abstract. The federal response to the COVID-19 pandemic in relation to the states has been marked by a paradox: while the federal government has provided substantial aid to the states, including general revenue support, states excessively cut their budgets early in the crisis. This paradox stems from the sizeable time gap between the onset of the pandemic and the arrival of general federal aid. The federal government provided much of its direct support to states long after the pandemic began and without prior assurance that such support was coming. Congress passed and President Biden signed the American Rescue Plan Act (ARPA) in March 2021, over a year after the onset of the pandemic.
 Both the Coronavirus Aid, Relief, and Economic Security (CARES) Act, passed in March 2020, and ARPA contained some novel elements as to federal-state fiscal relations. The CARES Act established an emergency-liquidity facility to help the municipal-bond market. ARPA provided states with general fiscal support to make up for revenues lost during the crisis. Regulations interpreting ARPA attempt to measure the size of state deficits caused by the crisis.
 This Essay argues that these innovations were successful, though imperfect, and should be built upon. In particular, I argue that the federal government could accomplish more, and likely spend less, by establishing a default federal-credit facility to aid the states during crises. The current system, which makes it hard for the states to borrow during an economic emergency, protects too much against the danger of moral hazard (i.e., excessive borrowing by the states) and protects too little against the danger of states making procyclical cuts during a crisis.
 This Essay’s proposal for a federal state-loan program has two basic components. First, during an economic crisis, the Federal Reserve would make loans to states on favorable terms, with the size of the loans restricted so as not to permit a bailout of state profligacy. Second, at the conclusion of the crisis, the Federal Reserve would forgive a substantial portion of the loans if states can certify that their loan funds were used for mitigating the crisis. Such a system represents an attempt to achieve a better balance between the harms of moral hazard and procyclical cuts.

Introduction

The federal response to the COVID-19 pandemic in relation to the states has been marked by a paradox: while the federal government has provided substantial aid to the states,1 including general revenue support, states have excessively cut their budgets.2 This paradox stems from the sizeable time gap between the onset of the pandemic and the arrival of general federal aid. The federal government provided much of its direct support to states long after the pandemic began and without prior assurance that such support was coming.3 Congress passed and President Biden signed the American Rescue Plan Act (ARPA) in March 2021, over a year after the onset of the pandemic.4

This Essay argues that one lesson to be learned from the COVID-19 economic crisis is that the federal government could accomplish more, and likely spend less, by establishing a default federal-credit facility to aid the states during crises. The current system, which makes it hard for the states to borrow during an economic emergency, protects too much against the danger of moral hazard (that is, excessive borrowing by the states) and protects too little against the danger of states making procyclical cuts during a crisis.

This Essay’s proposal for a federal state-loan program has two basic components. First, during an economic crisis, the Federal Reserve should make loans to states on favorable terms, with the size of the loans restricted so as not to permit a bailout of state profligacy. Second, at the conclusion of the crisis, the Federal Reserve should forgive a substantial portion of the loans if states can certify that their loan funds were used for mitigating the crisis. Such a system could achieve a better balance between the harms of moral hazard and procyclical cuts.

This Essay proceeds in three parts. In Part I, I provide relevant background on the structure of American federalism, the relief provided by ARPA, and ARPA’s limitations. In Part II, I draw lessons from the implementation of ARPA and other relief measures. Finally, in Part III, I build on those lessons and propose an automatic mechanism for providing general aid to the states during crises.

I. american federalism and arpa

A. American Federalism

Before proceeding to analyze ARPA and its relationship with the states, this Part sets forth the key features of American federalism that are relevant to the discussion of the government’s response to the COVID-19 economic crisis.

The first feature is that states are major employers (accounting for about 13% of employment in the United States),5 directly contribute to national GDP (10.9% of GDP in 2019),6 and provide most frontline governmental services.7 As for frontline services, about one-third of state expenditures go to education,8 which constitutes about 92% of all expenditures on education in the United States.9 States are also the major providers of funding for infrastructure either directly10 or indirectly11 through federal grants.

When states lay off their employees, as they did with over a million employees during the COVID-19 pandemic,12 severe harms result. Layoffs amplify economic distress, rendering spending cuts procyclical. And it is not desirable to cut services that are important regardless of economic conditions, such as education. This is not to say that the current set of services that states provide are sacrosanct and should not be reduced or changed to achieve efficiencies or address a structural deficit. Some cost-cutting measures during a recession, such as a hiring freeze, might even be sensible. But sweeping cuts triggered only by a national recession are unnecessarily harmful because they do not respond to the long-term financial situation of the state or the operations of the services in question.

