Distribution Reimagined: Determining a Law’s Ultimate Financing

David Kamin[*]

Abstract

Policymakers are considering enacting either a new spending program or a tax cut. Who pays for the policy? The answer to that question is central to understanding the consequences. Yet, it is a question which, for the most part, policy analysts and scholars have left under-explored and often answered incorrectly or at least incompletely. Analysts and scholars have approached the issue in a variety of ways, including simply believing what policymakers say is the financing in a given piece of legislation, not distributing any financing, assuming the financing is “distributionally neutral,” or assuming the financing comes from the income tax. These approaches are wrong and fail to capture the ultimate financing. This article suggests an alternative: causal analysis rooted in the political economy of the government. It is distribution reimagined. This article begins to explore what it means to analyze financing as a causal matter. Importantly, there will be no certainty in this endeavor—it is necessarily probabilistic and should take into account the inconsistency of policymakers over time.

The approach can fundamentally change our understanding of the distribution of policy. Policies that seem more progressive may be exactly the opposite when the ultimate financing is properly considered. And, it can lead to a re-evaluation of the wisdom of policy. Given the political economy of the United States, new benefits could be financed to a significant degree by those with lower incomes, despite what policymakers may say at the time, and this should be a serious concern for those who would prioritize resources for those with less. If policymakers continue to be averse to the types of financing sources that reduce this risk, it suggests greater reason to keep fiscal benefits targeted to those with greater need. In the end, irrespective of one’s policy preferences, this type of causal analysis to determine ultimate financing is critical to understanding the effects and wisdom of fiscal policies.

I. Introduction

The ultimate consequences of fiscal policies—decisions to tax and transfer—are almost never as certain as they are made out to be, are often different from those described, and should involve analysis that is rarely done. The missing analysis can have important consequences for ongoing policy debates, including the question of how universal or targeted government benefits should be, and others beyond that.

The issue—and, really, the challenge—is the financing. The fact that it matters how a policy is financed is not a new observation. In law schools, introductory courses to taxation often motivate the whole field with some lines about how the students are about to study how the government is paid for.¹ This is not to mention that perhaps the single most famous quote in the law about taxes involves the idea of financing: “Taxes are what we pay for civilized society,” said Justice Oliver Wendell Holmes.²

But what exactly pays for what? Or more precisely, and focusing on the issue of new policy, when fiscal policy is being changed, who is getting benefits and who is paying for those benefits? This is a question that has vexed scholarship and policymaking alike. However, the approaches often used—such as believing what policymakers say is the financing in legislation, assuming financing is distributionally neutral, assuming the financing comes from the income tax only, or not distributing any financing—are insufficient.³ They do not reflect an estimate of the actual ultimate financing.

Let’s take the first standard approach. This approach focuses on what policymakers say is the financing, and sticks to the boundaries defined by policymakers, especially in the legislation itself.⁴ If policymakers pair a benefit and a tax together in a law, then the question of what is paying for what has been answered, or so this approach goes.

To make this concrete in terms of just one recent policy proposal: the United States, unlike almost all countries in the world, lacks a national paid family and medical leave program.⁵ And there have been efforts to change this with proposals made by both congressional Democrats and the former Biden administration. However, there has been a divide on how to finance the benefit. The long-standing and leading congressional proposal developed by Senator Kirsten Gillibrand and Representative Rosa DeLauro used a 0.4 percentage point payroll tax that applies across the income spectrum and that is divided evenly between employer and employee contributions (though, as re-introduced in the current Congress, the legislation dropped any specific financing source).⁶ Payroll taxes apply to wages, and the proposal would have split the tax paid between employees and employers, though the economic consensus is that much of the actual burden of such a payroll tax would be on the employee.⁷ But when the Biden administration proposed national paid family and medical leave at the start of the administration, it used a combination of tax increases on high-income Americans and large corporations to finance it,⁸ and that was what was reflected in the bill that the House of Representatives eventually passed before paid leave (and other social measures) were cut from the legislation during Senate negotiations.⁹ On its face and using that standard approach of analyzing distribution based on the financing given in the legislation, the Biden plan was much more progressive.¹⁰

But analyzing distribution based only on what’s packaged in the legislation, while intuitive, is wrong. Policymakers’ statements and the boundaries of legislation do not necessarily define what is actually financing what. These are formalisms, not an answer to the question of the ultimate source of financing.

What does it take to determine the ultimate source of financing? It takes causal analysis.¹¹ The causal analysis should be rooted in the political economy of the government,¹² and the causal analysis should incorporate the potential for inconsistency in policy preferences and policymakers over time. The enacting President and Congress may want a certain kind of financing, but a very different President and a very different Congress might decide that. There will be no certainty in this endeavor—the exercise is necessarily probabilistic. Resources devoted to a given activity will have a range of possible financing, with some sources more likely than others.

To go back to the paid leave proposal: who is financing the benefit if it were to be enacted? We will not know for sure, irrespective of what policymakers say. As I will describe in this article, the approach of a broad-based payroll tax—relative to corporate and high-income tax measures—probably reduces the risk that the actual financing comes from programs important to low- to middle-income Americans and is more likely to be the actual financing. That is because it is a custom financing source tied politically to the benefit. By contrast, the corporate and high-income revenues are more likely to have been enacted anyway—they are not as politically tied to the benefit—and this means that the ultimate financing is more likely to be some other policies, including potentially highly regressive ones.¹³

So, if we use a standard approach of analyzing distribution—what is in the legislation itself—the payroll tax is more regressive in financing than the combination of corporate and high-income revenue raisers.¹⁴ But, reflecting on the actual ultimate financing, the end-result could well be exactly the opposite.¹⁵ Further, given the Biden approach, universal paid leave covering higher-income workers and financed by the federal government may be a questionable endeavor since there is greater risk that these benefits for those with higher incomes ultimately are financed from regressive sources.¹⁶

This analysis can matter not just descriptively but also in terms of whether a policy is wise to pursue. In this article, I focus on what I believe to be a crucial normative concern for judging U.S. fiscal policy: the effects on the living standards of those with low- to middle-incomes and with greatest priority for those with the lowest incomes.¹⁷ While that normative criterion will be familiar and agreed to by many,¹⁸ the analytical approach to distribution described here should be relevant to those with different values. Whether one is most concerned with overall efficiency or some other normative criterion, the issue of what pays for what remains central to the evaluation of any policy.

To be sure, there will be some policies that are so beneficial that their ultimate financing may make little difference in assessing whether they improve matters on net. But there are many other policies that would only be worthwhile with some, but not all, plausible financing sources.

Take, again, paid leave benefits for those with higher incomes. While this form of social insurance is in line with countries around the world and is a good policy, the financing matters in evaluating it. From my perspective and given my normative framework, the policy should not be pursued if those who end up paying for those benefits are those with lower incomes, rather than higher income Americans financing their own paid leave benefits on net. Thus, in assessing the wisdom of universal paid leave, it matters whether policymakers have credibly committed to financing that does not do that.¹⁹

Another way of viewing this is as a policymaker or social planner with limited control—control over a benefit but more limited control over how that is financed over time. This view should also apply in the reverse, namely, control over financing but not necessarily what it is used for. What policies should that policymaker pursue? The policymaker should only pursue ones where that policymaker believes the policy to be net socially beneficial, taking into account their lack of control over how it might be paid for.²⁰ And, the policymaker should not believe their own rhetoric about what they say is paying for it. They should assess what could be the ultimate financing, and take into account the broader political context.²¹

And, in the United States, this political context gives reason to be concerned that new benefits could be financed by those with lower incomes.²² Policymakers can commit to specific financing to some degree to reduce this risk. The degree of commitment should be understood on a spectrum. There is little to no commitment when a law is explicitly debt-financed; strong commitment when the financing is likely unique to the benefit involved; and something in between when it comes to financing that has a higher likelihood of being used for some other purpose.²³ It is this framework that distinguishes the previous congressional paid leave plans, which were paired with payroll-tax financing likely unique to the benefit involved, from the Biden paid leave plan, which was paired with policy that could finance other benefits.²⁴

This reimagining of fiscal policy then has important implications for a continuing debate in tax and transfer policy: namely, how universal versus targeted government benefits should be.²⁵ Should benefits be more like Social Security and the proposed universal paid leave program with very broad eligibility including for those with the highest incomes or be more like Medicaid and food assistance being highly targeted by income? The standard understanding is that broad benefits do not necessarily come at the expense of those with lower incomes—since the tax system can be adjusted to achieve any degree of desired redistribution in combination with the benefit system.²⁶ However, while that is a possible outcome, it is not a certain one. For that to be likely, policymakers must commit to financing which has that effect.

However, policymakers in the United States have, in recent years, been less willing to embrace the kind of broad financing measures that uniquely match to more universal benefits, like the payroll tax in the previous congressional proposals for paid leave that was dropped as President Biden attempted to legislate paid leave.²⁷ Among other issues, this financing has been interpreted to violate tax pledges made by leading policymakers on both sides of the political aisle.²⁸ The unwillingness to embrace broad financing measures raises a specific concern to pursuing more universal benefits in the United States unless policymakers change their approach to financing: that there is real risk that the benefits do come at the expense of those with less resources. This does not resolve the debate of universalism versus targeting, but it presents an issue that so far has received too little attention—and helps show the importance of reimagining how distribution of a policy is analyzed.

In titling this article as “Distribution Reimagined,” I am using the term “distribution” broadly to encompass the issues that come with how a policy is financed. That includes all effects on people’s well-being, whether it be resources directly paid to the government or changes in behavior affecting people’s well-being. So long as one cares about how a policy is financed, the analysis this article describes is relevant.

Importantly, this article focuses on policies where there are resource trade-offs. While some policies, especially investments in children, may generate economic returns so large that they finance themselves, many policies—including ones well worth doing for reasons other than generating a large economic return (such as fairness or other goals)—do not.²⁹ Further, the government does have a budget constraint, meaning that the government does not have the unlimited capability of deploying resources without getting them from somewhere (if the policies in fact are not paying for themselves).³⁰ We might argue about how expansive that budget constraint is and whether some free lunches are available just by the government borrowing funds at low interest rates for instance, but so long as the government is on course to fully use the budget available to it, the net deployment of resources in one area must come at the cost of resources elsewhere.³¹ That trade-off does not have to be immediate. In simplest form, the budget constraint is expressed through the “infinite horizon” and so the timing is not defined by this arithmetic,³² but it must happen at some point. It is this trade-off of resources and the distributional effects—benefit and financing—that is the focus here.

Part I of this article compares the current approaches for analyzing financing to the causal analysis that asks what is ultimately paying for what. Part II discusses how causal analysis is uncertain and begins to explore the normative implications of that uncertainty. Part III begins to bound that uncertainty in the U.S. context, explores the real risks in the United States that those with less do end up financing benefits, and considers ways in which policymakers can commit themselves when it comes to financing in ways that are more likely to avoid that outcome. Finally, Part IV brings this analysis to bear on the debate about whether benefits should be more universal or targeted, describing how more universal benefits generate greater distributional risk in the United States now and unless policymakers change their approach to financing.

