Reconsidering Expansion and Retrenchment in the American Welfare State

I. Introduction

It is a privilege to respond to Professor Samuel Bagenstos’s Frankel Lecture at the University of Houston. The Houston Law Review made an inspired choice in selecting him for this year’s lecture for several reasons. Professor Bagenstos is a leading public law scholar whose pathbreaking work has defined the legal academy’s study of disability law. Beyond his academic contributions, Professor Bagenstos has served at the highest levels of government, most recently as General Counsel at both the Office of Management and Budget and the Department of Health and Human Services, giving him a rare vantage point on fiscal policy, administrative implementation, and the challenges of governance during the COVID-19 pandemic. Finally, despite (or perhaps because of) the steady stream of controversies spilling out of Washington since President Trump returned to the White House, the legal academy has barely begun to engage with the significant changes brought by the One Big Beautiful Bill Act (OBBBA) just months ago.^[1] The timing of this lecture could not be better.

Professor Bagenstos argues that many of the COVID-era expansions of tax credits, health insurance, and other public benefits can be thought of as a national “move toward universal benefits.”^[2] For instance, as he rightly points out, by 2023, “the United States came the closest it has ever come to universal health care.”^[3] Prominent theories from political science would predict that these pandemic-prompted expansions of the American safety net would informally entrench themselves through policy feedback.^[4] Instead, as Bagenstos shows, these expansions expired and many of the programs were rolled back further under OBBBA.^[5] Bagenstos gives us an incisive account of the expansion legislation and the dramatic reversals. He then draws lessons about why the benefit expansions were not as “sticky” as political science would predict.^[6] Rather, he writes that the policymakers “who pressed for the various benefits expansions during COVID certainly bet on the durability of those expansions . . . [but] they made a very bad bet.”^[7] In his exploration of “what made the bet so bad,” he highlights the backlash from employers and the need to analyze not just the “material effects” of the policy, but the “social meaning” of the expansions.^[8]

In the four parts that follow, my response attempts to elaborate, reframe, complicate, and extend Professor Bagenstos’s analysis. I elaborate how the legal and bureaucratic structures of the COVID expansions undermined their universalist tendencies. I reframe Bagenstos’s characterization of the pandemic expansions as universalistic social policy with one that I think fits a bit better. In my view, many of the COVID expansions are more accurately characterized as economic stimulus blended with temporary public health protections. Disaggregating the legislation into two distinct deals should make us less surprised that the policies did not persist past the official public health emergency. Third, I want to complicate the story of how COVID-era expansion has given way to our current time of retrenchment with the help of two timeframes: the vanishingly short window during the 2024 election and a much longer perspective stretching back to the New Deal. Fourth and finally, I agree with Professor Bagenstos that OBBBA has the potential to dramatically reverse the trajectory of U.S. social policy. I want to offer a concurrence of how dramatic OBBBA’s impact could be, but for a different reason than he identifies, namely that OBBBA is best thought of as not just programmatic, but also systemic retrenchment.

II. Programmatic Federalism Undermines Legislative Universalism

Bagenstos places policy feedback at the center of his analysis. Benefits should have created durable political support, but the circuits of feedback did not hold. His account emphasizes employer backlash and shifting social meanings.^[9] Yet I question whether a kind of virtuous cycle of policy feedback that would move the United States towards universal benefits was likely to take root in the pandemic, given that some of the expansions he highlights (and others he leaves out) needed state and local government, as well as businesses, to implement them.

Policy feedback presumes straightforward transmission: government delivers benefits, beneficiaries experience those gains directly, they form new constituencies or bolster old ones, and political support hardens in such a way as to make it harder for opponents of the initial policy to make future cuts or return to the status quo ante.^[10] But COVID relief reached the public through a welfare state that Americans experience as fragmented, burdensome, and exclusionary. Many eligible households never received stimulus checks.^[11] Others gave up on unemployment insurance (UI) applications when antiquated state technology failed them.^[12] Even Medicaid’s continuous coverage rule guaranteed a painful unwinding once the public health declaration expired.^[13] In short, administrative burdens and our institutional design of devolved responsibility to state and local governments disrupted policy feedback before political support for these temporary extensions could take hold.