The second feature of American federalism I wish to highlight is that states are tasked with significant countercyclical spending responsibilities, such as paying their share of Medicaid and unemployment benefits. Indeed, over 40% of states’ expenditures go to public-welfare programs such as Medicaid, Temporary Assistance for Needy Families, and Supplemental Security Income, which are funded through a mix of state and federal dollars.13 By design, these safety-net responsibilities increase during a downturn.14

These two features of state expenditures are roughly separable. Suppose the federal government started paying for Medicaid and unemployment, just as it currently operates Medicare and Social Security.15 In such a world, states would retain primary responsibility to fund many other core functions of government, such as education, with their own revenues. They would therefore still exercise significant fiscal responsibility both in absolute terms and relative to other federal systems.

A third important component of our system of fiscal federalism is that states are bound by hard-budget constraints. Hard-budget constraints require states to balance their budgets every year, and do not allow them to borrow during an emergency.16 Such constraints are thought to be essential for subnational units in a federation.17 Without them, those units would be tempted to spend too much and then turn to the central government for bailouts when things went bad.

This is not just a hypothetical, as such behavior by subnational units has been observed in other federations and this one.18 In the 1830s, many states borrowed significant sums of money to finance grand infrastructure projects.19 When their projects failed, the states hoped that the federal government would intervene, but it did not.20 Two important lessons were drawn from these events. First, the federal government established a no-bailout norm.21 Second, states, having been burned by the no-bailout rule, placed balanced-budget rules in their constitutions.22 Thus, as a general rule, states can no longer borrow during a crisis to cover general government expenses or (increased) safety-net expenses.

The fourth feature, and it is in tension with the third feature of American federalism, is that the hard-budget constraint is somewhat porous. The federal government does regularly provide additional support to states for safety-net services during a crisis. This is arguably a partial bailout because, at least in theory, the federal government could insist that states continue to provide their share of the revenue for these programs themselves. For example, during the COVID-19 pandemic, the federal government provided aid to the states for their share of Medicaid and provided grants for pandemic-related services.23 It provided similar support to the states for Medicaid in the midst of the Great Recession.24 The federal government also sometimes helps with other government services that are not explicitly part of the safety net, as when it supported public education with approximately $40 billion dollars during the Great Recession.25 Further, the states themselves engage in various maneuvers, such as intrafund borrowing26 and even borrowing on the municipal markets27 to avoid cuts during a downturn.

The current regime of hard-budget constraints is beset with several tensions. First, the fact that states provide so many essential governmental services calls into question whether hard-budget constraints are appropriate. For instance, it makes little sense to force states to lay off teachers every recession. Second, it is unclear whether states—as opposed to the federal government—should be funding the social safety net at all. It seems even more perverse for states to be governed by balanced-budget rules as to safety-net expenses, given that safety-net expenses are supposed to increase during a recession when state revenues go down.28 Third, both the federal government and the states act as if they are implicitly aware of the issues with hard-budget constraints during downturns, but their actions to loosen the constraints are ad hoc.

B. The ARPA Saga

Though the federal government acted quickly when the pandemic hit to aid state and local governments, that federal aid came primarily in the form of earmarked funds to mitigate states’ increased expenditures. First, on March 18, 2020, Congress enacted the Families First Coronavirus Response Act.29 In light of increased state Medicaid expenditures to treat COVID-19, that Act temporarily raised the federal government’s share of Medicaid expenses by 6.2%.30

Then on March 27, 2020, Congress passed the Coronavirus Aid, Relief, and Economic Security (CARES) Act,31 a $2.2 trillion act that provided $150 billion in state and local aid.32 States and certain localities could use the funds for “necessary expenditures” incurred prior to December 31, 2020 that (1) resulted from the COVID-19 public-health crisis and (2) were not accounted for in their most recent budget.33 Because this aid was earmarked, it could not be used for general revenue shortfalls.34 Further, the amount of earmarked aid ($150 billion) appeared too small relative to projections made at the time as to state and local budget shortfalls, which were as high as $650 billion.35 Prominent voices, including Ben Bernanke36 and Jerome Powell,37 called for more aid for the states.

