The Yale Law Journal


The American Rescue Plan and the Future of the Safety Net

28 Nov 2021

abstract. In the year preceding the American Rescue Plan (ARPA), the unemployment insurance system in many states collapsed, leaving many workers to wait weeks and months to receive benefits. Millions more were ineligible despite losing significant sources of earned income.
 This Essay examines the pressures that made the unemployment insurance (UI) system crumble and aims to sketch a way forward. UI cannot survive without much more extensive federal funding, and this Essay explores several design options.
 I also propose tentative answers to the data issues and moral hazard worries that have been major obstacles to supporting the millions who work in part-time and gig jobs.
 Finally, I examine the constitutional and budgetary obstacles to safety-net reform. For example, current federal budget rules perversely penalize efforts to enact “automatic stabilizers” built to respond immediately to future crises. In effect, these rules damage our economic future in the name of preserving it.


By dollars and popular awareness, two key pillars of the American Rescue Plan (ARPA) are continuations of the Economic Impact Payments (better known as “the Checks”) and the unemployment insurance (UI) expansions begun in 2020.1 Both began in the CARES Act, as part of the national response to the massive workplace disruptions caused by the COVID-19 pandemic.2 But they are a mismatched pair. Where UI relief is elaborately targeted, capped, and constrained by an array of eligibility and filing requirements, the Checks were indiscriminately distributed to households earning less than $150,000.3 Behind the scenes as the ARPA was being drafted, though, it seemed that these two facts were tightly related. Americans needed the Checks precisely because UI—long intended to be the nation’s bulwark against this exact kind of widespread work interruption—had failed to reach many needy families.4

UI’s failure wasn’t an accident or an oversight. As Florida Governor Ron DeSantis observed of his state’s UI system, unemployment insurance in the United States is designed with “pointless roadblocks” to frustrate people so that they will abandon their claims.5 Known as the “ordeal” mechanism, this strategy is recommended by some economists as a way of supposedly targeting safety-net programs to those who need them most.6 The government’s choice to deliver safety-net programs jointly by states and the federal government, through our system of cooperative federalism, produces similar frustrations.7 Indeed, ordeals and federalism are interrelated: ordeals are not unique to cooperative programs, but the incentives and structures of cooperative federalism allow them to thrive.

In many ways, then, the Checks were a rebuke to our entire system of antipoverty law. In their vast scale, simplicity, and rapidity, they demonstrated the unique capabilities of an energized national government. Their perceived necessity demonstrated the limited capabilities of UI, the alternative path to delivering aid that we have largely chosen to follow until now.

This Essay explains how we arrived here and where our experience with the ARPA suggests we should go next. In Part I, I build upon foundational work by David Super and others to summarize the pathologies of how the United States delivers aid during recessions. Using unemployment insurance as a case study, I explain the ways in which state financial and political incentives drive program design. Since UI benefits are paid for largely by businesses and result in greater worker bargaining power, there is now a fierce race to the bottom in which states compete to slash UI.

Part II argues that the Checks were made necessary by our reliance on cooperative-federalism programs whose basic design makes them poorly suited to respond to recessions. While it was remarkable that the Internal Revenue Service (IRS) was able to deliver checks to millions in a relatively short period of time8 (all while its workforce was operating mostly remotely), it is neither desirable nor politically sustainable to expect the federal government to implement a wholly new federal poverty-relief program in every recession.

What we need instead are recession-proof safety-net policies. Step one is figuring out what recession proofing looks like. Proponents of a universal basic income (UBI) point to the Checks as an example of the promise of simple, lightly means-tested, ordeal-less programs. Still, we need to better develop the capacity to issue checks quickly and reliably. A robust federal-option payments system can deliver the Checks, bank the unbanked, and even offer recession-period government lending.

But checks and loans aren’t enough. Realistically, we are not going to have a UBI, even a temporary one, that comes close to replacing the income that most middle-income Americans lose when they are out of work (nor would most UBI supporters suggest payments that large). That’s what unemployment insurance does best.

Part III therefore aims to describe the future of UI. In earlier work, I outlined some plans for tinkering around the edges of the existing UI system, moderately tweaking incentives here and there.9 The ARPA experience shows those plans were not ambitious enough. The fundamental problem of unemployment insurance is state control and state fiscal incentives for maladministration.

