regulations.gov/comment/IRS-2024-0026-1696 [https://perma.cc/V6XL-EZA5].

[215].     Credit for Production of Clean Hydrogen and Energy Credit, 90 Fed. Reg 2224, 2275 (Jan. 10, 2025) (to be codified at 26 C.F.R. pt. 1). The vision seemed to be that the IRS would be able to verify reported leak rates with the EPA, because firms would soon have to report their leak rate to the EPA under a new methane-focused provision of the Greenhouse Gas Reporting Program. However, the EPA has now proposed to end most of the Greenhouse Gas Reporting Program. See generally Reconsideration of the Greenhouse Gas Reporting Program, 90 Fed. Reg. 44591 (Sep. 16, 2025) (to be codified at 40 C.F.R. pt. 98).

[216].     See supra text accompanying note 121 (defining market-mediated emissions).

[217].     See U.S. Dep’t of Energy, Assessing Lifecycle Greenhouse Gas Emissions Associated with Electricity Use for the Section 45V Clean Hydrogen Production Tax Credit 12 (2023) (discussing the use of “reasonable prox[ies]” to estimate the induced emissions of adding new loads to the grid).

[218].     Treas. Reg. § 1.45V-4(d)(3) (2025).

[219].     Letter from Janet G. McCabe, Deputy Adm’r of the Env’t Prot. Agency, to Lily Batchelder, Assistant Sec’y for Tax Pol’y (Dec. 20, 2023), https://home.treasury.gov/system/files/136/45V-NPRM-EPA-letter.pdf. [https://perma.cc/DP2F-ZAVC] (describing the proposed pillars as “a methodological proxy in lieu of quantifying certain indirect greenhouse-gas emissions associated with electrolytic hydrogen production”).

[220].     Wilson Ricks, Qingyu Xu & Jesse D. Jenkins, Minimizing Emissions from Grid-Based Hydrogen Production in the United States, 18 Env’t. Res. Letters 1, 6 (2022) (finding that “when a portion of the highlighted high-quality resource is procured for hydrogen production and cannot be used for grid supply, the system chooses to retire less coal capacity rather than developing significantly more expensive wind resources”).

[221].     Treas. Reg. § 1.45V-4(d)(3)(i). The logic of the nuclear reactor provision is that if the hydrogen project did not buy electricity from a financially struggling reactor, the reactor might retire and be replaced on the grid with a fossil resource.

[222].     Treas. Reg. § 1.45V-4(d)(2)(x)(B).

[223].     Inst. for Pol’y Integrity, supra note 122, at 11–12.

[224].     Section 45Y Clean Electricity Production Credit and Section 48E Clean Electricity Investment Credit, 89 Fed. Reg. 47792, 47804 (proposed June 3, 2024) (to be codified at 26 C.F.R. pt. 1).

[225].     Separately, enforcing the “first productive use” rule would be confusing in practice. One commenter warns of a gaming strategy where dairy farmers might move lagoons of cow manure into new anaerobic digesters in order to claim “first productive use” of the resulting methane production. See Fidelis New Energy, LLC, Comment Letter on Proposed Section 45V Credit for Production of Clean Hydrogen; Section 48(a)(15) Election to Treat Clean Hydrogen Production Facilities as Energy Property 14 (Feb. 26, 2024), https://www.regulations.gov/comment/IRS-2023-0066-29748 [ttps://perma.cc/Q48G-Q8YY].

[226].     Treas. Reg. § 1.45Y-5(e)(3) (2025).

[227].     See One Big Beautiful Bill Act, Pub. L. No. 119-21, § 70521, 139 Stat. 72, 276 (2025).

[228].     See Inst. for Pol’y Integrity, supra note 122, at 12.

[229].     See Energy Info. Admin., U.S. Dep’t of Energy, Renewable Fuels Module of The National Energy Modeling System: Model Documentation 2022, at 19, 76 (2022), https://www.eia.gov/outlooks/aeo/nems/documentation/renewable/pdf/RFM_2022.pdf [https://perma.cc/SKV6-HWPL]. See generally Alex Marten, Andrew Schreiber & Ann Wolverton, U.S. EPA, SAGE Model Documentation (2024), https://www.epa.gov/environmentaleconomics/sage-model-documentation-version-201 [https://perma.cc/JSG5-4TXQ]; Kenneth Hanson, Elise Golan, Stephen Vogel & Jennifer Olmsted, U.S. Dep’t Of Agric., Tracing The Impacts Of Food Assistance Programs On Agriculture And Consumers: A Computable General Equilibrium Model (2002).