Section 4003(b) of the CARES Act provided up to $434 billion in loans to state and local governments, as well as private businesses.38 The Federal Reserve used $35 billion of this funding as an equity contribution to launch a $500 billion lending program to the states and larger localities, called the Municipal Liquidity Facility (MLF).39 Under the MLF, the Fed could purchase “tax anticipation notes (TANs), tax and revenue anticipation notes (TRANs), bond anticipation notes (BANs), revenue anticipation notes (RANs), and other similar short-term notes issued by Eligible Issuers,” but the notes could not mature later than thirty-six months from the date of issuance.40 This was its critical shortcoming: three years was too short a payback period if the states were going to borrow to cover large operating deficits. In order to meaningfully help prevent dramatic, harmful budget cuts, the MLF should have allowed states to pay back these loans over a much longer term. In fact, only two borrowers took advantage of the program.41

In May 2020, the House passed the $3 trillion HEROES Act,42 a follow-up bill to the CARES Act that would have provided $875 billion in general aid to states and localities.43 The allocation formulas were based on states’ unemployment rates and Community Development Block Grants for cities.44 The Act also would have provided additional and explicit authority for the Federal Reserve (Fed) to make loans for up to ten years under “unusual and exigent” circumstances at the same low rate at which the Fed makes loans to banks out of its discount window.45 But this bill went nowhere in the Senate.46 A bipartisan group of senators introduced the SMART Act in May 2020.47 This bill would have provided $500 billion in general aid to state and local governments,48 but it never went beyond being introduced.49

Nevertheless, negotiations about another relief package continued through the summer and fall, with state and local aid often a major sticking point.50 No bill was completed until December 2020. The bill that passed in December neither included aid for the states51 nor provided any new powers to the Fed.52

As 2020 came to a close, even with a new President soon to take office, the states were right to be wary about receiving any general aid.53 For that to happen, Democrats needed to win both Georgia Senate runoff elections in January 2021. Even after they did, passing ARPA still required unanimity among the Democratic Senate caucus, which was hardly a certainty.54 Nevertheless, President Biden signed ARPA into law on March 11, 2021, providing $1.9 billion in aid.55

Like the CARES Act, ARPA provided specific, programmatic aid to states. But like the proposed HEROES Act, ARPA also provided states with substantial general revenue support: $219.8 billion to states, territories, and tribal governments and $130.2 billion for metropolitan cities, municipalities, and counties.56 In addition, ARPA increased funding for many special-purpose local governments like regional transportation authorities57 and school districts.58

Programmatic aid came in many buckets. For example, over $30 billion was made available for public-transit agencies, to be generally distributed according to already existing formulas,59 though ARPA specifically permits all of its support to be used for operating expenses,60 where the usual rule is 50%.61 Some funds, such as for rental assistance, required states to set up new bureaucracies and programs—at least new since the American Rescue Plan Act of 2021.62

By the time ARPA was being finalized in the early months of 2021, even proponents of more federal aid noted that the worst projections of state budget shortfalls had not occurred.63 There are many reasons for this outcome, including the wisdom of the generous unemployment support provided first by the CARES Act, whose support not only kept people out of poverty (and safely at home), but also allowed for continued consumption (and hence tax revenue).64

It is important to note that the more dire forecasts as to state revenue were based not only on reasonable models of previous recessions,65 but on actual tax receipts.66 Total state revenue declined by almost 50% in April 2020 (compared to April 2019) and was still down 10% in July of 2020.67 By the end of 2020, overall state revenues were only down 1.8%, with big variations, as revenues fell by more than 10% through the end of 2020 in seven states (and increased in twenty-two states).68 Given that states operate under balanced-budget rules, it therefore made sense to cut, rather than expand, during the spring and summer of 2020.

The relative size of the general revenue support provided by ARPA to the states appeared large compared to their projected deficits when Congress passed ARPA.69 Given that many states were no longer staring down giant budget holes, some state politicians made it clear that they would use the extra budgetary room to make state-level tax cuts, including regressive tax cuts.70 Congress wanted ARPA relief funds to support the general government services, particularly those most affected by the crisis; it did not want to fund regressive tax cuts.71 Accordingly, later in the drafting process, a proviso was added to the bill that prohibited using ARPA funds for tax cuts.72

One might reasonably ask why ARPA did not simply reduce the amount of funding upfront, rather than insert the clawback at the last minute. As a matter of policy and institutional structure, there were several sound reasons. First, different states had different needs relative to their budgets,73 but presumably politics demanded only one aid formula. Second, a broader survey of the costs imposed by the COVID-19 pandemic revealed additional needs beyond simply measuring the drop in revenue that the pandemic caused.74 For example, schools might have needed to spend extra to catch students up or local public-health agencies might have needed to spend more to vaccinate hard-to-reach populations. To do their jobs well, state and local governments needed to be made more than whole. Third, amid uncertainty about the course of the disease and the politics of further aid, Congress might have sensed that the amount provided would have to last the remainder of the crisis. And indeed, with the Delta variant driving a surge in cases in the late summer of 2021, the crisis seems far from over.