The solution is federally funded UI. Workers and advocates have developed a comprehensive set of substantive reforms needed to restore UI’s proper operations.10 Here, I emphasize the importance of federal fiscal support and outline several alternative paths to removing states’ financial incentives to cut benefits. I also offer some tentative takes on the hardest modern problem of UI, which no world government has yet really cracked: how to fund and deliver income protection for the modern gig workforce. The Pandemic Unemployment Assistance (PUA) program, which for the first time offered U.S. aid to some part-time and gig workers excluded from their state’s UI eligibility rules, has promise. But it also has important cracks that need caulking.

Finally, in Part IV, I examine some of the major legal obstacles to UI reform. Some of these are constitutional limits on Congress’s power to make conditional offers to states. The Supreme Court has pushed the law into a corner, where serious reforms to cooperative federalism are so hemmed in by doctrinal developments that cooperation no longer offers a viable path to the nation’s biggest problems.

Other barriers derive from obscure but important budget rules, such as the Senate’s “Byrd Rule,” which is poorly designed for coping with the threat of recessions.11 Even though commentators agree that it is optimal for governments to borrow during recessions and repay later in good times, modern budget rules do not allow them to do so. An automatic-stabilizer system will be “scored” by the Congressional Budget Office as losing money. Congress must therefore raise taxes in advance to pay for any safety-net improvements. This system further compounds the underlying political-economy challenge. Budget rules intended to build a fiscally responsible future should facilitate, not block, sound recession planning.

Though unemployment insurance is my case study, many of these lessons generalize to other crisis-fighting tools, such as health, housing, and nutrition assistance. To avoid relying on American Rescue Plans to respond to future public-health, climate, or financial emergencies—or worse, a situation where an ARPA is needed but the government is unable to pass one—we need more forward-looking legislation, along with constitutional limits and budget rules that give such legislation a realistic chance of success.

I. the recession ordeal

Recessions are key moments for government. Even for those who believe government’s role in the economy should be narrowed to the clearest of private-market failings, recessions offer a compelling case for government intervention.12 Most recessions pose a classic collective-action problem. As demand for goods and services falls, merchants lay off their workers, which in turn puts further downward pressure on demand, and so on. To reverse this spiral, customers need to work together to all spend at once. That isn’t possible, of course, so instead government policies that have the effect of stimulating and encouraging spending must take the place of coordinated individual actions.

A bit more controversially, recessions also increase the urgency of social insurance.13 It’s difficult to insure against job loss, because moral hazard and adverse selection would typically make the insurance contract prohibitively expensive.14 Private job-loss insurers would also have to maintain vast reserves to cover all the claims that would roll in during deep recessions. Private savings cannot easily substitute for insurance, either. Even for families with the capacity to save up for hard times, recessions pose unusual strain: as jobs become harder to find, workers face longer periods with no paychecks, and many of the people they might turn to for support, such as parents and siblings, may be out of work, too.

Unfortunately, American safety-net programs don’t reflect this well-settled understanding. Instead, most of them are built in ways that tend to systematically undermine the delivery of benefits during recessions. Most importantly, as David Super has explained, the cooperative-federalism architecture of safety-net programs makes them particularly vulnerable to economic downturns.15 We have turned over a lot of the responsibility for paying for financial support to states.16 But state budgets are much more procyclical than the federal budget: while strong during boom times, they crash during recessions.17 States are under fiscal pressure to fire workers and slash benefits precisely when the economy needs the exact opposite.18

Those pressures have been especially evident in the unemployment-insurance
system.19 UI is one of the earliest major cooperative-federalism programs.20 The federal government sets out some general ground rules, but states provide most of the money and establish nearly all of the policy details.21 In states that fail to follow the scant federal rules, federal taxes on employers in the state rise from $42 per employee to $420.22 Though federal taxes fund most of states’ administrative costs, each state establishes its own system for taking in individual claims, assessing them, and paying out approved benefits. Most states’ 2020 administrative performance could only be described as disastrous.23 Frustrated claimants waited months for their claims to be processed, unable to reach human administrators, as computer systems crashed.24