II. Approaches to Financing

Analysts and scholars have taken a number of different approaches to defining what fiscal policies pay for what, and, while some—especially in the context of deficit financing—have approached the analysis called for here, this has not been done in comprehensive fashion.

This Part begins by describing some of these pre-existing approaches before arriving at the causal analysis actually needed to answer the question of what really is paying for what.

A. The PAYGO Approach

One approach to distributional analysis sometimes used by scholars and analysts follows what I will call the “pay as you go” (or PAYGO) approach. The moniker comes from the rule that policymakers sometimes impose on themselves to pay for new tax and certain spending programs as they are enacted.³³ The focus here is not the particular nature of those PAYGO rules, but, instead, how a PAYGO analytical approach tends to govern distributional analysis of policymaking, both inside government and out.

1. How the PAYGO Approach Works

The PAYGO approach is simple: commitments of government resources and increases in those resources included in the same legislation should be analyzed together. One policy pays for the other, so distributional (and other) consequences should be seen as connected and normative consequences analyzed together.

The PAYGO approach dominates in the analysis of legislation. Take legislation considered and enacted under both President Trump and President Biden.

Start with President Trump. His most significant legislative achievement in the first term came in the form of tax cuts enacted in 2017, sometimes known as the Tax Cuts and Jobs Act (TCJA).³⁴ That piece of legislation included many different changes to the tax code, both corporate and individual. Some raised revenue, and others lost revenue. Overall, the legislation lost more than it raised to the tune of $1.5 trillion over ten years in the initial estimates—but this was netting approximately $5 trillion over ten years in revenue-losing provisions and $3.5 trillion over ten years in revenue-raising provisions and all packaged together.³⁵ And the analysis of the TCJA largely assumed that provisions were trading off against one another within the bill.³⁶

The same was true in President Trump’s second term as he enacted another major piece of fiscal legislation—the One Big Beautiful Bill Act (OBBBA).³⁷ That involved the extension of many of the same tax provisions in the 2017 legislation that had been set to expire, the addition of new tax cuts, and significant cuts to health programs and nutrition assistance, among other things.³⁸ And, the distribution of this legislation was analyzed in much the same way, based on what was included in the law itself and the trade-offs presented there.³⁹ Again, this means that provisions with no obvious relation to each other—other than their placement in the same piece of legislation—are largely analyzed as one.

Under President Biden, the Democratic Congress in place in 2021–22 considered a variety of versions of major domestic policy legislation, under the name for some time of Build Back Better but eventually enacted as the Inflation Reduction Act.⁴⁰ In its initial form, it featured a combination of social insurance, education, labor market, and climate policies—including the national and universal paid leave program—and was paid for with a combination of high-income and corporate taxes, tax enforcement, and savings from prescription drug reforms, almost all abiding by President Biden’s pledge to not raise taxes on those making less than $400,000 of income.⁴¹ In its final form, it was a more narrowly tailored bill focused on clean-energy oriented incentives and health care, and with similar financing, largely dropping the proposed tax increases on high-income Americans and reducing the increase on corporations.⁴² There was considerable debate about the wisdom of the legislation. But in all of forms of the legislation, it was largely assumed by various commentators and analysts that the taxes packaged in the legislation—with the legislation then characterized as “paid for”—were what was actually financing the benefits in the legislation.⁴³ This was true even though many of the financing measures were largely unrelated to the benefits delivered by the bill.

2. The Limits of the PAYGO Approach

The problem with the PAYGO approach is that the measures being packaged as financing may or may not be truly linked to the benefits being financed. If they likely are, then the trade-off has been identified. If they are not—and all of these recent legislative endeavors had benefits and financing that were not tightly linked—then the legislative boundaries are not accurately describing the trade-offs. While the PAYGO approach will always have a role in distributional analysis, and may be the only approach that official government scorekeepers should use given the nature of those institutions,⁴⁴ it can also be entirely misleading as to the actual distributional implications of legislation as a whole and individual provisions within that legislation. Analysts and scholars should approach the PAYGO method with that in mind.

Take a simple example of legislation. Provision A is a new benefit. Provision B is a tax increase that exactly pays for that benefit, and they are both in the same piece of legislation. The PAYGO approach would say the distributional effect of the bill is defined by “A+B.” But are A and B in fact causally connected? Let’s say that, absent A, B would’ve been enacted anyway and used to pay for C, which is a new benefit, and, with enactment of A, C no longer gets enacted since it lacks financing. In that case, the actual effect of enacting A is not “A+B;” instead, it’s “A+C.” C is what is ultimately trading off against A. Or it could be that B would’ve been enacted anyway to pay for a pre-existing benefit, D—and now D will be cut as a result. In that case, the ultimate effect of enacting A would be “A+D.”

The boundaries of legislation are defined by the interaction of political convenience and established legislative process. But those factors do not necessarily define what is in fact causally connected to and what runs contrary to the PAYGO approach of distributional analysis.

B. The Distributionally Neutral Approach

Another method—associated with the leading legal and economic thinker Louis Kaplow—is to assume distributional neutrality in the analysis of a discrete policy.⁴⁵

1. How the Distributionally-Neutral Approach Works

Kaplow urges, for analytical purposes, to assume that tax expenditures, public investments, and regulations are matched with distributionally-neutral income tax changes as financing.⁴⁶ Under his approach, it does not matter for distributional purposes what is packaged by policymakers with the benefit. In all cases, his approach assumes an income tax that pays for the benefit and in a way that holds distribution constant.⁴⁷

In setting out this methodology and comparing it to what I call the PAYGO approach to distributional analysis, Kaplow questions focusing on what policymakers say is paying for a policy. He notes, when it comes to what policymakers say:

[Offsetting financing] may not be specified at all, it may vary across proposals, or it may change. Specification [by policymakers] may also be meaningless because the tax adjustment being proposed (say, closing a particular tax loophole) is likely to be one with sufficient legislative support that it would be used to finance a subsequent proposal if the current one is not enacted.⁴⁸

This critique is similar in spirit to my own of the PAYGO approach to distributional analysis.

Take a policy like paid leave. The Kaplow methodology would essentially suggest that there is no meaningful difference between enacting a payroll tax with paid leave or a combination of corporate and high-income tax increases. The paid leave policy would be one and the same, offset by a distributionally-neutral financing rather than what policymakers suggest is the financing.

In describing his approach, Kaplow notes how it creates consistency in evaluating policies and focuses analysis on the efficiency effects—the effects on economic incentives—of the relevant policies. When it comes to consistency, he says: “[C]onsistency has virtues; otherwise, similar analyses of otherwise identical policies could reach widely different conclusions due entirely to disparities in ad hoc assumptions about how the policies will be financed.”⁴⁹ In addition to consistency, the exercise of policy analysis under his framework then focuses squarely on the efficiency of a given policy—rather than the distributional effect—and in a way that can be captured in cost-benefit analysis style approach. Kaplow continues: “As it turns out, this combination of policy and finance mechanism leaves labor supply unaffected. Since both distribution and labor supply remain unchanged, traditional principles like the cost-benefit test will provide proper guidance [in policy analysis].”⁵⁰ In a related and influential set of arguments, Kaplow and Steve Shavell have noted that, if distribution is held constant, regulatory policies should be pursued to maximize efficiency as measured via cost-benefit analysis, since they conclude that distributional maladies would be more efficiently addressed via income taxes than regulation.⁵¹

2. The Limits of the Distributionally-Neutral Approach

The Kaplow approach is analytically clean and focused, but as Kaplow acknowledges, it comes with a significant limitation: the method puts to the side the critical question of what is actually financing the given policy.⁵² Kaplow simply adopts distributional neutrality as a matter of assumption.

And the realism—or lack thereof—of that assumption has been a central critique of the Kaplow methodology.⁵³ This is a critique I join. There is no reason to expect that distributional policies are simply held constant over time. First, the effects of federal government policies on key aspects of distribution have been changing significantly over time. Take the effects of federal policies—both transfers and taxes—on poverty. One metric shows federal policies having reduced the poverty rate by only about four percent in the late 1960s; today, those policies cut the poverty rate by around fifty percent.⁵⁴ Similarly, when put in terms of overall inequality in the United States, federal tax and transfer policies have changed in their effect over the last few decades, generally increasing the degree to which they reduce inequality.⁵⁵ Second, taxes and transfers are contested.⁵⁶ There are active debates in the United States about changing distributional policies, including for instance significant changes to programs supporting those with the lowest incomes, and uncertainty about how those debates resolve.⁵⁷

However, my critique of the Kaplow methodology and the kind of distributional analysis I suggest may also be seen more as a complement rather than an alternative to what Kaplow suggests. Kaplow’s work acknowledges the challenge and importance of identifying who is actually paying for what. And he recommends his distributionally neutral approach in part because he assigns value to specialization in analysis. Those who might be good at analyzing the direct effects of a paid leave policy may be different than those who would be good at analyzing how it would be financed.⁵⁸ Understood in this way, this article and its suggested methodology is a complement, exploring what it means to take that second step. It is only in tension with Kaplow’s approach to the degree Kaplow suggests, as he does at times, that distributional effects of this or that policy may simply offset over time because of some settled “political equilibrium.”⁵⁹ History and current events suggest there is no such equilibrium in the United States,⁶⁰ and, if there were, it would need to be established analyzing the political economy of the country and through the kind of analysis this article suggests.

In sum, Kaplow’s approach can be seen as a helpful simplifying assumption for engaging in a partial and consistent analysis of the effects of policies, but it does not answer who is actually paying for those policies.

C. Marginal Cost of Public Funds

Another important strain of the public finance literature addresses how to evaluate fiscal policies, including the effects of financing, but does so through largely ungrounded assumptions of what finances marginal government policy. This approach is called the “marginal cost of public funds” (or MCPF).

1. How the Marginal Cost of Public Funds Works

As one book giving an overview of this method puts it, “The marginal cost of public funds measures the loss incurred by society in raising additional revenues to ﬁnance government spending . . . . It is the key component in evaluations of tax reforms, public expenditure programs, and other public policies, ranging from tax enforcement to privatization of public enterprises.”⁶¹ And on its face, this sounds exactly like the project of this article—what should we consider to be financing a given fiscal commitment and how do the financing and the commitment trade off against each other?