C. Consequences

As a consequence of the rollercoaster ride of COVID-19 relief legislation, states cut too much and invested too little, including in public health, in the second half of 2020.75 Projected budget deficits76 and the size of state reserve funds77 both drove these cuts.

Furthermore, the need to create new programs puts a lot of strain on state and local governments. For example, in order to keep people in their homes, the Consolidated Appropriations Act of 2021, passed in December of 2020, contained $25 billion for emergency rental assistance.78 The states have had a very hard time getting this support to the citizens that need it79 and hence improving implementation is now the subject of major support efforts by multiple federal agencies.80 Sometimes new crises will occasion the need for new programs, but starting new programs in a crisis should be minimized.81 If new programs must be set up in a crisis, then their success will require extraordinary federal-state-local cooperation.82

States’ cutting and underinvesting harmed not only the overall economy and the effectiveness of the government’s COVID-19 response, but also our most vulnerable populations.83 Many different strands of policy making have converged to increase inequality and vulnerability.84 Beyond the vulnerability caused by lack of resources generally, there is a long history of policies that have harmed certain communities, often, but not always, communities of color.85 A poor government response to a crisis thus has the tendency to disproportionately burden those communities that have been rendered more vulnerable by government action (or inaction) during “normal” times. This should not be an acceptable state of affairs.

The last-minute attempt to claw back any ARPA funds used to fund state tax cuts has generated litigation. If the courts strike down the clawback provision for broadly interfering with the states’ fiscal affairs, as a federal district court in Ohio did,86 then this could pose an actual roadblock for a future Congress seeking to provide ad hoc general aid to states with minimal provisos. Even if the federal government prevails, the litigation could still have a chilling effect. The federal government might win on standing grounds, as it did before a federal district court in Arizona,87 and thus the issue would not be reached on the merits. As a result, Congress could not be sure if it could claw back aid for general governmental purposes that was redirected to tax cuts and hence might just provide targeted aid instead. Ironically, the arguments against the clawback provision, which are nominally based on federalism concerns, would have the effect of disempowering the states because the end result of the litigation could well be that Congress would continue to provide aid to the states with specific requirements, rather than general aid to the states with only a handful of restrictions.

One collateral effect of the litigation and clawback is that the Treasury has promulgated an Interim Final Rule88 that contains a reasonable formula for assessing the deficit caused by the pandemic—and thus a permissible use for federal funds.89 That is, if a state’s deficit is larger than its allocation under ARPA, then the state is in no danger of the clawback because the Treasury will assume that all of the federal support went towards the general deficit. As outlined below, the federal government can and should build on this innovation.

In sum, the consequences of the delay in federal aid suggest that in the future, the federal government could improve both the efficiency and equity of its response to fiscal crises. Preventing cuts earlier in a crisis would allow for more efficient allocation of resources and allow for more flexible ongoing support, which could prevent overcautious belt tightening during and “after” the crisis. Further, because these cuts disproportionately hurt the disadvantaged, it would be more equitable to craft a more efficient response.

II. the limits of ad hoc fiscal federalism

The ongoing ARPA saga yields numerous lessons about the limits of ad hoc fiscal federalism.90 Though many of these lessons should have been learned already, such as the need for automatic federal support for the states, it is worthwhile to canvas these old lessons. In this Part, I begin by outlining how the current crisis showcases the limits of ad hoc fiscal federalism. Further, there are new lessons that we learned from the COVID-19 pandemic, and those new lessons interact with the old ones in interesting ways. One new lesson involves the success of the MLF. If providing liquidity to the states was a good idea in this crisis, and if there is, in general, a need for more automatic aid to the states, then why not make some kind of liquidity facility automatic? After outlining the lessons to be learned from past and present crises, this Part answers this question and presents the case for an automatic liquidity facility.

A. Lessons from ARPA

First, the federal government should return to the study of our federal system and its programs, including how to provide different kinds of support to states under different scenarios and the effects of different aid formulas.91 ARPA relied on the tools at hand to distribute funds, such as the number of unemployed in a state or the relative amount of aid a city would receive under Community Development Block Grant program, which were reasonable choices under the circumstances. But we do not need to settle for reasonable estimations of need based on programs designed for other purposes when it comes to the allocation of hundreds of billions of dollars. Historically, the federal government has conducted such studies through the Advisory Commission on Intergovernmental Relations (ACIR), but the Commission was disbanded in 1996 by a Republican Congress, with the acquiescence of a Democratic President.92 The ACIR should be reestablished and tasked with establishing thoughtful formulas for providing specific and general governmental aid to the states.93

Second, as part of its role of providing macroeconomic support during crises, the federal government should establish more automatic mechanisms to help the states.94 If the federal government will not take over Medicaid or unemployment insurance completely,95 then it should put into place a regular and predictable default rule for additional support in times of crisis.96 For example, if a state’s unemployment numbers get too high, then it should be able to count on the federal government to cover a fixed additional percentage of unemployment or Medicaid. Such formulas can always be changed, but given the bias in our system towards inaction, default rules providing baseline support are crucial to ensure timely aid to the states.