In part, the 2020 UI application ordeal was no accident, but instead the product of deliberate efforts to reduce the desirability of UI benefits. Like any insurance mechanism, UI could create moral hazard by potentially discouraging workers from returning to work when they could instead collect benefits.25 Empirically, several modern studies show this fear to be greatly overstated.26 It remains politically potent, though. And economic theory suggests that one way of reducing the supposed moral hazard of social-insurance programs is through the “ordeal.”27 Ordeals are designed to function as “costly screens” that sort the neediest applicants, who are presumably willing to pay a hassle cost to access benefits, from the less needy, who their designers assume are not.28

Whether or not it makes sense in normal times, the economic concept of the ordeal is incoherent during recessions.29 Again, during downturns, governments want to put spending money in people’s hands, because spending produces spillover benefits for the whole economy. Applicants do not internalize this broader economic gain, meaning that ordeals turn benefits away from those whom society wants to have them. Just as important, some moral hazard can actually be a good thing during downturns—that is, safety-net programs can give workers the freedom to turn down low-paying or dangerous jobs.30 In the long run, that freedom pays off for the economy. A good portion of the slow recovery from the Great Recession can be tied to the fact that workers took worse or lower-paying jobs, so that their next job was also less desirable, and so on.31

In short, UI’s ordeal is completely backwards. The optimal screen is much more forgiving when the benefits of spending are higher and the dangers of moral hazard are lower. Yet during recessions, state screens have instead become more demanding. Waves of new applicants and strained state budgets mean that the hassles and delays of applying for UI rise, instead of ease, at the moment when states should be aggressively enrolling beneficiaries.32

UI’s cooperative structure allows this bad design to persist.33 On the benefits side, there are few reasons for state officials to care about delivering effective UI. State economies are interdependent, so states have limited incentives to care about the lost spillover benefits their administrative failures may have on their neighbors.34 Unemployed workers are a case study in the political economy of neglect: they have few resources, they are scattered widely, and their needs are temporary. All these factors make it hard for them to organize for political success.

At the same time, there is a race to the bottom on the cost side, as state employers systematically prefer a narrow and ineffective UI. Reliable, adequate UI increases worker bargaining power.35 Federal law obliges states to fund their UI programs through a payroll tax on employers, so that states that pay out more in benefits must set higher tax levels.36 Employers respond to these facts by lobbying for lower payouts, and by threatening to shift jobs to states that will heed their demands.37

These incentives ultimately have excluded most workers from the unemployment-insurance system. For example, in Southern and Mountain states in 2019, fewer than one in six separated workers received UI benefits, often because they worked only part-time.38 That figure doesn’t even include many self-employed individuals, such as gig and domestic workers, who were not eligible for benefits at all. Those left out were disproportionately women, particularly Black and brown women.39 And already in the months since the ARPA, states are cutting deeper, with some proposing to slash the duration of benefits to as little as twelve weeks.40

Though federal law offers few rules for state UI, it does impose some U.S. Department of Labor oversight of whether states adequately administer whatever benefits they offer.41 This oversight plainly failed in 2020. Federal review can identify states that slash benefits to absurdly low levels, but it cannot as easily identify states where a series of tiny administrative obstacles have piled up to almost unscalable heights.

II. time for the checks?

The failings of the unemployment system in 2020 set the stage for the ARPA. It was evident in early 2020 that UI as it existed at the beginning of the COVID-19 pandemic would not be enough.42 Recognizing that state plans did not reach the country’s millions of gig and self-employed workers, Congress created Pandemic Unemployment Assistance (PUA).43 PUA made UI benefits available, under prevailing state rules, for individuals who would otherwise have been ineligible because they lacked traditional wages. Similarly, recognizing that state benefits levels were grossly inadequate for many households in traditional work arrangements, Congress also authorized the Federal Pandemic Unemployment Compensation (FPUC) program, which added a flat amount on top of the benefit award calculated through state rules.44

Though these expansions provided vital assistance to millions, they still left huge gaps. Millions of workers could not claim the generous FPUC top-ups because they could not satisfy their state’s requirements, either legally or simply because they couldn’t navigate the process of demonstrating their eligibility.45 Millions of other self-employed individuals similarly couldn’t obtain PUA, whether due to administrative barriers or technical restrictions, such as state rules requiring applicants to show that they were “available for work.”46

These gaps helped make the case for the Checks. In negotiations over the design of the ARPA, advocates argued that relying on the broken state UI system to sustain American workers through the spring and summer pandemic months would leave too many destitute, or force them back to work in unsafe conditions.47 To ensure that the Checks reached everyone, they were made essentially universal below an income threshold.48 Implicitly, then, the design of the Checks recognized that excessive screening for “need,” and efforts to detect and prevent moral hazard, would just reintroduce the very gaps the UI system had left: gaps misbegotten by ordeals of hassle and administrative waste.