However, MCPF—much like the Kaplow approach—tends to work by assuming the source of the financing. The typical assumption in this literature is that the financing in the United States is done via a proportional (i.e., flat rate) labor income tax and the distributional and efficiency consequences are calculated on that basis. This is irrespective of the nature of the project and its own distributional effects.⁶²

2. The Limits of the Marginal Cost of Public Funds

Thus, the MCPF approach comes with a significant limitation. The frequently-used assumption of a flat-rate labor income tax as financing is not grounded in the U.S. or any specific government’s political economy.⁶³ In addition to the fact that the income tax may not be the marginal financing for a new policy, the assumed income tax is not even the same income tax that the United States has enacted in law.⁶⁴

The critique is similar to that of the Kaplow approach—of assuming distributional neutrality—since there is a lack of analysis on what actually finances a given policy. But, unlike with Kaplow, the lack of grounding is even more opaque. In fact, Kaplow frames his distributionally neutral approach as an alternative and analytically superior way of engaging in policy analysis versus the MCPF approach.⁶⁵ He does so since the assumptions are clear; the method explicitly puts to the side the actual financing of a given policy; and his approach cleanly focuses on the efficiency effect of a given policy change without involving distributional considerations.⁶⁶

This weakness of the MCPF approach in practice—assuming particular financing without any strong grounding for doing so—has helped prompt a recent innovation in the public economics literature, which now offers an alternative: the marginal value of public funds (or MVPF).⁶⁷ Nathaniel Hendren and Ben Sprung-Keyser have defined this as an alternative way of measuring the benefit-to-cost ratio of a policy.⁶⁸ Their approach “take[s] the ratio of the benefits to net government costs to generate each policy’s marginal value of public funds.”⁶⁹ As they put it, it reflects the “bang-per-buck” of a policy.⁷⁰ One key advantage of their approach is that it does not assume a particular financing source. They recognize that the financing source is uncertain, and a given policy could be trading off against a wide range of fiscal alternatives ranging from taxes to spending programs. As such, the MVPF approach helps establish a menu and evaluate policy based on what one believes is in fact ultimately trading off.⁷¹ They do not put forward a theory of how to answer that question—the answer is the topic of this paper—but the approach effectively tees up that very question. In this way, it is like the Kaplow approach but avoids ever assuming a distributionally neutral offset. It is clear from the start that one has to engage this issue of trade-offs and then compare the MVPF of alternatives.

D. Distributing Increased (or Reduced) Borrowing

There are a variety of methodologies that analysts and scholars use when faced with a policy for which policymakers have not specified any particular financing. In this situation, the relevant policy is in a package that either increases or reduces borrowing, since it is not matched with any financing. There will still be a trade-off. If the government is already on track to make full use of the resources available to it, borrowing does not avoid a trade-off. Financing may simply come at a different time.⁷² And, to determine what the financing might be, some of these methods begin to approach the kind of analysis I recommend for policies more broadly.

The introduction of borrowing means the PAYGO approach to financing is impossible. Since legislators have not specified financing, analysts and scholars cannot rely on what policymakers have said to establish the source of financing, whatever the PAYGO method’s flaws. Some of the other approaches—such as Kaplow’s distribution-neutral financing and MCPF’s assumed financing from an income tax—can still work in this situation, since these methods do not rely on what policymakers say and assume a financing source. Nonetheless, changes in borrowing have pushed analysts and scholars—perhaps especially those who would otherwise rely on the PAYGO method—to begin engaging in the imaginative exercise of determining ultimate financing.

To be clear, the vast bulk of analysis of deficit-financed policies does not try to determine financing at all. The analysis, following the PAYGO approach, still sticks to what is within the four corners of legislation, but, since financing is not included, the analysis shows the policy without financing, or at least without full financing.⁷³ In that situation, the analysis tends to focus on who wins more or less not including financing (or who loses more or less if the legislation is largely deficit reducing). With debt financing alone, it is possible that almost everyone ends up being a “winner,” since those ultimately paying for the policy are not included, and the analysis and debate then become focused on who wins more.⁷⁴ There is still something meaningful to say about how progressive or regressive these changes are and focusing only on who wins more or less based on what’s within the four corners of the legislation, but it is limited in important ways by ignoring the eventual distributional effects of financing.⁷⁵

One set of scholars, led by the economist Bill Gale, has taken a different approach when it comes to deficit-financed tax legislation—and this is where the analysis begins to approach that urged in this article, except more broadly.⁷⁶ They calculate different financing scenarios and then display distributional tables showing the combined effects, explicitly engaging the question of what the ultimate distributional effects will be, recognizing that those effects will not be confined to the four corners of the legislation. It is a constrained form of the analysis being suggested here for distributional questions broadly—since it is done only to the degree tax legislation is explicitly deficit financed and with relatively little discussion of which financing scenarios may end up being most plausible.⁷⁷

For example, for the 2017 tax cuts, Gale and co-authors showed how, under some scenarios, the tax cuts could end up with many middle- and lower-income families losing on net financially from the tax cuts.⁷⁸ They specifically included three ways that the borrowing could eventually be paid for as illustration: equal-per-household, proportional-to-income, and proportional-to-income-taxes. Under the first two, the tax cuts—combined with financing—on average leave those in the bottom three to four quintiles worse off and the opposite toward the top of the income distribution. However, under the final financing option—proportional-to-income-taxes—the opposite is the case.⁷⁹ In this analysis, they focus on the eventual financing of the $1.5 trillion in unpaid tax cuts, and assume that the remaining $3.5 trillion of tax cuts and offsetting tax increases are trading off against one another as policymakers say they are, sticking with the PAYGO approach to the trade-offs.

The question then is what is the most plausible of the financing scenarios. Gale and co-authors sometimes leave that exercise to the reader,⁸⁰ but other times begin to point in a more specific direction. In an earlier analysis of the Trump-proposed version of these tax cuts, the authors noted that the two regressive scenarios more closely resembled the then-administration and congressional budget proposals, which sharply cut safety net spending, and, for that reason, pointed readers more in that direction.⁸¹ As I lay out later, such cuts to safety net spending are in fact a very real risk in the United States where policymakers have not committed to another specific form of financing,⁸² but the exercise should be seen as a probabilistic one and grounded in the political economy of the government involved.

The scholars Daniel Hemel and Kyle Rozema engage a related issue in the specific context of analyzing the distribution of tax expenditures and the home mortgage interest deduction.⁸³ They describe how calculating the distributional consequences of a tax expenditure unto itself shows only the effects of its repeal and not what the increased revenues would finance. It is the flipside of the same coin as debt-financed legislation. In this case, there is not legislation at all; just a provision as written into law, and repealing the provision would increase revenues. Looking at this alone is akin to looking alone at a piece of debt-reducing legislation, and with the four corners of the legislation including only repeal of the tax expenditure. Hemel and Rozema emphasize that analyzing the distribution of the tax expenditure requires something more: imagining what, if the tax expenditure is repealed, the revenue would pay for and with that determined by the political context.⁸⁴ To do that, Hemel and Rozema engage in an exercise similar to Gale and co-authors, showing a variety of potential financing scenarios (in this case, the finance being a new benefit) and illustrating how the repeal of the home mortgage interest deduction could range from having progressive effects to regressive ones depending on how the revenue is used.⁸⁵ The exercise in the end must be grounded in an analysis of the likelihood of the different scenarios—grounded in political economy—and Hemel and Rozema call for people to specify their assumptions and scenarios in analyzing the distribution of tax expenditures.⁸⁶ It is another step in the right direction of fiscal policy analysis, just one that must be built out and done more broadly.

A final strand of the literature on fiscal policy also engages this question though doing so not in evaluation of any specific policy but rather the total distribution of national resources. There has been a heated academic debate about trends in overall income inequality in the United States, and one element of that debate is how to distribute federal borrowing and specifically who will eventually pay for that debt.⁸⁷

The analysis involves distributing national income, and national income effectively excludes resources generated by borrowing under the assumption those resources will eventually be paid back—and so that then requires hypothesizing who will pay for government borrowing. One set of scholars assumes the deficits are financed in proportion to after-tax income across the distribution—meaning any deficit financed policy is assumed to affect everyone equally relative to income.⁸⁸ Another set of scholars, by contrast, assumes that deficits are financed in proportion to federal income and payroll taxes, a much more progressive financing source.⁸⁹ The first set of scholars, partly as a result, find greater overall inequality in the distribution of national income than the second set, since they assume borrowing is financed more regressively. Neither set of scholars engage in depth in justifying the different assumptions. To determine which is more convincing requires focusing on U.S. political economy—a task taken up in this article. And as I describe, there are significant risks, unfortunately, that low-income Americans end up financing reductions in the deficit and new fiscal commitments⁹⁰ and in a way that aligns with the assumptions of the first set of scholars or results that are even more regressive. Still, both sets of scholars recognize that it is a guess about what the political system will produce, though one that tends to get buried deep in these articles.

This distributional literature so far has recognized that deficit financing complicates the PAYGO approach to distributional analysis. For much of the literature, this becomes a limitation on analysis—something like “we can only say so much and what we can say is defined by what lawmakers wrote down so far or what the provision itself says.” For another strand of the literature, this has pushed in the direction of beginning to imagine what that financing might look like since the benefit will eventually cause some offsetting financing to be enacted. This article picks up on this strand, and both tries to better ground the exercise of imagining financing and to apply the method more broadly to all fiscal legislation, recognizing that—even where legislation is not deficit financed—benefits and financing may have effects beyond what is laid out in a bill. What causes what, ultimately, should not be understood purely in terms of a legislative package or the bounds of a specific provision. It should be understood in terms of causality, the topic of the next section.

E. Causal Analysis Rooted in Political Economy

The previous four sections explore methods developed in the literature so far for matching policies with financing, and the limits of those methods. This section develops the method recommended by this article: causal analysis, rooted in political economy.

What does it mean to do a causal analysis of the financing of a fiscal policy? It means analyzing what, as a probabilistic matter, will ultimately be used as financing, and would not otherwise be enacted. What set of fiscal changes would not be enacted “but for” the enactment of the given benefit?⁹¹

To go back to the example given earlier: provision A is a new benefit. There may be a spectrum of possible financing that could be used to pay for it, with different characteristics both in terms of the efficiency of the financing and the distributional consequences (and along any other dimensions of import). The probabilities will very much depend on the politics and fiscal position of the government involved. The analysis is surely uncertain, and that in itself has consequences.⁹² However, it is possible to begin to bound this uncertainty by grounding analysis in a government’s political economy.

Some ways of financing are more or less likely than others based on actions policymakers have taken so far, the statements they have made, and other evidence about the political economy of the government involved. Even if policymakers say nothing at the time about what is paying for “A,” we should be able to create some evidence-based range of plausible ways it could be paid for and begin to assign probabilities to that. Part III begins to do that in the United States context when it comes to the financing of a new benefit.

This approach does not entirely reject the idea that policymakers may be able to—at least to some degree—commit in terms of what is financing what. I reject the PAYGO approach not because policymakers have no ability to change the probabilities of what is financing what, but instead because the bounds of a particular piece of legislation are not fully defining of those probabilities.

Political commitment when it comes to financing should be understood on a spectrum. There is debt financing where policymakers say very little about what will eventually finance the benefit. That of course does little to commit, other than to the degree policymakers have articulated pre-existing positions on how the government should be financed generally. Now, take the opposite end of the spectrum: policymakers enact a benefit and its financing, and the financing is tightly linked in political logic to the benefit. Where such financing exists, it is “custom financing” that very likely would not have been enacted otherwise. That reflects a high level of commitment, though still is not certain. Finally, take a level of political commitment that is between these two poles: policymakers identify financing and enact it along with a benefit, but it is financing that has a higher probability of otherwise being used elsewhere. The financing comes from something like a “fixed menu” of financing, and even if it were not enacted here, there is some significant probability it would’ve been picked from the menu later on and for other purposes. This is the kind of financing that Kaplow identifies in passing when explaining why legislative packages do not define what is financing what.⁹³ Some weight should be given to what policymakers then say is paying for what, but less than if the financing truly appears to be “custom.”