[230].     See, e.g., Paul Caron, Moore v. The United States: Will The Supreme Court Join the Whack-a-Tax Shelter Game?, TaxProf Blog (Aug. 6, 2024), https://taxprof.typepad.com/taxprof_blog/2024/08/moore-v-the-united-states-will-the-supreme-court-join-the-whack-a-tax-shelter-game.html [https://perma.cc/945A-U8PQ] (describing the effort to crack down on tax shelters as an “endless game of whack-a-mole”); Brian Galle & Stephen Shay, Admin Law and the Crisis of Tax Administration, 101 N.C. L. Rev. 1645, 1690 (2023) (describing IRS enforcement of tax shelters as “whack-a-mole”).

[231].     See generally ACM P’ship v. Comm’r., 157 F.3d 231 (3d Cir. 1998); Black & Decker Corp. v. United States, 340 F. Supp. 2d 621 (D. Md. 2004).

[232].     The Chirelstein-Zelenak proposal offers more predictability than the economic substance doctrine for two reasons: It describes the circumstances of its application in mathematically precise terms rather than relying on the subjective standard of “change to the taxpayer’s economic position,” and it actively disallows the transactions it describes, rather than merely giving Congress’s imprimatur for the IRS to challenge such transactions at its discretion. See generally Chirelstein & Zelenak, supra note 194. Seealso I.R.C. § 7701(o)(5)(C) (“The determination of whether the economic substance doctrine is relevant to a transaction shall be made in the same manner as if this subsection had never been enacted.”).

[233].     Treas. Reg. § 1.45Y-5(d)(2)(iv) (2025).

[234].     Id. § 1.45Y-5(e)(3)(v) (mandating a negative carbon intensity score of -51 gCO2e/megajoule for combustion of methane derived from animal waste).

[235].     Id. § 1.45Y-5(e).

[236].     See generally Shelley Welton, Neutralizing the Atmosphere, 132 Yale L.J. 171 (2022); Albert C. Lin, Making Net Zero Matter, 79 Wash. & Lee L. Rev. 679 (2022); Yona, supra note 9; Daniel C. Esty & Nathan de Arriba-Sellier, Zeroing in on Net-Zero: From Soft Law to Hard Law in Corporate Climate Change Pledges, 94 U. Colo. L. Rev. 635 (2023); Salzman & Weisbach, supra note 124.

[237].     Renewable Natural Gas Production, U.S. Dep’t of Energy: Off. of Energy Efficiency & Renewable Energy, https://afdc.energy.gov/fuels/natural-gas-renewable [https://perma.cc/CTA6-QJ4P].

[238].     See LCFS Pathway Certified Carbon Intensities, Cal. Air Res. Bd., https://ww2.arb.ca.gov/resources/documents/lcfs-pathway-certified-carbon-intensities [https://perma.cc/M6ZS-23U2] (explaining the negative carbon intensities attributed to biomethane).

[239].     On counterfactual assumptions, one study found that if methane would otherwise have been vented, it would have a highly negative lifecycle emissions value. If, however, it would otherwise have been flared, it would have a negative lifecycle value only if the leakage rate is under 1 percent, which is improbable. Worse still, if the methane were intentionally produced for the sake of receiving the green subsidy (rather than already in existence), it would have a positive lifecycle emissions value. Emily Grubert, At Scale, Renewable Natural Gas Systems Could Be Climate Intensive: The Influence of Methane Feedstock and Leakage Rates, 15 Env’t Rsch. Letters 1, 2 (2020).

[240].     The risk of methane blending with natural gas arises because both are delivered via the same pipelines and methods for differentiating between the two gases once commingled are limited. See Inst. for Pol’y Integrity, supra note 122, at 17.

[241].     See Emily Grubert & Danny Cullenward, The New Hydrogen Rules Risk Opening the Door to Methane Offsets, Heatmap (Feb. 9, 2024), https://heatmap.news/climate/hydrogen-tax-credit-final-methane-offsets [https://perma.cc/A6UW-ZZ8A] (warning about the consequences of allowing methane offsets in the calculation of the Clean Hydrogen Credit).