The Checks were the right policy for their moment, but should they be a key piece of economic recovery in future recessions? In many respects, the argument for the Checks is not much different from the argument for UBI.49 If anything, the case for the Checks during recessions might be stronger, since as I’ve just noted in the UI context,50 the moral-hazard concerns sometimes raised about UBI are not as significant for a program that will only be deployed during recessions.51

Certainly, there are weaknesses in the Checks program that Congress could improve upon. We should make them more like “automatic stabilizers,” triggered by economic conditions, rather than depending on Congress to get the timing and amount of cash payments right in a moment of crisis.52

A more formal and reliable payment system would also help to speed delivery and reduce hassle costs. Getting cash to “unbanked” households often means that a significant fraction of the government’s support is instead captured by financial intermediaries, such as check-cashing operations.53 And basic but important components of direct payments, such as ensuring the correct identity and address of the payee, are hard to set up on short notice.54 Both of these flaws undermine the government’s ability to deliver timely relief to the neediest households.

A possible route forward, and one that offers a number of other advantages, is a much larger set of “public option” banking opportunities.55 If every household had, say, an electronic-funds account at the Federal Reserve, distributing the Checks would be much easier.56 Public banking would also facilitate a public-option credit program.57 Expanding access to affordable credit would be a huge boon to many families who now rely on lenders at the fringes of the legal market.58

Nevertheless, checks and loans are not enough. Sustained unemployment brings individuals and families tens of thousands of dollars of lost income, an order of magnitude larger than what the Checks provide. Realistically, the United States will not enact a UBI large enough to replace a year of lost wages.59 What does the real future of income security look like?

III. towards a federal unemployment-insurance system

Genuine preparation for the next recession will require a functioning UI system, and the path there is long. As this Essay has described, through state neglect and maladministration, unemployment insurance is legally available only to a small fraction of separated workers, pays them too little, and is hard to claim even for those who are entitled to it. Yet the state response to the remarkable failings evident at the time of the ARPA has been to make more cuts. This Part proposes a UI regime that is fully federal, or at least supported mostly through federal dollars.

A. The Basics

UI needs a comprehensive set of federal interventions. Federal law should at least guarantee universal standards for benefits and administrative adequacy. No component of the safety net can stand on its own. If states have incentives to minimize UI benefits, they have many options. They can narrow who can claim benefits, reduce the amount of their benefits, and impose ordeals that lower take up. Federal guarantees have to set floors on all three of these measures to be effective. More than that, federal law should also try to align state incentives with the national interest by adopting a more rational financing system.

A recent report from a coalition of antipoverty organizations—built from grassroots engagement with workers, polished by UI experts, and published by the Economic Policy Institute—laid out the essentials of what nationwide guarantees should ensure.60 Benefits in the United States are almost shockingly low compared to most of the developed world.61 In 2019, for instance, the average state made only twenty-eight percent of separated workers eligible for UI;62 in Florida, a remarkable eight out of nine workers were ineligible.63 On average, before taxes, UI replaces just thirty-eight percent of the weekly wage a worker was earning.64 Benefits last for as little as twelve weeks in some states, as compared to a year or more in many peer nations.65 All of these benefits should be brought up to modern standards as the experts propose.

In the case of administration, written guarantees alone are probably not enough, and there needs to be a substantial direct federal role in administering UI claims. The main problem is that guarantees are hard to enforce. Though states could be subject to standards for certain objective data, such as the average time it takes to begin paying benefits, many potential impediments to benefit uptake are essentially impossible to measure.66 How do we assess, for instance, the share of potentially eligible workers who do not apply because they are confused by the state’s rules and procedures, or vexed by its antiquated phone system? To guarantee effective administration of UI systems, the federal government likely has to provide them itself, such as by requiring states to use a streamlined national portal for beneficiaries to use when they apply or upload relevant materials.