In Part III.C., I explore financing in social insurance and tax reform to help differentiate custom from fixed menu financing in the United States. But, to get a bit more concrete on how this could work, let’s return to the example offered at the start of this article: the previous congressional approaches versus the Biden approach to financing paid leave.⁹⁴ As discussed, on its face and using the PAYGO approach to distributional analysis, the Biden approach is the more progressive one since it was financed with a combination of high-income and corporate tax provisions versus the congressional proposal of a payroll tax.⁹⁵ However, the congressional proposal had what I call “custom financing”: the payroll tax was more uniquely tied to the benefit as a political matter. Broad-based payroll taxes have not been proposed in recent years by leading policymakers as general financing mechanisms for the government. Instead, they have been only considered—and so far rejected—in limited circumstances like this to finance benefits tied to the payroll taxes.⁹⁶ That is not the case for corporate and high-income tax measures; these are proposed regularly by Democrats to finance government broadly.⁹⁷ As a result, the combination of paid leave benefits and payroll taxes is more likely to represent the actual financing trade off. By contrast, the combination of paid leave and corporate and high-income raisers is less likely to; paid leave in that case could effectively be financed by whatever those measures might have otherwise paid for. This includes programs that are much more concentrated in terms of their benefit toward the bottom of the income distribution. As a result, what appears on paper to be more progressive may not be at all. There is high uncertainty, and some greater probability of the benefit effectively being financed by regressive sources like cuts to the safety net.⁹⁸

While this form of analysis is necessary to understand the implications of policy and should affect how any policymaker should consider the effects of their actions, there will be some analysts in better positions than others to approach these questions. I am not calling for a change to how official government scorekeepers do analysis of policy proposals. For the most part, such analysis done by offices working for Congress and the executive branch proceeds using the PAYGO framework—respecting the boundaries of legislation and giving fiscal effects and sometimes other estimates including distributional ones.⁹⁹ Doing such analysis effectively requires clear rules and for policymakers to accept the relative impartiality of the analysis as done by nonpartisan staff. Engaging in the constructive speculation that is at the core of causal analysis would be in some tension with that. However, for the policymakers themselves, for outside analysts, for academics, for anyone interested in the real effects and wisdom of policy—this kind of analysis is necessary to answer what policies are ultimately doing.

But how much does this matter in terms of the wisdom of policy? The next Part begins to take up this question largely in the abstract, focusing on how the uncertainty of financing can have significant normative implications for the wisdom of enacting certain benefits. The two parts that follow then begin to bound the uncertainty in the U.S. political context and apply those lessons to the ongoing policy debate of how universal or targeted new benefits should be.

III. Uncertain Financing, Unwise Benefits?

Financing of a policy should be determined using causal analysis rooted in the political economy of the government and taking into account the uncertainty of that exercise. But how much does using this approach matter in evaluating the wisdom of policy and, specifically, the wisdom of a new benefit? That is what this Part begins to consider, and I describe how a benefit which would be wise to pursue in some circumstances may actually be detrimental, correctly evaluating the range of potential financing sources.

The basic point is that the financing could encompass bad outcomes, however defined, in terms of who pays and how, and that risk might undermine the case for giving a benefit in the first place. The question is not whether a benefit is worth pursuing if paired with the financing preferred by the policymakers enacting the benefit as under the PAYGO approach, or if it is paired with financing defined by assumption rather than analysis as under the distributionally-neutral approach. The question is whether the benefit is worth pursuing if it is paired with the kinds of financing that would ultimately be put into place and considering the uncertainty of that financing.

To put it a bit differently and biographically, I have served in both the Biden and Obama administrations. The leaders of those administrations enacted significant fiscal policies, and both administrations tried to enact more. The wisdom of those policies would depend on not just how those particular presidents and the Congresses with which they worked would prefer to finance the policies, but how future presidents, including President Trump and the Republican Congress now in place, would do so. And for the current Republican administration and Congress, the same is true—it might fall to future policymakers with very different fiscal priorities to decide on the financing of their commitments.

A. Time Inconsistency and the Social Planner

Core to the normative concerns that come with the uncertain causal chain of financing are two critical assumptions. The first assumption is that the social planner is not consistent over time.¹⁰⁰ By this, I mean that the preferences of those in charge of policy change over time because of the electoral cycle or for other reasons. The second critical assumption is that some social planners are better than others so that a “good” social planner must consider the actions of “bad” future social planners.

Let’s, for a moment, assume a consistent and consistently powerful social planner.¹⁰¹ In that case, the social planner does not have to be concerned that a benefit will be matched with a source of financing that is a poor one along whatever dimensions the social planner cares about. After all, the social planner will be in charge. In deciding whether to pursue a benefit, the social planner can simply ask what kinds of financing the planner expects to use—and what are the expected costs associated with such financing. There might not be certainty—perhaps the planner has not fully decided—or there could be uncertainty about the consequences of the benefit or its financing. But as a normative matter, the fact that financing may still be uncertain in a variety of ways does not fundamentally change the analysis of the wisdom of a policy. The social planner should still compare the expected benefit of the policy to the expected social cost associated with the financing that the social planner would choose. And the social planner can feel reassured that the planner would not choose high-cost financing (from their perspective) if lower cost financing is available.

Consider paid family leave again, and what it looks like to assess financing if there were a consistent social planner. Perhaps there is some uncertainty over how the benefit will be really financed. But the planner knows that they will always choose what they expect to be the best financing. The planner should make realistic plans—which is certainly a challenge¹⁰²—and the planner should understand some of the lessons of this paper. Pairing two policies in a bill still would not mean they necessarily reflect the actual trade-off. But the planner would not need to be concerned that, because of a fiscal commitment they make today, a future social planner would choose what the current planner believes to be a bad source of financing. The first planner would still be in charge, and they would choose whatever source they believe to be the best source of financing.

But let’s assume, instead, a social planner who is more limited in power and particularly a social planner who might control the decision of whether to deliver a benefit but may not be the one to decide the financing of that benefit. (The opposite could also be true and involve a similar set of analysis—a planner who controls financing but not necessarily the benefit derived from it.) What then? The uncertainty of financing takes on a different dimension. If the planner cannot commit to what is ultimately the financing, then the social planner has to consider that it may be someone else—another social planner—that might be the one in control of the financing. The social planner would then have to consider not just the financing that the social planner prefers but the financing that some other social planner would potentially enact to pay for the benefit.

This concern—of optimal policymaking in the case of time inconsistency—has received some attention in other contexts including in the tax law and public finance literature. For instance, a recent vein in the literature on optimal taxation of capital gains has explored the implications for tax reform of inconsistency in policymakers’ preferences over time when it comes to the tax rate—leading some, who support higher capital taxes, to recommend reforms that do more to collect taxes upfront and in ways that are more politically committing.¹⁰³

This brings this same perspective to a different if related issue: what is paying for what. A social planner makes decisions that could lead another social planner with different values to change their approach. And when it comes to fiscal policy, it means that one social planner may enact a benefit, and it may be another social planner who is the one deciding on the financing.

To begin to build out a highly stylized example: let’s say there is a commitment of fiscal resources that a planner (Planner A) believes is modestly good for society on net if matched with the planner’s preferred financing. However, if matched with another social planner’s (Planner B) preferred financing, Planner A believes that it would be bad for society—and the policy should not be pursued. Should Planner A make the fiscal commitment? From the perspective of Planner A, it depends on the probability that it is Planner A’s financing versus Planner B’s financing that ends up paying for the benefit. And Planner A should not pursue the policy—from Planner A’s perspective—if the probabilistic assessment means that the policy is likely to produce a net social loss, despite the fact that there are ways to finance the benefit which would lead to a net social gain.

B. Beginning to Contextualize the Concern

To begin to make this more concrete: let’s take a social planner in the United States who is concerned especially with the welfare of those with lower incomes in society (as I am) and is analyzing whether to enact a policy that generates $110 of benefit for the top quintile in the United States for $100 in resource costs, regardless of whoever finances it.¹⁰⁴ The policy is, thus, net beneficial in the sense that it expands total available resources.

Consistent with the last section, let’s assume this is a social planner of limited power. The social planner is deciding—and has control over—whether to deliver the $110 benefit, but does not have full control over the financing, and it is possible that a social planner with very different priorities would be the one making that decision. Should the social planner pursue enactment of the $110 benefit? The answer is “not necessarily,” despite the fact that there are combinations of benefit and financing in which the program would be socially beneficial.

To further illustrate why, let’s add two additional assumptions. These assumptions are not critical to the basic point I am making here about the implications of an inconsistent social planner when it comes to policies that are marginally beneficial. They simply help illustrate from the perspective of a social planner with a particular set of beliefs.

The first assumption is that the social planner is a utilitarian and believes that there is a declining marginal utility of additional resources as those resources rise—meaning there is greater value for additional resources for those with less than those with more. The utilitarian principle requires that society be organized so as to maximize social welfare as calculated by summing the utility of all members of society.¹⁰⁵ Nonetheless, the utilitarian approach puts priority on resources for those with less than those with more if the utilitarian believes there is a diminishing marginal utility of resources, as empirical evidence suggests there is.¹⁰⁶ The question is then how much marginal utility drops as resources rise—sometimes referred to as the elasticity of marginal utility with respect to income.¹⁰⁷ This is an area of some dispute, to say the least. For these purposes, I adopt an assumption that the U.S. Office of Management and Budget (OMB) under the Biden administration recently suggested as a central estimate of this parameter for purposes of government distributional analysis based on its review of the literature.¹⁰⁸

Importantly, for this illustration to carry weight, the planner could have other values as well as or in place of being a utilitarian—the key really is that the planner prioritizes resources for those with less. And there are many theories that would do so if for different reasons.¹⁰⁹

To give a sense of what this looks like, if a dollar given to the top quintile in the United States is given a weight of one, then a dollar given to the bottom quintile would be given a weight of fourteen times that under this assumption.¹¹⁰ A dollar to the middle quintile would be given a weight of five times that and so on. The same would be true of a dollar taken from any of these relevant income groups. Table 1 illustrates this weighting.

	Average Income (2019)	Distributional Weighting, Top Quintile = 1
Lowest Quintile	$38,900	14
Second Quintile	$54,900	8
Third Quintile	$74,800	5
Fourth Quintile	$104,400	3
Top Quintile	$252,100	1
All	$102,400	N/A

Source: author’s calculations based on Congressional Budget Office data (see footnote 110) and assuming an elasticity of utility with respect to income of 1.4 (see footnote 108).

The second assumption is that a wide range of financing is possible. I use four illustrative financing scenarios, from most to least progressive. The first assumes the top quintile pays for its own benefit. The second replicates the current distribution of federal taxes. That is a progressive system by almost any definition, and the top quintile pays almost seventy percent of the benefit. The third scenario assumes the benefit is financed in proportion to after-tax and transfer income. That is less progressive but still almost half would be paid for by the top quintile. In the final scenario, the financing represents an across-the-board cut to social insurance (programs like Social Security and Medicare) and means-tested benefits (programs like Medicaid and nutrition assistance). This is the most regressive scenario explored—though even more regressive (such as cutting means-tested benefits only) are imaginable and, as discussed in the next Part, have in fact been proposed and recently enacted.¹¹¹

Table 2 summarizes the share of financing that would come from each of these sources.

