No New Tax Cuts? Examining the Rescue Plan's New State Tax Limits

The American Rescue Plan Act (ARPA), enacted in March 2021, allocated $195 billion to state governments to help them respond to the COVID-19 pandemic, with funds remaining available until 2024.¹ The funding provisions give states broad latitude in how to spend the money to combat the virus and the economic turmoil left in its wake.² But that discretion is limited in one respect that has proved controversial: The funds cannot be used to “directly or indirectly offset” state tax cuts. The relevant language says:

The reaction to this section — alternatively described as the tax mandate or offset provision, and which we will call the tax provision — has been fierce. Soon after enactment, the attorneys general of several Republican-led states wrote that an aggressive interpretation of the provision “would represent the greatest attempted invasion of state sovereignty by Congress in the history of our Republic.”⁴ Ohio filed suit for a preliminary injunction against the tax provision, contending that it exceeded Congress’s power under the Constitution’s spending clause. Ohio argued that the tax provision “gives the States a choice: they can have either the badly needed federal funds or their sovereign authority to set state tax policy.” Ohio continued, “In our current economic crisis, that is no choice at all,” but is instead “a metaphorical ‘gun to the head.’”⁵ Several other states quickly filed similar suits,⁶ which “emerged as the Biden administration’s first major legal battle.”⁷ Many of these lawsuits are now on appeal.

Our purpose in this article is not to comprehensively evaluate the merits of these suits. Instead, we seek to explain and contextualize the tax provision. We have several related observations — observations that make the tax provision more likely to be consistent with the spending clause and hardly unprecedented from the perspective of existing policy.

First, in key respects, the tax provision, as implemented in Treasury’s recent rule,⁸ is a variation of a familiar animal: a maintenance-of-effort (MOE) provision. These provisions — which are designed with the sensible goal of helping to ensure that federal spending “supplements and not supplants” state spending — are common, with scores of examples scattered across numerous titles of the U.S. Code.⁹ MOE provisions usually take the form of a condition attached to new federal grants for, say, higher education, and require that the state spend at least as much of its own funds on higher education going forward as it did in a base year, or face losing the new funds.¹⁰ Although money is fungible, these provisions encourage states to not slash their own education budgets or use federal funds for non-education purposes.¹¹ With ARPA’s tax provision, the twist is that the MOE incentive applies at a higher level of generality — broader than the level of a specific program, and only in areas where the state uses the funds. As we explain below, the tax provision thus offers states greater flexibility compared with traditional MOE provisions because the state may shift its own spending (and indeed federal funds) across covered activities. In theory, there can be tradeoffs — not all of which are present in this case — between broader MOE provisions and conventional, program-specific MOE provisions. But because of the inextricable link between state revenue and spending, we find that traditional MOE provisions put at least as much pressure on state tax policy as the tax provision. If a state wants to cut taxes — and wants to do so in part by reducing its spending in a program covered by a traditional MOE provision — it will lose federal grants, usually dollar for dollar.

Second, and more generally, we argue that it is not unprecedented for the federal government to reduce its spending on a state if the state cuts its taxes, as would happen for offsets under the tax provision. This already occurs indirectly through the state and local tax deduction: When state and local taxes are deductible at the federal level, states have more of an incentive to maintain or increase taxes than they otherwise would. Put differently, when a state or locality cuts its taxes by $1, usually less than $1 ends up in the hands of its residents because the cut also reduces the residents’ potential federal income tax deduction for state and local taxes by $1. More directly, there is a close parallel between the tax provision and the State and Local Fiscal Assistance Act of 1972, which for over a decade gave states grants based in part on their tax effort, which was defined as the total state and local taxes divided by total state income. In other words, when states cut their taxes, their fiscal assistance grants fell. Moreover, the magnitude of federal funds at stake under both the 1972 act and the state and local deduction is comparable to state and local assistance under ARPA. These comparisons suggest that there has long been a complicated interplay between federal law and state tax incentives — with large federal policies often applying pressure on state taxes — that is not captured by broad and simple claims of a strict separation between the state and federal levels.