All of these details should respond automatically to economic conditions, such as by expanding benefits and administrative capacity in recessions. Otherwise, we will be trapped in a spiral of state cuts and haphazard federal responses. The more the federal government steps in with dramatic boosts like the CARES Act and the ARPA, the less likely states are to provide adequate benefits themselves.67 As a result, even more significant federal responses in the future will be necessary. And at some point, due to politics or simply legislative inertia, Congress might fail to deliver, leaving the safety net threadbare.68

B. The Challenge of the “Self-Employed”: Lessons from the PUA

One challenge in making UI available to a larger share of workers is that traditional UI does not offer benefits to individuals who are self-employed.69 That was the gap that led to the adoption of the PUA program, which extended traditional UI benefits to contractors and gig workers. Traditional UI provided nothing to the millions of gig workers who did not qualify as employees under applicable state law.70 Employers pay UI taxes on employee wages, but not on payments to self-employed contractors.71 UI’s financing system thus encourages businesses to try to characterize their workers as contractors, whether through legal arguments or real changes to work processes.72

Simply extending some version of PUA may not be the right answer. Historically, the United States did not insure the self-employed because of some combination of moral hazard and data problems.73 In general, only workers who lose work through no fault of their own can claim UI benefits, on the theory that otherwise it would be too tempting for some to quit their jobs and claim UI.74 Whether or not one agrees that this is a serious concern, it plainly was not much of an issue during the COVID-19 pandemic, when it was obvious that most gig and domestic workers, visiting nurses, and the like were losing work because of forces beyond their control.75 Extending PUA to normal times would require a more serious engagement with the moral-hazard debate.

The UI system does have other mechanisms for restraining potential moral hazard, but it was these mechanisms where the data issues made insuring the self-employed difficult. To collect benefits, an applicant must continually show that they are looking for and available for work.76 If these systems operate effectively, they should mitigate any moral-hazard concerns, since a worker who is intentionally remaining at home would not be able to collect UI.77 In practice, to verify worker claims that they applied for and were available for work, a state UI office usually contacts the employers the worker says she applied to, and asks whether the beneficiary was offered a job.78 If she declined an offer, she then must explain why she should still be considered “available for” work.79 These verification steps are often not practical for, say, a self-employed handyman or home-care worker whose work consists of multiple part-time jobs that are arranged informally.80

Determining how much income a self-employed individual has actually lost poses another challenge. UI systems generally determine benefits based on prior earnings.81 Of course, benefits are only open to those who are out of work (or, in some cases, those who lost one job and returned to work part-time).82 Policy makers have struggled with how to reliably verify earnings for applicants who do not have a relatively disinterested third party (their employer) to report compensation.83 For instance, a self-employed individual paid mostly in cash could potentially exaggerate their prior earnings, understate their current earnings, or both.

Technology may be able to solve many of these problems, as the PUA experience suggests. PUA allowed workers to establish their prior earnings by submitting a recent tax return—on which they presumably would have had little incentive to inflate their income.84 Workers then attested to or used quarterly tax filings to establish their lost income, with the expectation that these claims would then have to be verified when they filed their year-end tax return.85 The attestation system was probably imperfect; reports suggested significant fraud in the PUA system, although much of it may have been the result of organized criminal enterprises rather than opportunistic

Many of these steps could be automated for greater ease and reliability.87 With some small changes to tax-privacy law, tax information could be shared automatically with UI benefit offices.88 A significant fraction of so-called contractors receive regular payments through large business intermediaries, such as gig platforms or temp agencies.89 ARPA now requires most platform operators to report income for workers earning at least $600 through the platform, so that more workers will now have verifiable tax data on their earnings.90 With little marginal cost, electronic-payment reporting could be made in real time, or at least biweekly, providing an ongoing snapshot of worker earnings.

Technology could also help workers establish that they are available for work, even if they lack a traditional employer who can confirm their status. A platform operator or a temp agency can verify whether their workers are accepting new gig offers. Workers who are not would then have to provide a good reason to retain their UI benefits, as in the traditional system.