Share of Financing, %	Top Quintile Only	Federal Tax System	Relative to Income	Cut to Social Insurance + Means Tested
Lowest Quintile	0%	0%	8%	25%
Second Quintile	0%	4%	10%	22%
Middle Quintile	0%	9%	14%	20%
Fourth Quintile	0%	18%	20%	18%
Top Quintile	100%	69%	48%	16%
Top Quintile	100%	100%	100%	100%

Source: author’s calculations based on Congressional Budget Office data (see footnote 111).

So, what is the result here under the various scenarios? Measured in terms of the value of a dollar to the top quintile, social welfare is improved only if the top quintile finances its own benefit. Not surprisingly in that case, society is better off by $10—the top quintile receives $110 and pays $100 in cost. In the three other scenarios, society is worse off. This is measuring in terms of the value of a dollar to the top quintile. And society is worse off by between about $100 and about $600. This is effectively saying that, under these other financing scenarios, it is as if we instead enacted a policy that took between $100 and $600 from the top quintile, and with no matching benefit. Another way to conceive of what this illustrates is that there would need to be that amount of additional benefit for the top quintile for this to even be a break-even policy from a social perspective. For any of the financing scenarios other than “top quintile only,” the policy would need to generate not just $110 of benefit for the $100 in financing costs that are then distributed but between over $200 to over $700 in benefits—a very high bar. Table 3 below summarizes these results.

Change in social Welfare from $110 Benefit to Top Quintile Costing $100 (Measured in Terms of Social Value of Resources to Top Quintile)	Top Quintile Only	Federal Tax System	Relative to Income	Cuts to Social Insurance + Means Tested
Lowest Quintile	$0	-$1	-$104	-$337
Second Quintile	$0	-$34	-$89	-$186
Middle Quintile	$0	-$50	-$79	-$109
Fourth Quintile	$0	-$60	-$68	-$61
Top Quintile	$10	$41	$62	$94
All	$10	-$105	-$277	-$599

Source: author’s calculations.

Thus, if one assigns any significant probability to a financing scenario that does not involve anything but highly concentrated financing at the top for the benefit going to the top quintile, this looks like a losing proposition. Even if there is a ninety percent probability that the financing comes only from the top and a ten percent probability that the financing is consistent with the second scenario—a progressive tax system—it is still a (slightly) negative proposition. If probabilities of any significance were assigned to the other scenarios, it would look even worse.

Despite being highly stylized, the example should begin to raise concerns—and, at least, point to the importance of this exercise. What is the social planner to do? If the planner cannot commit to a specific form of financing in this case, then a policy that might otherwise be socially beneficial may not be. In this example, it is only socially beneficial if the benefit is paid for by the top quintile, and if one is highly confident that would be the case, or alternatively, if the benefit is multiple times the amount needed to finance it.

Importantly, this basic dynamic would be true irrespective of one’s normative framework so long as there is inconsistency between the planner deciding on a benefit and the one then picking the financing. Let’s say a given social planner were principally concerned with efficiency and believes some forms of financing lead to large, socially costly economic distortions while others do not, and that planner was considering whether to enact a benefit. If the benefit only made sense when paired with a specific form of efficient financing, but there was a significant probability of some other, less efficient form of financing being used, then a similar concern would arise. Under that planner’s normative framework, it may not be sensible to pursue the benefit. The social welfare benefit, from the planner’s perspective, may end up being negative.

However, while the basic exercise should be of interest to all irrespective of one’s values in fiscal debates, this author is principally interested in the normative consequences that assume greater priority for those with fewer resources. The question then is what we should expect given the political economy of the United States (or any relevant government), and I explore that further below.

C. What About Current Versus Future Generations?

So far, this stylized analysis has made a critical simplifying assumption. It treats benefits and financing as happening at the same time. However, this may not occur at all. The government’s inter-temporal budget constraint suggests only that a benefit will be financed eventually,¹¹² and the time-period may be no time soon. The lowest quintile may eventually help finance a benefit provided today, but it may be the lowest quintile now, five years from now, ten years from now, one hundred years from now, or more (or some combination of all of that). The math of a government budget constraint says nothing about when. It is another, uncertain element of the causal analysis. Included in the question of ultimate financing is what generation is the one doing the financing. But does the fact that a benefit could be financed now or in the future make the delivery of that benefit more or less attractive?

For purposes of both these stylized examples and this article more generally, I assume that the possibility of future generations paying for the policy does not change the moral calculus. This reflects less conviction and more acknowledgment that there are a complex set of trade-offs between future and current generations, which lead me to conclude—at least for purposes of this article—that it is not clearly better or worse for the lowest quintile of the future to finance a policy versus the lowest quintile today.

This question of intergenerational equity has—not surprisingly—prompted a vast literature and public debate around these trade-offs.¹¹³ To briefly summarize some of the arguments: on the one hand, there is a positive rate of return on resources saved today—meaning future generations could get more consumption in dollar terms out of those resources than the generation today. Another way to see this is in terms of borrowing and interest. A benefit given today comes at the cost of not just that dollar but also the interest charged on the borrowing, reflecting that opportunity cost of capital. On the other hand, future generations are likely to be better off on average than the current generation due to income growth and so would value marginal resources less. Further, people tend to have a preference for getting value today over value tomorrow,¹¹⁴ which could justify further discounting the future (though some argue that taking this preference into account is immoral across generations).¹¹⁵ In addition, there is the chance that future generations simply do not exist due to global disaster in the interim, which can grimly justify discounting their well-being.¹¹⁶

It is not just about how resources should be valued in the abstract across generations but how they should be valued now in light of current policies already in place. For those who believe that we should be borrowing more against that future, it might mean that benefits today for those higher on the income spectrum come with lower risk than they would otherwise. The “high income” of today might be prioritized relative to those “lower income” in future generations who may end up paying. In a sense, the high income of today might be treated as low income across time. However, for those who believe we owe more to the future rather than less, the opposite would be the case—future generations paying rather than today’s generation might raise more in the way of moral concerns.

There is no easy resolution to these trade-offs and the long-standing debate,¹¹⁷ and, for purposes of this article, I assume that the trade-offs are a wash at present. So, the risk of the lower-income Americans financing (or the gain from higher-income Americans financing) is treated the same irrespective of what generation they are in. But, whatever one’s view, the exercise of imagining the actual financing still matters. And, even if one believes that the future should be deprioritized for resources relative to today, it remains possible that a benefit given to someone higher on the income spectrum today would be paid for by people who, under almost any moral framework, should receive greater priority—someone in the future who still has less in the way of resources however measured.

IV. Lower-Income Americans at Risk

This part will begin to try to bound the uncertainty of financing, specifically in the context of U.S. federal fiscal policy. The focus of this part will be on distribution and on the risk that a benefit might be financed by those with less in the way of resources. If that risk is very low, then the normative concern raised in the previous part should fall away. We do not need to be concerned that a future policymaker would choose to finance a new benefit with a high-cost form of financing (from the perspective of this author). If the opposite is the case, then there might be a higher bar for enactment of any new benefit.

Unfortunately, looking at the political economy of the United States does not assuage this concern but rather the opposite—there is a risk that lower-income Americans finance a new benefit and one that is not for them. This is for several reasons. First, there have been long-standing and now legislatively successful efforts to enact financing policies that do in fact burden those with less in the way of resources. This is effectively the fiscal position of key conservative policy leaders.¹¹⁸ Second, while more progressive policymakers tend to publicly support financing government from taxes concentrated on the very top of the income spectrum, that is unlikely to prove to be a sufficient financing source—which further raises concerns about who would actually finance a benefit in the end.¹¹⁹ Third, while there are ways for policymakers to enact financing that might better commit to a specific financing source, and ones that do not result in those with less paying for benefits for those with more, they are forms of financing generally rejected by U.S. political leaders in recent years, though not always.¹²⁰ This Part considers each of these issues in turn.

A. Financing from Lower-Income Americans

In the United States, low-income Americans stand at risk of effectively paying for government commitments. On the spending side of the ledger, social insurance and means-tested programs are contested and at risk of being slashed as a way to help finance the government. On the tax side of the ledger, conservative leaders have recently shifted from a “no new taxes” approach to supporting relatively regressive financing in the form of broad-based tariffs, even as they seek to significantly reduce revenues on net.

Starting with social insurance and means-tested benefits: these are both fiscally significant and critical sources of support for middle- to lower-income families. The Congressional Budget Office data suggest that families in the lowest quintile alone received approaching $600 billion in 2019—or about three percent of GDP—in federal social insurance and means-tested transfers.¹²¹ That amounted to over fifty percent of the total resources for those families.¹²² This is a combination of social insurance like Social Security, Medicare, and Unemployment Insurance (over 15% of their resources), health coverage from Medicaid and the Children’s Health Insurance Program (over 25% of their resources), and other means-tested programs like the Supplemental Nutrition Assistance Program (over 10% of their resources).¹²³ For the bottom three quintiles, the number was approximately $1.5 trillion, or just under 10% of GDP, and about 35% of total resources for these families.¹²⁴ Table 4 shows the value of social insurance and means-tested benefits both in dollar value and as a share of family resources as of 2019 by quintile.

Social Insurance and Means-Tested Benefits in the United States in 2019	Dollar Value (Billions of $)	Share of Household Income
Lowest Quintile	$569	56%
Second Quintile	$510	36%
Third Quintile	$461	24%
Fourth Quintile	$414	16%
Top Quintile	$359	6%
All	$2,313	17%

Source: author’s calculations based on CBO data (see footnote 121).

These benefits have been and are contested.¹²⁵ This paper will not describe in detail the evolving fiscal positions of the parties in the United States, which will surely continue as fiscal issues play a prominent role in current legislative debates. What I mean to simply illustrate is that there is a real risk that a new benefit could in fact be, in part, financed by those across the income distribution including at the bottom.

Let’s take the legislation just enacted by the current Trump administration and Republican Congress, or OBBBA.¹²⁶ In a major legislative push, President Trump and the Republican Congress extended the tax cuts first enacted in 2017, expanded some of those tax cuts, and, critically, also enacted major cuts to the social safety net, among other reductions.¹²⁷ While the legislation would still add over $3 trillion to deficits through 2034, it included over $1 trillion in cuts to health programs, mostly Medicaid, and almost $200 billion in cuts to nutrition assistance, among other areas.¹²⁸ Due to these cuts to the social safety net, the Congressional Budget Office estimates that the new law will result in a reduction of over three percent in resources available to the bottom decile, even as the law significantly increases the resources available to those higher on the income spectrum.¹²⁹ And, while the law was uniquely regressive in its effects,¹³⁰ the cuts to the social safety net in the bill reflect long-held beliefs within economically conservative circles about the wisdom of the social safety net.¹³¹

The point here is not that such financing is inevitable in the United States. Notably, the first Trump administration tried and failed to enact cuts to the social safety net. That administration had made an all-out legislative effort to repeal many of the Affordable Care Act’s health provisions including the Medicaid expansion.¹³² But the administration could not get Congress to agree—though they famously came within one vote of moving forward a limited repeal in the Senate and the defeat might even have depended on the vote being held later in the day rather than earlier.¹³³ And that effort reflected the fact that only several years earlier the safety net had been significantly expanded with the passage of the Affordable Care Act itself under President Obama.¹³⁴ So, the outcomes of debates with regard to the social safety net are uncertain, and it is a question of risk, and how much an earlier commitment of resources increases the chance of such financing being enacted. We have seen a version of that risk materialize in the form of the just-enacted fiscal legislation using the safety net as a financing mechanism, and, looking ahead, such risk unfortunately remains.