ARPA and Treasury’s Rulemaking

ARPA is a $1.9 trillion piece of legislation with dozens of individual programs spread across hundreds of pages. The programs include an extension of supplemental unemployment benefits, tax rebates of up to $1,400 for individuals, and an expansion of the child tax credit.¹² One of the most significant ARPA provisions was to give directly to state and local governments through two fiscal relief funds — new sections 602 and 603 of the Social Security Act — administered by Treasury. Section 602 creates a nearly $220 billion fund for states, territories, tribal governments, and the District of Columbia; section 603 creates a more than $130 billion fund for local government entities.

Most federal grant programs are program or area specific, but the ARPA state and local funds dispense federal money with unusually broad limits. The contrast can be seen in ARPA itself. Elsewhere in the act, for instance, Congress created a large variety of program-specific grant schemes, including to support midsized food processors and distributors, rural healthcare, socially disadvantaged farmers, and the supplemental nutrition assistance program.¹³ ARPA’s state and local funds, by contrast, are limited only to broad eligible-use categories. Funds must be used:

• “to respond to the [COVID-19] public health emergency . . . or its negative economic impacts;”

• “to respond to workers performing essential work” during the emergency;

• “for the provision of government services to the extent of the reduction in revenue” of the political entity as compared with the last fiscal year before the COVID-19 emergency; or

• “to make necessary investments in water, sewer, or broadband infrastructure.”

Should the recipient use the funds outside these categories, the Treasury secretary may withhold future funds or recoup them to the extent of the violation.¹⁴ But only the funding for states and territories contains the tax provision — reportedly inserted at the behest of Sen. Joe Manchin III, D-W.Va., in response to Republican Gov. Jim Justice’s proposal to phase out West Virginia’s income tax¹⁵ — that has produced recent controversy and litigation.

On May 17, 2021, Treasury published an interim rule implementing (among other things) the funding provisions of sections 602 and 603, and on January 6 it published a final rule, effective April 1.¹⁶ Both the interim and final rule describe the “eligible use” provisions of 602(c)(1) and 603(1) and include examples of eligible uses that count as responses “to the public health emergency . . . or its negative economic impacts.” Treasury implemented the tax provision by outlining an annual reporting process by which states calculate whether they have used ARPA funds to offset a tax cut.¹⁷ First, states must identify planned changes in law that will (according to their own scoring methods) reduce tax revenue; those changes only matter if they exceed a de minimis safe-harbor level. Next, states calculate the amount of tax revenue recorded in the year for which they are reporting. Finally, if the revenue recorded is less than the amount of revenue recorded in the fiscal year ending in 2019 (adjusted for inflation), states may identify other state law changes that permissibly offset the shortfall between the value of the tax cuts they reported at the first step and any other general revenue growth. These permissible offsets are state tax increases and spending cuts if the cuts are not in areas in which the state is using its section 602 ARPA funds.¹⁸

In other words, under the tax provision, a state may cut taxes without any recoupment if it has sufficient offsetting revenue growth relative to fiscal year 2019 or has made offsetting spending cuts outside the areas in which it spends the section 602 funds. The tax provision will thus generally bite only when a state cuts both taxes and its own spending in areas where it also spends section 602 funds.¹⁹ As we explain in the next section, in practice this means that the tax provision is broadly analogous to a familiar MOE provision that treats the areas in which a state spends section 602 funds as a single program.

Comparing Section 602(c)(2) and Traditional MOE Provisions

Suppose a state spends $10 million a year on higher education, and the federal government wants to encourage the state to spend $1 million more. The fungibility of money makes it impossible for the federal government to ensure that a $1 million grant given to the state for higher education will in fact achieve this purpose. The state may simply take the federal funds and reduce its own spending by an equivalent amount, and total education spending will remain $10 million. To avoid this common difficultly, the federal government does not typically rely on the flypaper effect or attempt to quixotically trace fungible grant dollars.²⁰ Instead, it frequently uses MOE provisions to help ensure that federal funds achieve their intended purpose. In the simple education hypothetical described above, an MOE provision might require the state to maintain its own spending on higher education compared with a base year or forfeit grants equal to a spending shortfall. “Maintenance of effort” as a term of art has appeared in the U.S. Code more than 100 times, and usually takes the basic form described.²¹ Matching fund requirements — which are likewise common in the code — work similarly.²²