[242].     See Letter from Sen. Sheldon Whitehouse et al., to Janet Yellen, Sec’y, U.S. Dep’t of the Treasury et al. (Sep. 11, 2024), https://www.whitehouse.senate.gov/wp-content/uploads/2024/09/Letter-to-Biden-Administration-re-45V-Hydrogen-Tax-Credit-09-11-2024.pdf [https://perma.cc/8JMV-5MB3] (arguing for the same restriction); see also Rocky Mountain Inst., Comment Letter on Proposed Section 45V Credit for Production of Clean Hydrogen (Feb. 26, 2024), https://www.regulations.gov/comment/IRS-2023-0066-29657 [https://perma.cc/YX3D-QV88].

[243].     One Big Beautiful Bill Act, Pub. L. No. 119-21, § 70521, 139 Stat. 72, 276 (2025).

[244].     See Treas. Reg. § 1.45Y-5(e)(3) (2025) (continuing to require that taxpayers take into account “avoided emissions”).

[245].     See David Kamin, David Gamage, Ari Glogower, Rebecca Kysar, Darien Shanske, Reuven Avi-Yonah, Lily Batchelder, J. Clifton Fleming, Daniel Hemel & Mitchell Kane, The Games They Will Play: Tax Games, Roadblocks, and Glitches Under the 2017 Tax Legislation, 103 Minn. L. Rev. 1439, 1490–93 (2019) (discussing opportunities under the 2017 tax law to blend income across countries and providing numerical examples).

[246].     See I.R.C. § 951A(a).

[247].     See Stephen E. Shay, J. Clifton Fleming Jr. & Robert J. Peroni, Designing a 21st Century Corporate Tax—An Advance U.S. Minimum Tax on Foreign Income and Other Measures to Protect the Base, 17 Fla. Tax Rev. 669, 706 (2015) (recommending that any minimum tax be determined on a per-country basis).

[248].     See Martin A. Sullivan, Economic Analysis: Will GILTI Save U.S. Multinationals From GLOBE?, Tax Notes (Dec. 23, 2019), https://www.taxnotes.com/featured-analysis/economic-analysis-will-gilti-save-us-multinationals-globe/2019/12/20/2bq6t [https://perma.cc/XP9G-BDL7] (criticizing GILTI for undermining any consequences for activities in low-tax jurisdictions).

[249].     Rocky Mountain Inst., supra note 242, at 18.

[250].     See Mathias Zacarias & Joseph Majkut, 45V or 45Q? How Tax Credits Will Influence Low-Carbon Hydrogen’s Development, Ctr. for Strategic & Int’l Stud. (Feb. 15, 2024), https://www.csis.org/analysis/45v-or-45q-how-tax-credits-will-influence-low-carbon-hydrogens-development [https://perma.cc/5RJK-AZFB](“45V should be the tax credit of choice for all projects with lifecycle emissions below 1.5 kg CO2e per kg H2. These projects are granted a tax credit of $1 per kg H2, higher than what could be attained through 45Q.”).

[251].     This argument pertains only to carbon sequestered in geologic storage. If the taxpayer gets credit for carbon captured and sold to third parties (an option allowed under the Clean Electricity Credits), then another additionality problem surfaces because there is a risk that carbon utilized by third parties will ultimately still escape into the atmosphere. In the OBBBA, Congress amplified this risk by increasing the credit value for carbon utilization up to parity with the credit for geologic storage. See One Big Beautiful Bill Act, Pub. L. No. 119-21, § 70522, 139 Stat. 72, 279–80 (2025).

[252].     The Clean Electricity Credits Final Rule does not allow carbon captured in the course of producing input fuels to count against the emissions rate, but it does allow carbon captured at the generation facility to count against the emissions rate. See Treas. Reg. § 1.45Y-5(f) (2025).

[253].     See Joshua D. Blank & Ari Glogower, The Trouble with Targeting Tax Shelters, 74 Admin. L. Rev. 69, 84 (2022) (describing how attempts to target tax shelters with activity-based rules can be overinclusive and impose unnecessary burdens on fully compliant taxpayers).