Again, ideally, states should not retain a role in Medicaid or unemployment insurance. However, if states are to retain some role because it is thought that state involvement allows greater diversity in program design to meet the unique needs of each state, then the formulas will be complex. The formulas will need to be not only efficient and fair between states, but also consistent with, for example, meaningful state choice within a certain range. There is—or should be—a close relationship between the first two lessons canvassed so far, namely the need for more study and for automatic rules. If we are going to make rules automatic (as we should), we should choose ones that were at least designed for the task at hand and then have some mechanism to assess their success.

Third, the COVID-19 crisis reaffirmed that states with diversified tax bases and ample reserve funds were more fiscally stable and needed to make fewer destructive procyclical cuts. This is, first of all, a lesson for the states. It also indicates that it is in the interest of the federal government for the states to have many tax bases97 and large reserves98 because the federal government should not want state cuts to exacerbate national crises. One surprising aspect of the revenue dynamic during the COVID-19 pandemic was the relatively strong performance of income taxes, particularly progressive income taxes.99 The takeaway is not that income taxes are the best state tax, but rather that a diverse set of tax bases in a state helps limit the potential economic downturn during crises.

The fourth lesson, thus far applied most directly to the federal government, relates to issuing treasury bonds with longer maturities.100 At least at the moment, the current crisis has been accompanied by historically low interest rates.101 Rather than borrow for ten or twenty years at these low rates, the federal government should lock in current low borrowing rates further into the future and also spread the payment obligation for these investments so that future generations may benefit from them.

We can glean one last lesson from the success of the MLF. As noted above, the MLF provided support to the municipal market, but on such noncompetitive terms that there was relatively little use of the liquidity facility.102 Nevertheless, studies of the municipal-bond market during the crisis indicate that the municipal market returned to relatively normal functioning and interest rates once the MLF was in place (and after the passage of the CARES Act).103 Furthermore, early studies indicate that this resumption of normal functioning had real impacts on state and local employment.104 Thus, paradoxically, the MLF seemingly helped the regular municipal market offer better terms than the MLF itself offered. That is, the knowledge that the MLF was there as a backstop with its lesser terms, or perhaps the existence of the MLF signaling that someone was paying attention, allowed the municipal market to exit panic mode and continue to function. Given the limited fiscal capacity of state and local governments generally, a similar liquidity support would be helpful in future downturns.

B. The Case for Updating the Current Fiscal Rules of our Federation

Several of the lessons in the previous Section point to a further lesson. If the federal government should borrow at low rates during a crisis, then why shouldn’t the level of government that most directly provides the majority of government services borrow at low rates? If there should be automatic rules to help states cover their share of safety-net programs, such as Medicaid, why shouldn’t there be automatic rules to cover other general governmental services, such as education? The federal government should intervene in order to automatically provide some general fiscal support to states during a crisis.

To be sure, if the federal government took more safety-net responsibilities from the states and helped state revenue functions, as it should, then these reforms would reduce the need for state borrowing. Nevertheless, the states’ borrowing needs would not be eliminated completely as to general services. As a matter of political economy, the states are unable to save enough for large crises—nor should they, given the urgency of other needs.

In any event, it is unlikely that the federal government will directly take over the provision of all safety-net services. Allowing the states to effectively borrow at the federal government’s borrowing rate to pay for their share of these services is thus a justified, if convoluted, workaround. That is, instead of just borrowing and spending the money itself on, say, 100% of the unemployment program, the federal government would borrow money and then send a portion of that money to the states to pay for their share of unemployment benefits. Either way, the safety-net services are being paid for by the federal government, which can borrow at a lower cost because of its greater fiscal capacity.105

Congress must approach systematic support for states thoughtfully, of course. ARPA correctly recognized states’ need for general support, but its ad hoc nature—including its delayed timing and the uncertainty surrounding its enactment—reduced its effectiveness. Establishing systematic institutional support for state revenues would violate the hard-budget constraint on subnational governments.106 As implemented in the United States, the hard-budget constraint means that the federal government will not bail out states and that the states have bound themselves with balanced-budget rules. At least formally, these rules do not generally permit for borrowing in an emergency. As such, the current rules would be classified in the literature as “first-generation fiscal rules.”107

At least some federalism scholars now advocate for second-generation rules (and even third-generation standards).108 The second-generation rules tend to be more comprehensive. They do not exclude pension debt; they are granular, with the inclusion of numeric guideposts; and they are flexible, in the sense that more borrowing is permitted during downturns.109 It is this last feature that is crucial for our purposes. Second-generation rules understand that there are two harms that need to be balanced—moral hazard if borrowing is too easy, versus procyclical cuts if borrowing is too hard during a crisis.