When a state accepts federal funds that have an MOE provision, it discourages state tax cuts, just as it puts some pressure on the state to maintain spending. State spending and tax revenue are inextricably linked. (We generally assume in our analysis that states are required to balance their budgets, and thus that spending and revenue are the same.)²³ In the education example, for instance, if the state would prefer to spend the $1 million federal education grant on the state police, it cannot do so without giving up the funds — which would happen if it used the federal dollars for state police or reduced its own higher education funding below the $10 million baseline. But the state can reshuffle taxes and spending outside higher education without consequence: It can add funding to the state police by reducing non-education spending or by increasing taxes. Likewise, if the state would prefer to use the $1 million in federal education spending to reduce state taxes, it cannot do so without giving up the funds — though the state can still reduce spending outside higher education to cut taxes without consequence. As a result, every binding MOE provision reduces the attractiveness of state tax cuts by reducing the attractiveness of spending cuts. The state can only cut taxes by reducing spending outside the MOE-covered activity or giving up the federal grant money.²⁴

Historically, we have found no evidence that this familiar legislative mechanism has created controversy under the spending clause. Nor is it obvious why these provisions would. The spending clause gives Congress broad power to create spending programs that pursue the general welfare, including attaching conditions to federal funds accepted by states and local governments. Under long-standing Supreme Court precedent, conditions on federal funds are constitutional if they relate to the purpose of the spending, are clearly stated, and are not unduly coercive.²⁵ So long as MOE provisions require maintenance in the federally subsidized areas — reflecting the federal purpose of supplementing rather than supplanting local spending — they should satisfy these general requirements, a fact we see reflected in the limited case law on the subject.²⁶

Does the sheer size of the ARPA program matter? We think probably not. In its most relevant spending clause precedents, the Supreme Court has looked to both the size and preexisting nature of federal spending in deciding whether a program is unduly coercive in violation of the spending clause, while emphasizing that the federal government retains significant latitude in attaching conditions to the use of new funds.²⁷ With preexisting funds distributed to the states, the federal government has created a kind of understandable reliance interest. But for funds that are new — and in which the states do not have an established reliance interest — it would be anomalous to craft a constitutional doctrine that reduces the federal government’s authority as the size of the program increases. Such a doctrine would suggest, paradoxically, that the largest federal programs are the ones over which the federal government should have the least control.²⁸ Also, as we note below, the size of the ARPA program is of the same order as previous federal programs that applied a similar amount of pressure on state taxation — suggesting that ARPA’s size, distributed over the course of several years, is no anomaly.

Moreover, in contrast to the baseline MOE provision sketched above, ARPA gives states even wider discretion in how to use the funds allocated in section 602. It is worth comparing the ARPA funds to a system in which Congress provided activity-by-activity MOE provisions for the kinds of uses outlined in section 602(c)(1) — $X for vaccination programs, $Y for essential workers, $Z for infrastructure projects, and so forth — conditional on states maintaining their existing spending in each category. In such a case, if a state wanted to put the $X earmarked for vaccination programs toward essential workers, it would be unable do so, nor could it cut its own vaccination funding below the baseline without losing grant money. Given the breadth of effects from the pandemic, a set of activity-by-activity MOEs covering pandemic responses could well have locked in a large fraction of state budgets.²⁹

By contrast, section 602 gives states considerably more choice over how to spend in the ARPA-covered categories and how to allocate federal subsidies, because states are free to reallocate funding under the broad ARPA umbrella — or simply not use the federal funds in a particular area. As implemented by Treasury, the tax provision acts as a kind of MOE provision that treats all the relevant categories of state spending in 602-covered areas as a single aggregated activity.³⁰ A state tax cut is offset indirectly by section 602 funds only if the state has reduced total spending on section 602 ARPA-covered activities in which the state is using the funds. Viewed through this lens, the tax provision looks strictly preferable for states to activity-by-activity MOEs, because it gives states, which we would expect to have more expertise about local spending needs and desires, more flexibility to allocate federal funds as they see fit. That additional layer of flexibility is broadly akin to the benefits that cap-and-trade regimes enjoy over command-and-control ones, with the former more flexibly allocating an entitlement based on local conditions, rather than with more central planning.