[254].     Chirelstein and Zelenak identify the following among noneconomic losses clearly contemplated by Congress that their rule would overinclusively disallow: “the noneconomic loss created by the combination of tax-free imputed rental income and deductible qualified residence interest, see I.R.C. § 163(h)(2)(D) (Thomson 2005); the charitable contribution deduction for appreciation never taxed to the donor, see id. § 170(e); the deduction for amounts contributed to individual retirement accounts, id. § 219(a); and the limited expensing of long-lived assets used in a taxpayer’s trade or business, id. § 179.” Chirelstein & Zelenak, supra note 194, at 1956 n.55.

[255].     See Treas. Reg. § 1.6011-4(b) (as amended in 2010); Blank & Glogower, supra note 253, at 77 (describing how the reportable transactions regime requires the IRS to spend resources investigating transactions that turn out to be legal). For example, taxpayers must disclose any transaction that results in a large tax loss ($10 million in a single year, in the case of corporations, and $2 million in a single year, in the case of individuals)—even though such losses might be fully bona fide. Treas. Reg. § 1.6011-4(b)(5).

[256].     Cf. Julie McNamara, These Are the Critical Issues to Track with the New “Tech-Neutral” Clean Electricity Tax Credits, Union of Concerned Scientists: Equation (Aug. 21, 2024), https://blog.ucs.org/julie-mcnamara/these-are-the-critical-issues-to-track-with-the-new-tech-neutral-clean-electricity-tax-credits/ [https://perma.cc/MG8L-ZL65] (“In the face of uncertainty over input assumptions, analytical boundaries, and/or counterfactuals that have the potential to fundamentally shape the analytical outcome, a conservative approach is appropriate to ensure the tax credits do not inadvertently subsidize a net-harmful outcome for the climate.”).

[257].     See generally Lynn A. Stout, Type I Error, Type II Error, and the Private Securities Litigation Reform Act, 38 Ariz. L. Rev. 711, 711 (1996) (defining the two types of errors).

[258].     These would be variance errors rather than bias errors. See generally Scott Fortmann-Roe, Understanding the Bias-Variance Tradeoff (2012), https://scott.fortmann-roe.com/docs/BiasVariance.html [https://perma.cc/JTL7-K7L3] (distinguishing between bias and variance errors).

[259].     See supra Part II.A (discussing the evolution of carbon accounting to correct past oversights). See generally Yona, supra note 9.

[260].     See, e.g., Charles Cannon, Pieter Gagnon, Gavin McCormick & Wilson Ricks, Assessing the Impact of Voluntary Actions on the Grid 4 (2024) (expressing a lack of consensus on the consequential emissions impact of procurement of clean energy on the grid, which is the central channel of emissions reductions in the Clean Hydrogen Credit).

[261].     Especially once we have already achieved high levels of wind and solar penetration, it tends to be more expensive to produce additional energy with low emissions compared to default fossil fuel alternatives. See James H. Stock, Driving Deep Decarbonization, Fin. & Dev., Sep. 2021, at 12, 14–15 (“The prospect of sustainable aviation fuel competing with petroleum jet fuel . . . is daunting. . . . [N]ew wind and solar power generation is less expensive than coal and natural gas in some but not all parts of the country.”); see also Kenneth Gillingham, Carbon Calculus, Fin. & Dev., Dec. 2019, at 6, 8 (“In the short term, there are some inexpensive ways to reduce emissions, but deeper cuts run up against quickly rising costs.”).

[262].     It is widely understood that the purpose of Pigouvian taxes and subsidies is to make clean energy cheaper relative to fossil fuels, not just cheaper relative to other goods and services. See John Bistline, Neil Mehrotra & Catherine Wolfram, Economic Implications of the Climate Provisions of the Inflation Reduction Act 4 (Nat’l Bureau of Econ. Rsch., Working Paper No. 31267, 2023) (describing the purpose of both the IRA subsidies and carbon taxation as to “lower the relative price of clean to fossil fuel power generation”).

[263].     To be clear, there is no budgeted limit on the possible subsidy expenditure under the Inflation Reduction Act. In the long term, however, Congress’s appetite for green subsidies is presumably subject to whatever budget constraint that applies to fiscal policy generally.

[264].     Cf. Grubert & Cullenward, supra note 241 (explaining that if biogas is allowed a negative emissions rate, a plant that blends biogas with natural gas could qualify for tax credits).

‍ ‍