Second-generation rules are not just an academic exercise. The Fiscal Compact of the European Union (EU), adopted in 2012, contains such second-generation rules, and the states of the EU committed to putting these rules into their national laws.110 These rules require, among other things, keeping budget deficits below 3% of GDP, aiming for a medium-term structural deficit of 0.5%, and putting into place an automatic correction mechanism.111 However, an escape clause from these requirements is permitted in “the case of an unusual event outside the control of the Contracting Party concerned which has a major impact on the financial position of the general government . . . provided that the temporary deviation of the Contracting Party concerned does not endanger fiscal sustainability in the medium-term.”112 Appropriately, the EU agreed to trigger this escape clause and suspend its fiscal rules through the end of 2021 on account of the current pandemic and recession.113

Putting into place second-generation rules is partially a state-law issue, a point that I will return to in future work. But states only put their formal rules into place after the federal government informally established its “no-bailout rule.”114 If the federal government were to establish the following simple default rule, loosely based on the EU Fiscal Compact and the Treasury Interim Final Rule governing the clawback provision, many states would presumably change their laws so that they could borrow funds:

The Treasury may extend loans with a term of thirty years at Treasury’s cost of funds to the states, but the principal of the loans cannot exceed 80% of the state’s certified projected budget crisis deficit (as opposed to a structural deficit) nor can payment of the loans cost more than 1.5% of the state’s last pre-crisis budget.115

A rule like this would represent an improvement over the status quo. First, it would be predictable and head off procyclical cuts. Second, because the amount of aid is based on a reasonable but not overly generous formula, this rule does not undermine state and local incentives to save for a rainy day.

But this simple formula is far from ideal. For one, it might turn out to provide too little aid or, in retrospect, too much.116 It does not take into account heterogeneity in states’ needs.117 Moreover, it does not take into account different interest-rate
environments.118 A different approach is warranted. One potential approach would be to make the rules more complicated, while another would be to put in place a fiscal standard, like states can borrow so long as the resulting deficit is not “excessive.”119

Both of these approaches have merit. As for the more refined rules, not only is there a significant body of literature on what such rules might look like, but there are entities, like the U.S. Treasury and the Fed, that have the capacity and relative independence to develop and enforce such rules. Of course, in the end, the rules will be rules. If they are too complex, then the scheme will be hard to operate. Moreover, as complex as the rules may be, they will always fail to take into account some aspect of a real crisis.

Standards have the opposite problem. They are flexible, and in some cases straightforward, but their flexibility also creates uncertainty. Further, though the courts have considerable experience applying standards, it is hard to imagine them charting a new path defining “excessive debt,” rather than following reasonable judgment calls made by federal regulators. Will a court enjoin the Fed from giving too much relief or claw it back after the fact? Alternatively, will a court force the Fed to lend more to a state? Judicial political economy and competence and any number of doctrines, such as the political-question doctrine, suggest that the answer is “no.”120 There are innumerable ways to measure when debt is excessive, so a court would thus be replacing the policy judgment of policy makers with its own policy judgment.121 Certainly, the historical experience of state constitutional jurisprudence indicates that courts are likely to defer to state legislatures when it comes to the need and size of borrowings.122

III. a way forward

In this Part, I propose that Congress should establish a new federal state-loan program, and address potential critiques. My proposal builds on—and institutionalizes—a proposal made by Thomas Cochran right at the beginning of the crisis in April
2020.123 One great merit of Cochrane’s proposal is that, even in the midst of the crisis, the federal government appeared to have the institutional capacity to implement his proposal.