A simple numerical example illustrates the contrast. Suppose that the activities covered by section 602 are $20 billion of a state’s preexisting budget and that there are four equally sized activities within that budget — roads, schools, healthcare, and police — with $5 billion of state spending on each. If the state were subject to four separate MOE provisions, a reduction in state spending in any of the four areas would result in the state giving up grant money. The MOE provisions would be violated whether money went from schools to healthcare, from schools to a non-covered activity like corrections, or from schools to a tax cut. By contrast, the tax provision requires loss of grant money only if total state spending on the four activities falls below $20 billion and there was a tax cut without sufficient organic revenue growth or spending cuts outside the ARPA areas to cover it. Cases in which the tax provision would require recoupment of grant money appear to be a strict subset of those in which activity-by-activity MOE provisions would require the equivalent loss of grant money.

Indeed, Treasury’s implementation of the tax provision appears to give states one final advantage over activity-by-activity MOE provisions: Under section 602, a spending category only counts toward the maintenance total if the state uses federal funds in that area. Suppose that, absent the section 602 funds, the state would like to increase funding of healthcare to $7 billion, decrease funding of roads to $2 billion, leave the remaining two areas at $5 billion, and cut taxes by $1 billion. If the state directly spent the section 602 funds in all four areas, it might violate the tax provision because its spending has dipped below $20 billion. By contrast, if it directs all its section 602 funds into eligible healthcare uses under section 602 and continues to spend at least $5 billion on healthcare, it will not face recoupment under the tax provision because it has not cut spending compared with the status quo in its chosen section 602 areas, even if it has cut spending in potential section 602-eligible areas.

Not all MOE provisions that operated closer to the level of a state’s budget would necessarily have the structure of section 602 as implemented by Treasury. For example, it is possible to imagine an MOE provision that conditions federal grant dollars on a state maintaining total revenue or spending. This provision would present difficult and interesting tradeoffs when compared with a set of activity-by-activity MOE provisions. In the example, for instance, if the state’s total tax revenue was $30 billion, the state would not have to give up grant funding under the activity-by-activity MOE provisions if it cut spending in a non-covered activity like corrections by $2 billion, and at the same time cut tax revenue from $30 billion to $28 billion. By contrast, if the MOE provision conditions funding on total revenue or spending, the state’s actions would require giving up federal grant money. Under the total spending MOE, the state would get the freedom to move spending and grants around that it would not have under activity-by-activity MOEs. But the cost to the state would be that it cannot cut spending in a non-covered activity to cut taxes without giving up grant money.³¹

The motivations that apply for an ordinary MOE provision — a desire to ensure that federal spending does not supplant state spending in a single program — can also apply at more general levels. The federal government may reasonably wish to encourage state spending in a broader category such as pandemic response, or even wish to encourage total spending, as may be especially desirable during an economic downturn. Or the government may simply view a state tax cut as a sign that federal help is less needed.³² In our view, these differences between ordinary MOE provisions and ones applying at more general levels are ones of degree, not kind.

Federal Law and State Taxation in Wider Context

Attaching MOE provisions of matching requirements to federal spending programs is, while common, hardly the most salient thing that the federal government does. But the general implications of those efforts — encouraging states to alter or maintain tax and spending policies — are more common than many observers may realize. We conclude by briefly discussing two more general examples of how federal policies can interact with a state’s incentives in setting its taxes: the familiar state and local tax deduction and the State and Local Fiscal Assistance Act.