It is not just on the spending side of the ledger where lower-income Americans now face risk. In a way that is unique to the era of President Trump but may not be isolated to his term, the Republican Party has embraced one form of taxation: broad-based tariffs. Whereas in recent decades, Republican leaders have for the most part rejected any new taxes,¹³⁵ that is not the case with President Trump, who implemented a historic set of tariffs through executive action, driving the tariff rate to the highest level since the 1930s.¹³⁶ President Trump and those in his administration have sold tariffs not just as a way of affecting trade flows but also as a source of revenue in place of the progressive income tax.¹³⁷ Full replacement of the income tax is not even mathematically possible,¹³⁸ but expansion of tariffs as a source of revenue is. The new tariffs President Trump has put in place are expected to raise on the order of $200 billion per year on average over the next decade if they remain as of September 2025.¹³⁹ And while President Trump has suggested that other countries will pay those tariffs, the vast bulk of the evidence suggests that tariffs would in fact be paid largely by Americans up and down the income spectrum, and that the burden would fall disproportionately as a share of income on those at the bottom of that distribution.¹⁴⁰ For instance, the Budget Lab at Yale finds that the new Trump administration tariffs will reduce the incomes of those in the first and second deciles of income by roughly 2% to 3% and the top decile by about 1%.¹⁴¹

The point is not that these relatively regressive financing sources—that affect lower income Americans—are inevitable sources of financing for a new government benefit. There is opposition to measures such as these. And even if they are implemented, it is possible they get undone. And, on the other hand, it is possible that some of this financing might—unfortunately—get enacted anyway. Still, all of this demonstrates that there is some significant risk that such measures affecting low-income Americans would be the marginal financing source, unless policymakers can credibly commit otherwise and in a way that is constraining on future policymakers. To go back to the examples in the last section of a benefit for the top quintile: given the risks of financing coming from lower-income Americans and absent some strong and credible commitment otherwise, the policy begins to look like a net loser for someone prioritizing the well-being of those with less.

B. Why Pledges to Finance Only at the Very Top Are Not Credible

That should be enough to show that there is some significant probability that lower-income Americans might finance new benefits. There are policymakers who regularly take power and who believe that to be the best source of financing. However, there is an additional, if somewhat less important, reason to believe this. While progressive policymakers so far have largely opposed efforts in this regard, they have yet to credibly identify policies that, as a mathematical matter, could even finance the existing government, let alone new benefits.

Democrats in Congress and Democrats in the White House have recently been only willing to support financing coming from the very top of the income distribution, around the top two to three percent of households, and large corporations.¹⁴² Both Presidents Obama and Biden promised not to increase taxes at all below high-income thresholds, sharply constraining the kinds of revenues they proposed both to reduce the deficit and to finance new initiatives.¹⁴³ There is a political challenge to enacting those measures, meaning they may not end up being the financing. Republicans, of course, oppose these measures and have proposed their own sources of financing.¹⁴⁴ But there is an additional challenge to that: math. There is likely not enough revenue available at the very top of the income distribution plus among large corporations to do much beyond financing our existing government commitments. There is simply not enough to finance our existing government commitments and also finance new commitments—including, for instance, reversing the recent large safety net cuts enacted under President Trump. To be clear, there is substantial revenue to be had from those at the top, and important reforms to be done, as this author has detailed elsewhere.¹⁴⁵ But it will likely not be enough even if some of the more aggressive measures end up being adopted.

How much can be raised from the top two to three percent of households plus large corporations? I have elsewhere done an illustrative accounting based on proposals made during the Biden administration plus repeal of the tax cuts in OBBBA for those making over $400,000. Roughly speaking, I estimate that this forward-leaning set of proposals would raise on the order of $760 billion per year, or 1.9% of GDP, from 2031–35.¹⁴⁶ And, given realistic political limits in Congress even if there were a progressive governing majority and legal risks at the Supreme Court with some of these proposals, I conclude that this exercise suggests a rough limit of an additional one to two percent of GDP in revenue from this top two percent to three percent of households plus large corporations.¹⁴⁷ That is a significant number, and achieving reforms like this would be a major step forward. However, this additional revenue has to be put in the context of the government’s pre-existing commitments and additional priorities.

The United States is on a likely unsustainable fiscal trajectory. Based on the latest projections from the Congressional Budget Office updated for passage of OBBBA plus imposition of tariffs, debt held by the public will rise to 120% of GDP by 2035 from the current level of about 100%, exceeding the highest level on record.¹⁴⁸ Stabilizing the debt as a share of the economy would require an adjustment to taxes or spending or both on the order of $700 billion per year from 2031–35, or 1.7% of GDP.¹⁴⁹ That alone equates to roughly all (or more) of what even a forward leaning revenue package focused on the top two to three percent of households plus large corporations would raise. Further, that means there wouldn’t be revenue available from these sources to finance additional priorities, including reversing the very regressive financing recently enacted or imposed under President Trump. Reversing the social safety net and other cuts in OBBBA would cost on the order of $290 billion per year from 2031–35, or 0.7% of GDP. Reversing the regressive and ill-advised new tariff regime (to the extent they remain in place) would then roughly double that number—adding over an additional $250 billion, or another 0.7% of GDP, in cost.¹⁵⁰ The bottom line is that there is probably not enough revenue available from the top two to three percent plus large corporations to both cover pre-existing commitments and any new significant priorities.

To be sure, the Biden budget does not represent all that is possible in revenue-raising at the top of the income spectrum, but it is a menu featuring very significant changes to the code. It is the math of such menus that has led a number of outside commentators, like Paul Krugman and others, to arrive at this same conclusion—that taxes would need to rise on those beyond the very top of the income spectrum to finance the existing government and any additional major fiscal priorities.¹⁵¹

This is another reason to be concerned that financing could come from those with lower incomes. Even those leaders in the Democratic Party who tend to oppose such financing have not supported a credible set of proposals to fully finance the existing government and any new significant priorities. Neither have Republicans, to be clear, even with their proposals to cut back elements of the social safety net and tariffs. But this dynamic further raises the risk of burdens falling on those with lower incomes. While it is much more plausible to raise sufficient revenues from those toward the top—if policymakers are willing to raise taxes beyond the top two to three percent—to finance the existing government and commitments beyond,¹⁵² no leading policymakers have been willing to propose that in recent decades. And even when it comes to the progressive policy leaders of the future, what will they choose: raising taxes on the ninetieth percentile to finance a new commitment or, alternatively, cutting back benefits for those most vulnerable? I know what I hope, but I do not know with confidence what they would do. After all, if anything, the one thing that both parties have agreed on in recent years is cutting taxes for the somewhat well to do.¹⁵³ That could reverse, but that’s what it would be—a reversal.

***

Where does this leave us? The bounding exercise thus far further raises concern about the normative implications of new benefits beyond those going to those worst off—and absent policymakers being able to credibly commit to the financing that will pay for it. Going back over the financing possibilities laid out in Part II.B., even replicating the current burden of federal taxes seems too optimistic as a central estimate in light of Republicans’ desire for more regressive policies and Democrats’ unwillingness to raise taxes below the very top and large corporations so far. And it is certainly too optimistic to assume that a new benefit would be financed at the very top alone.

C. The Sliding Scale of Commitment

However, as suggested earlier, policymakers do have some ability to affect what actually finances a new policy as a causal matter. The risks laid out in the prior section of a new benefit potentially being financed by those most vulnerable can be mitigated at least to some degree. It is best understood as a sliding scale of commitment, and this section begins to try to differentiate points on that scale.

One pole on that scale is no commitment at all on financing. That is when policymakers entirely defer that decision by relying on changes to debt projections. In that case, the bounding exercise I engage in above becomes most relevant. Policymakers in that case have explicitly chosen a policy of no pre-commitment beyond their rhetorical positioning on the financing of the government generally.

But policymakers sometimes do identify financing, and they can do so in ways with greater or lesser commitment, as I have discussed.¹⁵⁴ The opposite pole to no commitment is what I have called “custom financing”—financing that is unique to the benefit. The middle of the spectrum is “fixed menu” financing—financing that had a greater chance of being used elsewhere.

How to differentiate “custom financing” from “fixed menu” financing? The basic differentiating factor is that, without the benefit involved, policymakers would be very unlikely to consider that financing source. Through the economic and political narrative, the policies are then causally linked. Financing that only gets proposed in connection with the specific benefit involved would more likely be “custom.” Financing that appears with any regularity in a variety of different proposals would come from the “fixed menu.” As a very general matter, “custom financing” will tend to link those who are paying and those who are benefiting—this is what generates the political logic linking the financing and benefit.

This distinction is sufficiently important that I will pause here to give examples in two key policy areas, social insurance and tax reform, and try to situate those examples as either closer to “fixed menu” or, alternatively, “custom financing.”

1. Social Insurance

Several social insurance programs in the United States have been paid for in whole or in part with broad-based taxes tied partially to the benefit—and which seem more likely to be “custom financing” for the program.¹⁵⁵ For instance, Social Security is financed with payroll taxes imposed both on workers and their employers,¹⁵⁶ although actual economic incidence is likely mostly on the employees.¹⁵⁷ On the whole, the program itself engages in significant progressive redistribution in addition to providing insurance against certain events such as disability, early death of a worker, or the resource strain of living to an old age in retirement—since those with lower incomes get more in benefits per dollar paid in.¹⁵⁸ But there is a connection between payroll records and taxes paid and the benefits received, even if that connection is not one-for-one. This then creates a political logic linking the benefit and the tax. President Roosevelt famously reflected on the use of the payroll tax as being strongly tied to the benefit as a political matter. In a private conversation with staff in which he defended the use of payroll taxes, he is reported to have said:

We put those pay roll contributions there so as to give the contributors a legal, moral, and political right to collect their pensions and their unemployment benefits. With those taxes in there, no damn politician can ever scrap my social security program. Those taxes aren’t a matter of economics, they’re straight politics.¹⁵⁹

Even with President Roosevelt’s explicit thoughts on the strength of the linkage, the causal link here is not certain—it is possible that at least some of these payroll taxes may have been imposed irrespective of the social insurance benefit and to pay for other, alternative benefits, or that the benefits might have been enacted and sustained with an alternative financing source. But there is a significantly higher probability of a causal link than is the case with other forms of financing. The key differentiating factor that Roosevelt himself identifies is a sense that the financing is, as a “legal, moral, and political” matter (in his words)¹⁶⁰ justified by the benefit and paying for the benefit.