A. The Proposal

In the spring of 2020, at the peak of the COVID-19 crisis, when states were making cuts rather than investments, matters looked most severe for state budgets. There were calls, including from members of Congress, for the Fed to make loans to the states on more generous terms than those outlined under the MLF as originally designed.124 More generous loan terms for the states could have been achieved by the Fed itself to some extent through tinkering with the MLF program (authorized under Section 13 of the Federal Reserve Act)125 or by using the Fed’s power to buy short-term notes under Section 14 of the Federal Reserve Act.126 Most straightforwardly, Congress could have given the Fed the explicit power to lend for general revenue purposes through amending the Federal Reserve Act. At least some members of Congress tried: this was Section 110801 of the HEROES Act.127

Suppose the Fed had made long-term credit available to the states on favorable terms in the spring of 2020. And suppose that the Fed used a rough formula, like the one arrived at by the Treasury in connection with the ARPA clawback, to police the size of the borrowing. On the one hand, this would have been a positive step and may have encouraged more borrowing, more investing, and less budget cutting. Because this would have been a borrowing program, one would expect some discipline on the part of at least some states.

But on its own, the state-borrowing program would have fallen short of providing effective aid to every state. First, as a new program, it would have been difficult to design and process applications in a timely manner. Second, even though the terms would have been favorable to the states, paying back the loans might have appeared worrisome when matters continued to deteriorate.128 Third, as discussed, the states had a reasonable, if uncertain, expectation of receiving direct support from the federal government. Therefore, they might have chosen to wait for support to arrive rather than taking out a loan.

How, then, to improve matters? First, the program would have needed to be designed and ready to go before the crisis so that it could have been administered expeditiously, and so that the states could have relied on it. More time allows for a more refined rule to better meet the needs of each state. Further, given the significant uncertainty of forecasts during crises, the rule should have allowed the states to borrow for only part of the projected deficit at a time, such as for one fiscal quarter.129

Second, along with setting up this program as a default, a further default should have been established for loan forgiveness. This is the big insight of Cochran’s proposal.130

I propose a backwards-looking loan forgiveness rule that would kick in after five years. For example, 85% of the loan would be forgiven in five years so long as the borrower could certify that the borrowing was spent filling holes caused by the crisis, including general budget holes as determined by a reasonable formula. Forgiving some percentage of the loan better balances against moral hazard. States should maintain reserve funds to support themselves during downturns and should be wary of borrowing too much. Thus, requiring that they pay 15% of the total downturn—after application of a formula whose results they might not like—seems reasonable.131 The statute and regulations would resemble the Interim Final Regulations132 that the Treasury produced in connection with the ARPA clawback. If these rules were upfront in the statute, then the legal challenges made to the ARPA clawback rule would be largely mitigated. This is because at the heart of the current legal challenges is the claim that the current statute does not make it clear what is expected of the states.133

This program would leverage the relative independence and expertise of the Treasury or the Fed to provide automatic support to state and local governments in case of emergency. Note that this independence and expertise can (and should) play three important roles. First, this expertise can help draft the right rules. Second, the expertise and independence could help assure the integrity of the program. That is, it could help forestall the concern that politically favored borrowers will get special treatment. Finally, the independence and expertise can be harnessed to tweak the rules in light of experience or surprising circumstances. Not surprisingly, another line of argument in the second-generation fiscal-rules literature maintains that it is better when the rules are not entirely set in advance but can be adjusted by an independent and competent institution.134

B. The Critiques

In this Section, I respond to three critiques that my proposal might elicit. One critique focuses on the details of a potential state-borrowing program. Perhaps states would not borrow enough because this reform is too risky (what if Congress rescinds the debt forgiveness?) or perhaps my proposal is too generous (why not borrow to the maximum and dare Congress not to forgive?). As to this kind of critique, I have little to say. My main point is that there is a need for deeper thinking about state finances during a crisis. We almost backed into a reasonable system as a means of balancing moral hazard with procyclical cuts during the COVID-19 downturn, and it was a system that we have the institutional capacity to implement. That there could be a better system and that my thumbnail sketch could be improved upon is a given, and I hope to spur such discussions.

Another kind of critique rejects one of the semi-implicit premises of this Essay’s argument and views its proposal as misguided. Specifically, one might believe that moral hazard is such a serious problem that state shortsightedness would overwhelm the benefits of preventing procyclical cuts. I know of no theoretical or empirical warrant to justify such confidence that overborrowing is going to do more harm than underborrowing, especially when the proposed framework is attempting to achieve a balance. Certainly, as discussed above, the experience of the early twenty-first century suggests there is real harm in underborrowing, hence the move in the literature to second-generation fiscal rules and beyond.

A third critique challenges this reform as inconsistent with the American tradition of federalism and public finance. It asks, wouldn’t this ban be an unprecedented intrusion into state fiscal autonomy by entangling federal law and bureaucrats with state and local governments?