Federal law has long allowed taxpayers to reduce their federal tax bills by deducting a portion of their income that they pay in taxes to state and local governments — an amount recently (and somewhat controversially) capped at $10,000 by the Tax Cuts and Jobs Act.³³ Most Americans pay less than the cap in state and local taxes — an average of $8,500 in 2017 — and, that same year, about half of all states had average individual state and local tax deductions that were below the $10,000 cap.³⁴

Because the SALT deduction reduces the burden of state and local taxes on taxpayers, it provides an effective subsidy for states and localities to increase their taxes — a subsidy, moreover, that varies in its attractiveness from state to state. If a state cuts its income tax by $1, less than $1 will end up in the bank account of many of its citizens, because those citizens would be paying more in federal taxes after the concomitant reduction in their SALT deductions. The state would be giving up a portion of the subsidy implicit in the SALT deduction.³⁵ When the federal government capped the deduction in 2017, it diminished the size of that effective subsidy, though it still varies substantially by state. Increasing the cap — a proposal to increase the cap to $80,000 has been in the news³⁶ — would increase the subsidy. The deduction and these potential changes amount to a substantial and costly federal subsidy for the states. In 2019 the Joint Committee on Taxation estimated that repealing the SALT deduction cap — and returning to the prior, long-standing system in which state and local taxes were fully deductible — would cost the federal government more than $77 billion in annual revenue.³⁷

The constitutionality of the SALT cap, like the constitutionality of section 602, has also been challenged, with the states of New York, Connecticut, Maryland, and New Jersey suing to enjoin its enforcement on constitutional grounds. Both the Southern District of New York and the Second Circuit found the claim meritless.³⁸ But the point we hope to emphasize is independent of the merits of that suit: Regardless of what deduction regime is used, any deduction regime affects the states’ incentive and ability to raise or lower taxes.

A second example of the interplay between federal funds and state taxes is the State and Local Fiscal Assistance Act of 1972, which created a long-standing system of revenue sharing between the federal government and state and local governments. The idea for this revenue sharing program was hatched in the 1960s by economist Walter Heller, signed into law by President Nixon, and extended in 1976, 1980 and 1983, before finally expiring in 1986.³⁹ The act and its extensions contemplated the federal government sharing revenue with the states with few strings attached. Section 103(a) of the original act, for instance, required only that the funds be used for “priority expenditures,” including the “ordinary and necessary maintenance and operating expenses” of government services.

While the act occasionally arises in discussions of fiscal federalism, it is less commonly remembered that its allocation formula disbursed funds based in part on a local jurisdiction’s “tax effort,” which the law defined as a ratio of state tax revenue to state income. States that raised more revenue from the same ability to pay — and thus displayed a greater tax effort — received a greater share of federal revenue. This allocation scheme had a policy goal like what the policymakers have for a standard MOE provision: to allocate federal funds without displacing the states’ own taxing and spending efforts. In his message to Congress on revenue sharing, President Nixon declared that “the revenue effort adjustment is designed to provide the States with some incentive to maintain (and even expand) their efforts to use their own tax resources to meet their needs.”⁴⁰

Like the SALT deduction cap and section 602 of ARPA, the revenue sharing program was not without controversy, and aspects of the act’s implementation were challenged in court.⁴¹ But everyone understood that a state cutting its taxes resulted in a reduction in federal grants under the act, and we have found no argument that the law exceeded Congress’s constitutional authority under the spending clause. The magnitude of the 1972 act and the revenue sharing program that followed is also roughly comparable to the size of the ARPA funds. The 1972 act distributed more than $30 billion in 1972 dollars — or about $200 billion in 2022 dollars.⁴² Over all the years it was in place, the revenue sharing program distributed more than $400 billion (in 2022 dollars) to states and localities; the ARPA provisions creating the state and local funds appropriate a total of $350 billion.⁴³

Conclusion

The purpose of this article is to put ARPA’s tax provision in context. Viewing it as an MOE provision that sits at a level higher than a single activity, we think, helps clarify what it gives states and what, if anything, it takes away. It is not unusual for states to give up federal grants or subsidies when they cut taxes — an observation that may be helpful in thinking more generally about whether ARPA’s tax provision is consonant with the spending clause.