That does leave the question of what to do with more traditional self-employed individuals, ranging from handymen to sole-proprietor management consultants. Payment verification is still feasible. Workers could voluntarily enroll in a reporting system through their checking account, along the lines of a U.S. Treasury proposal to require banks to report account inflows and outflows for certain customers.91 Establishing availability for work is a tougher nut to crack. One approach would be to recognize that the moral-hazard risk is higher for this population, and to offer a lower replacement rate (i.e., make the benefits less appealing), or impose some kind of individual-worker-experience rating so that claiming benefits results in higher taxes later.92 Since moral hazard is not usually a significant concern during recessions, these added safeguards could be waived in downturns. Alternately, we could grant full benefits, but phase them out at higher household incomes—in effect accepting some moral-hazard costs in exchange for being able to redistribute money to the neediest families.93

A Job-Seeker’s Allowance (JSA) would also help fill gaps for many gig and self-employed workers. Current proposals for a JSA would grant a modest weekly benefit, around $200 per week, to anyone who can show they are looking for work, regardless of past earnings.94 Although primarily intended for recent graduates, parents returning to the workforce, and the like, a JSA could also serve the self-employed fairly well. Since a JSA would not require proof of past wages, the self-employed applicant would not have to worry about documenting her cash earnings. And since the self-employed applicant has to show that she is available for work (albeit work through a traditional employer), moral-hazard concerns are minimized.95

These changes should come with financing reforms. Adding millions of beneficiaries to the UI system will pinch existing budgets. It would be unfair to tax traditional employers to pay for benefits for gig workers, whose “employers” contribute nothing. Indeed, to the extent that the absence of UI benefits currently discourages some gig or contractor work, expanding eligibility might lead some workforces to shift away from the relative security and benefits of employee status.96

The Economic Policy Institute (EPI) Report proposes a two-pronged solution. First, it recommends imposing UI tax on the payments that large workplaces make to contractors, as if those contractors were employees.97 Intermediaries, such as internet platforms, would have to count all their payees as their own contractors. In addition, to ensure that smaller employers do not use legal shenanigans to escape UI tax, the EPI report proposes a nationally uniform, simplified legal test for determining who counts as an employee, with a strong presumption that most workers should in fact be treated as employees.98

C. How to Pay for It

None of these federal guarantees can be fully effective if states retain a strong fiscal incentive to minimize benefits. Even if federal law aims to ensure nationwide minimum benefits and obliges states to make use of some federal administrative tools, there are yet other paths for states to drag their feet,99 such as by appointing administrative-law judges who will reject a high proportion of claims. For this reason, some advocates have argued that we should simply have a national UI system.100 While I offer some reasons to retain state involvement in the next Part, I don’t particularly want to argue against nationalization.

Instead, I’ll simply observe that federal action could mostly eliminate states’ incentives to compete on UI tax rates, which in turn would soften (but probably not eliminate) political pressure to hold down benefits. Most straightforwardly, the federal government could fully fund UI, while leaving some programmatic and administrative details to states. That, after all, is exactly what the ARPA did, and closely resembles the everyday operations of both the Supplemental Nutrition Assistance Program (SNAP) and certain unemployment benefits for former federal workers.101

A fiscally similar approach that may sound less radical to some politicians would be to provide extensive federal guarantees for state UI debts. Currently, when states spend more in UI benefits than they have saved, they can borrow from a federal fund.102 My proposal is to write a statute that would automatically wipe these debts clean in states that are hit hard by recessions (for example, in states where unemployment exceeds 120% of its recent historic levels). In effect, states would know that they do not have to save up for times of serious need and can instead rely on federal funding in any crisis, allowing for a much lower tax rate.

Some UI experts have instead favored “forward funding” incentives, in which the federal government tries to encourage states to do a better job building up their savings accounts.103 These efforts have failed.104 The typical incentive, which promises states future benefits in exchange for savings today, is ineffective. If state officials cared about the future of their UI program, they would already be saving.105 I have proposed alternative structures in which state officials are incentivized today in order to commit to taking fiscally responsible action later.106 Those structures are more theoretically promising, but they remain untested and require a federal government with the tools and political will to hold the states to their commitments. That has been notably absent in the UI space.