Such a critique is misguided in at least two ways. First, it misconstrues how the current system works. Federal law and federal bureaucrats are deeply involved in state and local finance and operations in numerous ways.135 For example, the federal government has required states to show a “maintenance of effort” in order to receive additional support for Medicaid during the pandemic,136 as well as educational support.137 The federal tax law regularly requires states to conform to federal rules, such as the calculation of “cash flow deficit[s],” an amount that sets an upper limit on short-term borrowing.138 Federal securities law similarly constrains states, sometimes occasioning deep dives into their debt positions and practices to ensure accurate disclosure.139

Or consider support for public transportation. As we saw above, ARPA, like the CARES Act, provided more funding using preexisting formulas for grants. These grants come with substantial rules, require states to make certifications as to the use of funds, and submit to audits after the fact.140 At least some federal-transportation grants can contribute up to 80% of the cost for a capital project,141 which means that it is likely that the local transit agency borrowed on the municipal market for its share of the project and thereby subjected itself to the federal tax and securities rules just discussed.

These examples are not all crisis-related. Rather, they reflect the general way that state and local governments perform their financial functions in our federal system. In a crisis, as noted above, state and local governments typically engage even more deeply with federal guidance for two reasons. First, new federal programs set up during a crisis involve a lot of work on the part of state and local officials to properly uptake the federal funds. Second, because the size, timing, and nature of federal aid is uncertain, states must wait and see what the federal government will do during crises. This waiting on tenterhooks is itself an (unhelpful) form of federal control.

On the part of the states, they already engage in many maneuvers to soften the impact of balanced-budget rules.142 In short, there is not a current practice of completely rigid hard budget constraints, though that is not to say the constraints as they are do not have teeth. If the constraints were not significant, excessive cuts during a downturn would not be a problem.

The critique based on the tradition and practice of American federalism related to hard-budget constrains is thus flawed on principle. A more precise, if leading, question to be asked about the design of our current system would start by acknowledging the significant magnitude of current federal-state entanglements, though generally ad hoc and accreted over time.It would also acknowledge that,, consistent with the theory motivating second-generation fiscal rules, that states are straining against balanced budget rules in an ad hoc way and also getting considerable federal support.Once these facts are accepted, then the question is would it not be better to have a thoughtful consolidated approach, even if only so that approach can be deliberately tweaked if it fails in some way?

I think the answer to this more precise question is “yes,” which brings me to the theoretical response to the objection on federalism grounds. It is tempting to count each connection between federal and state finance as a further blow to state autonomy. This interpretation strikes similarities to a sullen teenager’s view of autonomy, in which every rule is a restriction. But that view is mistaken. While certain rules undermine autonomy (for example, no going out after 10 PM), others enable autonomy. Certain limitations enable some choices while cutting off others.If one’s goal is to have the kind of meal that can only be prepared with time and care, then one must also submit to the limitation of ordering what is on the menu.143

The same analysis is true for the federal-state relationship. Numerous federal rules constrain state-revenue autonomy.144 But while some are purely limiting, others arguably enhance autonomy. Compare the Mobile Telecommunications Sourcing Act (MTSA)145 with the Internet Tax Freedom Act (ITFA).146 Both of these bills constrain states by imposing taxes. While the ITFA forbids certain taxes on internet access,147 the MTSA simply sets ground rules for states to follow when taxing mobile telecommunications.148 Further, the states supported MTSA (unlike the ITFA),149 illustrating how certain restrictions can be consistent with autonomy and even chosen.

The current state of emergency financing of general state services, where the states do not know what is coming, is debilitating. In some ways, this is worse than a simple prohibition like the ITFA. In the eyes of a teenager, which is more undermining, helicopter parenting or hard-to-predict toggling between helicopter parenting and free-range parenting? Some teenagers might say the latter—as might some states.

Conclusion

There are different ways of looking at ARPA and the states. On the one hand, it has provided an essential backstop, especially in light of the Delta variant surge. On the other hand, ARPA represents more of the same, with lessons still not learned from the Great Recession. Specifically, the series of ad hoc federal interventions could have been far more effective if executed systematically.150 This Essay has argued that ARPA has done enormous good and is not simply another ad hoc measure. Rather, aspects of ARPA and the preceding relief bills contained novel structural elements that could and should be built upon. Thanks to the MLF, the federal government now has some experience lending to states. Thanks to ARPA’s general support for the states and its clawback provision, the federal government now has some experience in supporting general state services—but not in a way that rewards profligacy. This new learning should be institutionalized into a new federal fiscal system that no longer overcorrects for the danger of moral hazard on the part of the states.

Professor, UC Davis School of Law. Many thanks to Ellen Aprill, David Gamage, Gladriel Shobe, Tracy Gordon, Katie Pratt, Erin Scharff, Ted Seto, and all the participants at the Loyola L.A. Tax Policy Colloquium.