Other social insurance programs are financed from the general fund and more in a “fixed menu” fashion. Medicaid for instance—the health insurance program for low-income Americans—is financed wholly through the general fund,¹⁶¹ and expansions, such as in the Affordable Care Act, were at least partially financed with revenues unrelated to the program and health care entirely—such as increased taxes on the wage and investment income of high-income Americans.¹⁶² The Biden administration’s efforts to expand social insurance and safety net programs such as the paid leave program in Build Back Better were also financed in “fixed menu” fashion, relying on a list of largely unrelated taxes on high-income Americans and large corporations including an expansion of that same net investment income tax that had been enacted in the Affordable Care Act.¹⁶³ In these cases, there is less commitment from policymakers to identify financing that is tightly linked to the benefit. Some weight should be given to what policymakers package in the legislation—the financing may not otherwise have been enacted, at least not anytime soon. But it should be given less weight.

This brings me back to the example given at the start of this article: paid leave.¹⁶⁴ The financing proposed in the previous congressional proposals for paid leave—a payroll tax—is, by political logic, directly connected to the benefit, much the same as Social Security is. Workers are given a similar deal: if you pay into a system, you earn a benefit (if there is a circumstance making you eligible for family or medical leave). The system is progressive, giving more to those with less on average, but, nonetheless, there is that strong political connection. Leading policymakers have not been otherwise proposing new payroll taxes to cover general government services or other non-work-related benefits (outside of Social Security itself); it is limited to this context.¹⁶⁵ So, while it is possible that this financing could have been used for other purposes, it is less likely. By contrast, the Biden administration’s approach of using a combination of high-income revenue and corporate revenue raisers¹⁶⁶ is less committing—this is “fixed menu” financing. And the trade-off is less likely to be that financing and more likely to be something else as decided by a future policymaker. (Notably, in its most recent iteration, the sponsors of the paid leave act in Congress dropped any specified financing source—a break from their previous versions. This, of course, represents a shift away from any commitment on financing.¹⁶⁷)

To make the trade-off clearer, compare the distribution of a payroll tax, like that proposed by congressional Democrats in the past, to a corporate tax increase of the kind that President Biden had proposed to finance paid leave (along with other high-income revenue raisers). As noted at the start and on its face, the payroll tax is clearly more regressive. The very nature of the political link to the benefit comes with a downside, at least from the perspective of this author: everyone with payroll pays. We have instruments that are considerably more targeted at the top. Nonetheless, as shown in Table 5, a considerable portion of the total burden comes from those with higher incomes since they have so much of the payroll.¹⁶⁸ Still, the corporate income tax is likely considerably more progressive. One set of estimates suggest that the bottom two quintiles pay twelve percent of the burden under the payroll tax,¹⁶⁹ but only six percent of the burden under the corporate income tax.¹⁷⁰

However, neither might be the ultimate financing for the benefit, and that risk is considerably greater for the corporate income tax since it is fixed menu financing—not custom financing like the payroll tax—and further, corporate and high-income tax increases alone are unlikely to be sufficient to finance the existing government and new commitments.¹⁷¹ This means other sources, some highly regressive, are more likely to be the sources. Compare the distribution of a cut to social insurance and means-tested programs. Almost fifty percent of the burden would come from the bottom two quintiles alone,¹⁷² and there are even more regressive sources possible, like cuts largely to means-tested programs (like what happened in the recently enacted fiscal legislation under President Trump¹⁷³). It does not require a much higher chance of such a regressive source being the actual financing for the payroll tax to ultimately result in more progressive outcomes than the corporate income tax. All it takes is a significant reduction in the probability of very regressive financing sources like these being used, and while there is no way to fully prove that out, that is a plausible description of what might occur given the U.S. political economy. Counterintuitively, in this context, the payroll tax could generate more progressive outcomes, and ones that at least reduce the probability of some of the most regressive sources of financing.

Payroll Tax Versus Other Financing Sources (Share of Burden)	Payroll Tax	Corporate Income Tax Increase	Cut to Social Insurance + Means Tested
Lowest Quintile	4%	2%	25%
Second Quintile	8%	4%	22%
Middle Quintile	14%	9%	20%
Fourth Quintile	22%	15%	18%
Top Quintile	52%	70%	16%
Note: Top 1%	11%	35%	1%

Source: see footnotes 168, 170, 172.

2. Tax Reform

Tax packages can, of course, involve a combination of tax cutting and tax raising, and, in the context of tax reform, the same question arises of what is ultimately paying for what and the degree to which that is contained within the four corners of the legislation. As with social insurance, a key factor is any linkage between benefits and financing. Returning to TCJA, much of which had been set to expire after 2025 but was permanently extended in the recent fiscal legislation¹⁷⁴: the initial law, as noted earlier, was in significant part deficit-financed, costing $1.5 trillion over a decade initially.¹⁷⁵ That reflected policymakers choosing not to commit to the eventual financing for that significant portion. But, it also combined $5 trillion in gross tax cutting and $3.5 trillion in gross tax raising, partially financing the legislation on paper.¹⁷⁶ However, some of these financing provisions were much more likely to be causally connected to the tax cuts than others—and again, it in part comes back to the political link between benefits and those provisions serving as financing.

Contrast the two largest financing provisions in the 2017 law: the elimination of personal exemptions and the limitation on itemized deductions including the mortgage interest deduction and the cap on the deductibility of state and local taxes.¹⁷⁷ The first was very likely causally connected to a number of the tax cuts in the bill, and custom-made for the benefits for which it is paying. The second—the limitation on itemized deductions—was less likely to be causally connected and might have been enacted in some similar form anyway in the future.

Starting with the personal exemption, it allowed a write-off of a specific amount from income per person—scheduled to be $4,150 per person in 2018.¹⁷⁸ The 2017 law repealed those personal exemptions and raised $150 billion per year initially as a result of the repeal of the provision.¹⁷⁹ For a family in the lowest income bracket of ten percent, this raised their taxes by $415 per person.¹⁸⁰ This financing, though, is causally connected to two other reforms giving tax cuts. The 2017 law increased the standard deduction which also allows a family to take a write-off from their income, but, unlike a personal exemption, this requires a family to choose not to itemize deductions (part of the policy goal was to reduce the number of itemizers). For a married couple, the increase was $11,000,¹⁸¹ more than offsetting the loss of $8,300 in personal exemptions for the two adults. When it came to children, the 2017 law increased the Child Tax Credit by $1,000,¹⁸² and this too offset the loss in personal exemptions for dependent children. Altogether, these three provisions were expected to lose a small amount of net revenue, less than $10 billion per year.¹⁸³ And, they were intimately linked in the reform in terms of who saw the personal exemptions withdrawn and who received offsetting benefits.

The basic economic and political logic of the reform was to simplify the tax system by reducing the number of those who itemize their deductions by swapping personal exemptions (which can be taken in combination with itemized deductions) with a larger standard deduction (which cannot be taken in combination). Thus, the increase in the standard deduction and Child Tax Credit was truly a swap—with those benefiting broadly paying for that benefit—with the repeal of the personal exemptions, and a repeal of those exemptions would have been highly unlikely to happen otherwise. It was custom financing, and the provisions were rightly considered as a package in the causal sense.

The limits imposed on itemized deductions (for those who continue to itemize) operated very differently than this. There were several new limits in effect. The most important was a cap on deducting state and local taxes.¹⁸⁴ There were also other consequential limits including a significant cut back in the deductibility of mortgage interest.¹⁸⁵ According to the official estimates, these limitations raised on the order of $80 billion per year when initially in effect from 2018 to 2025,¹⁸⁶ and they did so disproportionately from the very top of the income spectrum.¹⁸⁷ The 2017 law overall still offered regressive and deficit-financed tax cuts according to the traditional distributional estimates,¹⁸⁸ but this financing was assumed to partially offset that.

However, unlike the repeal of personal exemptions, a reduction in the value of itemized deductions was and is imaginable in other contexts. President Obama, for instance, had proposed limiting the value of itemized deductions and other tax benefits to finance some combination of new benefits and deficit reduction,¹⁸⁹ though that proposal was not adopted. So, reducing the value of itemized deductions—coming more from the “fixed menu” model of offsets—could well eventually have been used for other purposes, whether it be to finance current commitments or to pay for new ones that were more progressive than what was in the 2017 law (and now made permanent). It is by no means certain that it would have been adopted by now or in the future. These caps are controversial, especially the limit on the deductibility of state and local taxes,¹⁹⁰ but there was no particular political logic for their use in the 2017 law versus paying for other priorities. In this key sense, the 2017 law’s provisions may be more regressive than traditional distributional analysis might suggest. The law’s regressive benefits may not have really been partially paid for by the highly progressive limits on itemized deductions—at least to the degree reflected in traditional distributional analysis—but, instead, other sources that are considerably more regressive. As discussed in the prior two parts, there is real risk it is those with less who end up footing the bill as a causal matter.

****

In sum, recent political economy of the United States suggests that, absent policymakers being able to commit otherwise, there is risk that new benefits come at the expense of the most vulnerable and not from highly progressive sources. Policymakers do have ways to alleviate this risk, with the risk lowered to the degree policymakers identify any financing that does not burden lower-income Americans and the risk further lowered if the financing is more custom to the benefit involved and could not be used for other purposes. However, U.S. policymakers have—in recent decades—sharply constrained the kinds of financing they have been willing to enact. This has made it harder to pair benefits and custom financing, and this then brings us to the final part, using these reflections to inform the ongoing debate about how universal versus targeted government benefits should be.

V. A New Perspective on Universalism Versus Targeting

This part turns to the long-standing debate about universalism versus targeting in government benefits,¹⁹¹ and seen through the lens of the kind of distributional analysis this article advocates. The approach to universalism versus targeting represents something of its own divide between the Biden and Obama eras in designing expanded federal benefits, leaving an open question on what progressive governance might look like in the years ahead. The Biden administration and perhaps especially the Congress with which it worked in the first two years had ambitions for more universal programs than did the Obama administration and its governing coalition in its first two years.¹⁹² This perhaps reflected in part a shift in thinking in progressive circles toward support for more universal over more targeted benefits,¹⁹³ even as the debate on this topic has been longstanding.

The basic question is whether it is better for government benefits to go broadly to the population—like Social Security and Medicare do—or be more targeted by income, like means-tested benefits such as Medicaid and nutrition assistance. Those engaging in this debate tend to share the normative perspective offered in this paper, with the assumption being that to the degree that more universal benefits come at the expense of those with fewer resources, they would be a bad idea or at least that would be a major detriment.¹⁹⁴ So, the disagreement tends to be empirical: whether or not that is what would result (or if universal benefits could even lead to more resources going to those with less).

This Part offers one new and important observation: that one key factor in this debate is the degree to which policymakers can and do commit that those lower on the income spectrum do not end up paying for benefits for those higher on the income spectrum. When appropriately reimagining fiscal policy and who actually pays for what ultimately, this issue becomes critical to the exercise of assessing whether it is best to pursue a more targeted versus more universal benefit.