A third alternative is making state taxes essentially costless for states to impose. For example, until 2001, most states imposed a substantial estate tax because the federal government provided a dollar-for-dollar credit against federal estate taxes for every dollar paid to the state.107 In effect, the state’s tax would have cost an estate zero dollars, and instead simply moved money from the federal treasury to state coffers. A similar UI design might impose a substantial federal tax on wages (and contractor payments), with a one-hundred percent credit for taxes paid to a state. States would have no incentive to compete on tax rates since lowering state rates would not save their businesses any money.108

IV. implementing the new unemployment insurance

Implementing a set of reforms this sweeping will naturally face serious obstacles, with legislative inertia probably being the most formidable. In this Part, I analyze some additional legal and procedural hurdles, including Supreme Court doctrine and congressional budget rules. These barriers are not unique to UI, and I’ll also use the discussion to reflect on the need for reforming many of them to clear the way for real national readiness for the next crises.

A. Cooperation or Federal Control?

First, Congress must decide whether to encourage states to meet its new universal guarantees or instead transfer full control of UI to a federal agency. Historically, the Roosevelt Administration chose UI’s cooperative federalism in part to avoid perceived constitutional constraints.109 Ironically though, today there are compelling policy reasons to continue state-federal cooperation, even as modern constitutional doctrine makes nationalization the safer legal strategy.

The benefits of cooperation are familiar to federalism scholars. If federal involvement can remove or mitigate incentives to race to the bottom, sharing policy authority jointly between governments offers several key advantages. The federal government is better at raising money, while local governments are often better at spending it.110 Local officials typically have better information about the highest-value spending priorities.111 Coordinated spending can also mitigate the crowding-out problem where a local government might be tempted to slash its own transfers to a group supported by federal dollars.112 While the benefits of local experimentation are often oversold, guided local experiments in which a central planner helps steer and assess a series of policy variations can offer powerful evidence about the most effective options.113

The Supreme Court’s decision in National Federation of Independent Businesses (NFIB) v. Sebelius, the Obamacare case, puts all those benefits in jeopardy, at least for large-scale cooperative projects.114 Although the logic of the NFIB decision was fairly opaque, as best as commentators can tell, NFIB held that “new” federal conditions on transfers to states are invalid if the transfers are large enough to “coerce” a state by requiring them to “participate in a separate and independent program.”115 Though the Supreme Court held that UI was not “coercive” in 1937,116 it is unclear how committed the current Court would be to that eighty-year-old precedent.117 To remove the race-to-the-bottom incentive, Congress most likely has to take on a much larger share of UI costs, as I described above in Section III.C, which could make the program appear more coercive. While Congress could instead try to move states with more gentle nudges, recent evidence suggests that such nudges would often be ineffective. Nearly half the states, for instance, cut off federal aid to state UI recipients even though the federal government was paying one-hundred percent of the cost of those benefits.118 Much more modest financial incentives are even less likely to prompt states to strengthen UI.

This is the familiar paradox of cooperative-federalism doctrine. When the Court steps in to “protect” states from federal commandeering or coercion, it pushes Congress towards two extreme options: abandon its plans or make them wholly federal with little meaningful policy role for states.119 Sometimes this gambit will foil legislative efforts, but sometimes it will instead move us towards nationalization. Depending on which prevails more often, efforts to protect state autonomy may instead result in less.

Congress could mitigate this risk by making its new transfers unconditional and relying instead on the conditional taxes already upheld in 1937,120 though those would require some revamping. Again, if states fail to stick to federal requirements, taxes on state employers shoot from $42 to $420 per worker. The difficulty is that the Department of Labor apparently views this penalty as so dire that they have never been willing to impose it, making it a fairly weak deterrent.121

Smaller penalties might actually be more effective because the Department of Labor would be more likely to impose them.122 For instance, the UI statute might direct the Department of Labor to calculate a state’s savings from failing to comply with a federal rule and set the tax penalty at 110% of that amount. It is true that many states recently accepted an even larger penalty when they turned away fully funded benefits. But that decision was likely driven by state employers (who else in the state was excited about turning away free money?) and that is exactly the constituency that would have an interest in avoiding tax penalties.123 To further increase the credibility of the threat, private parties could be allowed to commence penalty actions themselves, or at least petition the Department of Labor to do so, with the Department obliged to answer in court if it declines a nonfrivolous petition.

This alternative both relies on and highlights a fundamental incoherence of the Supreme Court’s “coercive” spending doctrine. The Court focuses its coercion analysis on the share of a state’s budget threatened by a federal condition.124 But as a political-economy matter, threats to the general budget are relatively ineffective tools, given that the budget is a public good shared by all legislators.125 Effective suasion imposes private costs on individual lawmakers.126 Nations in trade wars know this: when they impose tariffs on each other, they aim at concentrated, politically powerful industries that are most apt to lobby their governments.127 Certainly, the coercion doctrine could evolve to take account of these realities. But doing so would just reveal the hollowness of the Court’s notion of coercion: ultimately, a “coerced” legislator is just one who does not like the political consequences of her decision. There is no obvious reason the Constitution protects state legislators from making hard choices or prohibits them from taking a deal they think is fair.128

For now, though, the law of cooperative federalism seems to favor a federal takeover of most UI policymaking. Any meaningful incentives Congress offers states to stick to its new universal guarantees might be challenged as coercive. Moderate tax penalties on state employers have the best chance of threading the Court’s current doctrinal needles while still offering some chance of efficacy, but none of the cooperative options are risk-free. This is not to say that the federal government must also supplant the thousands of experienced, unionized state workers who now administer UI. The United States could, for instance, contract with state workforces for the delivery and administration of benefits.

B. Budget Rules and Automatic Stabilizers Don’t (Yet) Mix

Budget rules pose another obstacle to UI reform—or any other expensive policy that economic conditions would automatically trigger. Many federal budget rules are intended to tie Congress’s hands, preventing it from impoverishing the future to benefit the present.129 That makes some sense: legislators with limited terms in office are unlikely to prioritize later generations over present voters.130 Instead, though, budget rules end up blocking a forward-looking Congress from planning for tomorrow’s crises.

Some key examples are the “Byrd Rule,” a component of the Congressional Budget Act of 1974, and “pay as you go” (PAYGO) rules in the House and Senate. Under the Byrd Rule, provisions that will increase the federal deficit more than ten years from their enactment must receive sixty votes to pass in the Senate, instead of fifty.131 PAYGO bars each House from adopting legislation that would on net lose money under its respective accounting rules—although often that limit can be procedurally sidestepped.132

These constraints can offer significant roadblocks to any automatic stabilizer. In modern recessions, governments have largely borrowed to pay for economic
stimulus.133 Budget rules instead require that automatic stimulus must be paid for in advance. When the Congressional Budget Office (CBO) estimates the costs of recession-fighting legislation, it assumes that each future year has some chance of recession.134 If a reformed UI bill would result in an extra $500 billion in spending in a recession year, and recessions occur once every ten years on average, the CBO will score the bill as costing $50 billion per year. If so, then to escape PAYGO or the Byrd rule, the bill must also include at least $50 billion per year in new revenues. Needless to say, having to adopt a large new tax together with any safety-net program greatly increases the difficulty of enactment. In effect, budget rules have made it harder for Congress to plan for the future, in the name of protecting the future.

One might instead argue that budgetary rules are a way of saving for the future. But it turns out that they don’t do so effectively. Budget experts note that “lockboxes” or “rainy day funds” could also provide a mechanism for transferring money from good times to bad—if they actually worked.135 The difficulty is that it is usually easy for a legislature to override its own rules and withdraw from its rainy-day fund before the rains come.136 PAYGO and the Byrd Rule are even worse: they are basically lockboxes with no lock at all. They force Congress to raise money for automatic stabilizers in advance, but nothing stops Congress from spending that money on something else before the need for stabilization arrives.137

Automatic stabilizers should thus be exempt from PAYGO, the Byrd Rule, and similar provisions. Arguably, any long-run investments ought to be exempt: since Congress systematically underinvests in the future, institutional rules should systematically favor them.138 Borrowing to invest in, say, climate mitigation is a net win for the very future generations that the Byrd Rule is supposed to protect. The counterargument is that an exception so sweeping would swallow the rule. Congress could plausibly argue that most spending has at least some future returns. But automatic stabilizers, which by definition cannot come into being until some future event, do not pose this problem. Carving them out of budget rules does not risk a slippery slope.


The ARPA was an urgently needed emergency repair, but it very nearly did not happen. To make our safety net more resilient to future crises, we should build institutions that deliver aid without the need for massive federal legislative action. The UI system, in particular, is headed for a ditch, driven by state fiscal and political pressures. Congress should take state legislatures out from behind the wheel and implement a federal unemployment-insurance system. But successfully rebuilding UI and other systems for the future first requires rethinking key budget rules and accounting for constitutional constraints.