A. How Universal Should Benefits Be?

The debate over universalism versus targeting tends to raise several issues, and this article will not delve deeply into these—but rather focus on this issue of distribution and, in particular, financing. This is not to say the issue of financing alone should be definitive, even if it is important. To name a few other of the key issues often raised: the debate often focuses on whether political support is more stable for benefits that are universal.¹⁹⁵ There is also the issue of the administrative challenge of targeting versus the ease of universal benefits, both when it comes to costs for the government bureaucracy and the costs for the population eligible to participate.¹⁹⁶ And there is the issue of the potential stigma of targeted benefits and social cohesion that may come with more universal ones.¹⁹⁷ These issues will not be addressed here, and are certainly important ones.

This section focuses on who pays for a universal benefit. This too—not surprisingly—has been raised in the context of this debate. A universal benefit by its very nature will give more benefits to those with higher incomes than a targeted benefit. Thus, holding all else equal, greater universality would tend to suggest less redistribution to the bottom of the income spectrum than if the same amount of resources were used with greater targeting. However, it may not be right to hold “all else equal.” A more universal benefit could be matched with taxation that targets the total redistribution society might want. So long as the financing is adjustable—and in a way to which policymakers can commit—then universality does not necessarily come at the expense of greater benefits for those most vulnerable and could in fact help the most vulnerable, for instance, by sustaining political support, making the system simpler to navigate, or other reasons.¹⁹⁸ The question is whether policymakers will in fact adjust financing in this way.

However—and to the main point of this article—that is uncertain. The benefit for those higher on the income spectrum might be effectively paid for by them with higher taxes, or it might not be. It could be paid for by those with lower incomes. The risk of that depends on the degree to which policymakers commit to financing that is custom to the benefit involved—and so that those with higher incomes are likely to pay for their benefit and not those who are more vulnerable. Putting this in terms of the inconsistent social planner: those who engage in these debates may at times be in the position of a social planner—or at least able to influence those who are in that position—but they must also take into consideration the reaction of other social planners with very different values who may at times control the process.

This is less a concern for targeted benefits, assuming that one believes additional redistribution is desirable. In that case, a lack of commitment to the financing source does not necessarily change whether the policy should be pursued. Take a new, targeted program that helps lift more Americans out of poverty. And assume no commitment to a specific form of financing—there is a risk that it is paid for by those with low incomes and also a possibility it is paid for by those with higher incomes. At worst, those with the lowest incomes are effectively left no worse off than they were before if they end up paying for the policy; at best, they are better off. This might be complicated by winners and losers within the income class, but the broader point remains—the downside risk is limited. This is unlike a benefit going to someone higher on the income spectrum. Those with low incomes do not directly benefit from that but could end up paying for it.

In sum, a key and under-appreciated issue in the universalism versus targeting debate is the degree to which policymakers can commit to offsetting the benefits in universal programs flowing to those with higher incomes through taxes paid by them in the causal sense. As the prior part laid out, it is in fact possible for policymakers to do so. They have done so in the past—with the financing of Social Security being a quintessential example of that. But the fiscal debates of recent decades have made this much harder—in turn making universal benefits riskier to pursue absent a changed approach to financing new programs.

B. U.S. Fiscal Policy: The Challenge of Universal Benefits and Constrained Financing

As noted above, U.S. tax policy in the last several decades has been sharply constrained in terms of the revenue sources that policymakers have been willing to use. For the most part, even where policymakers have been willing to increase revenues, they have only been willing to do so for the very top of the income spectrum in the case of Democratic lawmakers or, most recently when it comes to the Republican party, in the form of tariffs with the burden as a share of income highest on those with less.¹⁹⁹

This article does not fully lay out the trade-offs when it comes to policymakers limiting revenue increases to these sources. I mean only to add an additional and important implication: so long as this political constraint holds, broad rather than targeted benefits will tend to come at greater risk distributionally. By this, I mean that there is greater risk that those with lower incomes will end up effectively financing benefits for those who are better off—because policymakers will be unable to offer financing that is truly custom and represents a political commitment of what is paying for what. Such custom financing will tend to violate the tax pledges made by both Presidents Obama and Biden given the link between benefit and financing.²⁰⁰ On net, the middle class and lower-income Americans might still be left better off by the combination of benefits and taxes, but the tax pledges focus on gross taxes paid, not netting against any benefits received, and abiding by them rules out any broad form of financing.

In the Obama administration, the distributional risks were largely avoided by a focus on targeted benefits. The Affordable Care Act represented the single largest expansion in social insurance and the safety net in decades,²⁰¹ and it was a targeted expansion, combining an expansion in Medicaid for those with the lowest incomes and tax credits for those with higher incomes but phasing out as incomes rose.²⁰² It was, in part, financed in the “fixed menu” model as described above—such as using the net investment income tax to pay for it.²⁰³ However, there is not the same distributional conundrum when it comes to targeted benefits like this. Fiscal benefits are flowing to a group that should receive priority (at least in many normative frameworks), and, while there is risk that the same prioritized group could end up helping to finance the benefit or that some within the group end up as winners and others as losers, they are still likely, as a group, to be left better off overall. In the end, in terms of the tax and health care policy that the administration enacted across its eight years, net benefits did not go beyond the sixtieth percentile of the income distribution, and eighty percent of the benefits went to those in the thirtieth income percentile or below.²⁰⁴ Even if we are not certain on how those benefits are financed, the risk of those with the least resources being left worse off is low because of the targeting.

Under the Biden administration, and especially the Democratic Congress at the start of the administration, there was a push for benefits with greater universality as a general matter—though much of this agenda failed to pass or was blocked by the courts. In addition to paid leave for which there was a real legislative push, there was a proposed childcare benefit, capping costs of childcare expenses at no more than seven percent of income for the vast majority of families, giving substantial benefits to families toward the top of the income spectrum.²⁰⁵ That, too, eventually fell out of the enacted legislation.²⁰⁶ The Administration, through executive action, tried to broadly reduce student debt burdens. While there was some targeting that increased relief for those coming from families with the lowest incomes, and relief extended only to those with an annual income of $250,000 or less for a married couple, that still extended benefits up to over the ninetieth percentile of income.²⁰⁷ In the end, this effort was stopped in court.²⁰⁸ To be sure, there were targeted benefits as well,²⁰⁹ but the push toward more universal programs reflected a broader intellectual agenda in certain progressive circles.

Of course, the Obama and Biden administrations are now in the history books. But the question on approach is not—future progressive administrations will face the question of whether to pursue more targeted or more universal benefits.

The challenge is that the push for greater universality has not so far been matched, for the most part, with a push for broader financing. The tax pledges have continued to bind and constrain progressive policymakers. Custom financing of a kind that is truly linked to broader benefits has been largely dismissed under the last two Democratic presidents. The financing for the most part is “fixed menu” financing combining revenue raisers focused on large corporations and high-income Americans. Many of these are good policies and should be pursued, but policymakers cannot credibly say that these are—with very high probability—the actual financing sources of a new benefit as a causal matter. It is more committing than simply debt financing, but it is less so than broad-based measures better linked to the benefits.

In some sense, the point here is intuitive and aligns with a number of arguments that have been made in the past: broader benefits will tend to require broader taxes—or other financing—to pay for them.²¹⁰ However, this article is also making an additional point, and one that is largely unstated in the universality versus targeting debates of recent years. Absent policymakers being able to credibly commit otherwise, benefits that are more universal come with the risk that those more vulnerable ultimately pay for benefits for those better off. It is taking a bet on the future resolution of fiscal fights. There are scenarios in which the resolution is more progressive taxes; there are resolutions in which it means cuts to the safety net; there are resolutions in which it involves higher tariffs (an outcome that would have seemed low probability even a few years ago). We do not know. This is a conundrum that policymakers could partially solve by embracing financing that is more unique to the universal benefit. They have in the past, and there are versions of proposals today that feature relatively strong commitment. Senator Gillibrand and thirty-six other senators in the last Congress continued to co-sponsor her paid leave legislation with a payroll tax as financing (even as the current version of the legislation does not specify a financing source).²¹¹ But, at the presidential level and when it has mattered most, there has been a sharp constraint.²¹²

There are two important implications then when it comes to universality versus targeting. First, if the financing constraints of recent years remain binding, then this puts a thumb on the scale toward more targeted benefits over ones that are more universal given policymakers’ inability to commit to custom forms of financing. Second, if greater universality in new benefits continues to be a goal, then this suggests even more importance in changing the pledges that recent Democratic presidents and others have made to allow for the kind of financing that reduces the chances of the benefits coming at the expense of those worse off. It is not just about building a revenue base to finance these benefits and resolving concerns about a long term and unsustainable rise in federal debt—though there is that. It is also about resolving the uncertainty about how they will be financed and not leaving that to future fiscal fights.

Progressives (including this one) who support these benefits should understand that they are in fact social planners of limited power (as is anyone else involved in these fiscal debates). There might be moments of some influence on the process—where a confluence of electoral outcomes puts more progressive policymakers in power. And, at that point, the question arises concerning what to do. The answer should not depend on just how those progressives would prefer to finance a benefit they may try to enact, but instead, how the benefit might ultimately be financed—given the politics of the country and the times at which policymakers of very different political persuasions may be in power. From this perspective, a policy that may seem beneficial may not be—because of the danger of the financing. Failing to engage in the imaginative exercise is simply burying one’s head in the sand. And while their goals may not align with the normative preferences of this author, the same, of course, is true of conservatives—to best achieve their own goals, they should imagine how a different social planner may react to their decisions and in part control the financing of it.

VI. Conclusion

This article seeks to push policymakers, academics, analysts, and others to truly wrestle with question of what ultimately pays for what in fiscal policy. To reimagine distribution. Whatever one’s view of the appropriate role of fiscal policy, this form of analysis should be central to the endeavor. I have played out the implications based on the normative framework that I hold, including giving further reason to target new benefits given the U.S. political economy. But evaluating the ultimate financing for a fiscal policy is relevant whatever one’s views.

To be sure, the reimagination of fiscal policy will provide anything but certainty—this is instead a call for fully engaging what is the inherently uncertain process of enacting a fiscal policy whose financing will not ever be fully known and will in part be subject to the preferences of other policymakers and planners. The uncertainty and inconsistency should not be seen as excuse to not engage in this exercise. Ignoring the challenge does not make it go away. Failing to develop a theory of what ultimately pays for what risks enacting fiscal policies that are only worthwhile based on one’s preferred financing rather than the financing that would actually occur. One should focus on the financing that may ultimately be. That is what matters for assessing the effects of a fiscal policy—what really has paid for what—and whether that policy has improved the country.

[*] Charles L. Denison Professor of Law, New York University School of Law. I am grateful for feedback received from participants in the NYU Law Summer Workshop, the Tax Law and Policy Series at the University of British Columbia Allard School of Law, the 2024 UVA Fall Invitational Tax Conference, the 2025 Tax Policy Conference at Columbia Law School, and the National Tax Association Annual Meeting. Particular thanks to Ryan Bubb, Wei Cui, Daniel Hemel, Susie Morse, Alex Raskolnikov, Natasha Sarin, and Daniel Shaviro. Thanks also to the outstanding work of the editors at the Harvard Journal on Legislation.

View PDF Version