Part of the problem with legislating universalism in the United States is that Congress has hampered its own spending power. On Capitol Hill, budgetary rules in both chambers and antimajoritarian procedures in the Senate force Congress to rely on sunset clauses, reconciliation rules, and fiscal baselines that all but assure expiration.^[14] Reconciliation rules and pay-as-you-go requirements make it much more likely for Congress to craft legislation in such a way as to not create permanent increases in discretionary programs.^[15]

And even if Congress weren’t so constrained in fiscal policy, it is undermined by our system’s federalism. Congress legislates, but often state agencies deliver the benefits. That decentralization and devolution to state and local governments guarantees that even legislation approaching universalism will be blunted by the patchwork of public administration scattered among fifty states, D.C., five territories, 574 federally-recognized tribes, and countless counties, cities, and other municipalities.^[16] Take pandemic unemployment assistance (PUA), which Bagenstos admits included “significant administrative burdens” and that “state programs at times found themselves overwhelmed.”^[17] Several states, including Florida, Kentucky, Michigan, and New York, could not respond to the volume of applications for UI in the first months of the pandemic.^[18] New Jersey asked applicants to file online before 7:00 a.m. or after 10:00 p.m. so as not to overload the system.^[19]

Two other pandemic-related programs Bagenstos acknowledges, the Emergency Rental Assistance and Pandemic Electronic Benefit Transfer (EBT), are also worth considering. Bagenstos leaves them out, understandably so, since emergency relief programs are not his focus, but they are additional examples of how state and local implementation undermines federal expansion of benefits.^[20] For the Emergency Rental Assistance Program, 89% of that funding had not reached renters over a year into the pandemic.^[21] Or take Pandemic EBT—a program, built on the Supplemental Nutrition Assistance Program (SNAP), to help families experiencing food insecurity.^[22] Tennessee delivered the additional Pandemic EBT to households who were already enrolled in SNAP or the Temporary Assistance for Needy Families (TANF), but overlooked families who were not already enrolled in either program but still eligible for Pandemic EBT based on school meal eligibility.^[23] As a result, Tennessee’s mistake impacted nearly 400,000 children, delaying approximately $60 million in federal funds.^[24]

Nor should we assume states will willingly implement national directives and accept federal funding to expand programs. Such an assumption flies in the face of the COVID-era expansions. Let’s return to Bagenstos’s example of PUA.^[25] Twenty-six states opted out of the expanded unemployment assistance program before the federal government’s cutoff date of September 6, 2021.^[26] Even though these weekly enhanced benefits of $300 were financed by the federal government, states argued they needed to forego this funding because of labor shortages.^[27] Texas’s decision to cut off PUA in June 2021 meant the Lone Star State relinquished roughly $6.5 billion in federal funding.^[28] Nearly one million Texans were impacted, 767,830 of whom lost all UI benefits.^[29] And sometimes, the line between incompetence and willful failure is difficult to discern. In the case of Florida’s PUA experience, it is difficult to determine whether, like other states, it was simply overwhelmed by the sheer number of applicants or if the program was designed to fail.^[30]

The implementation of Pandemic EBT tells a similar story. Some states decided not to renew emergency declarations in 2021, which cut off SNAP recipients from receiving additional benefits. Twelve states also opted out of continuing Pandemic EBT, denying additional food assistance to children.^[31] Florida refused to apply for the summer 2021 Pandemic EBT funding for months, delaying an additional $820 million in federal funds for approximately 2.7 million children.^[32] And when Congress converted Pandemic EBT into Summer EBT to make sure kids do not go hungry when schools are closed, more states refused the federal funding.^[33] To date, thirteen states still refuse to participate in the program.^[34]

Incidentally, this is not to say that the federal government always does a better job than states. The debacle of healthcare.gov in 2013 suggests otherwise.^[35] But Professor Bagenstos is, in my view, on the strongest ground when he emphasizes the COVID expansions that did not depend on state and local implementation, like those of the Child Tax Credit (CTC) and the Earned Income Tax Credit (EITC).^[36] Maybe the stimulus checks also emerged relatively unscathed from implementation snafus, although I would suggest that the issues that plagued the Paycheck Protection Program suggest that even federal agencies cannot always be expected to implement legislative expansion in an evenhanded way.^[37]

Rather, my research and practice experience on state implementation of federal welfare programs make me deeply skeptical of social policy feedback working in the way political scientists have suggested, but perhaps for a different reason than the ones Professor Bagenstos offers. I do not doubt that Presidential administrations and Congress will, in times of crisis, have the impetus and the vision to expand benefits, but the patchwork of delivery, endemic to state and local implementation, will repeatedly undermine that vision. Universalism in the American welfare state, then, is an ever-receding horizon. Even if the President and Congress chart the course to that destination, we need hundreds of ships of state to row in the same direction. And past experience suggests we should not expect them all to arrive.

III. Disaggregating the Pandemic Expansion

Bagenstos describes the pandemic-era expansions as edging the U.S. welfare state closer to universalist principles.^[38] I read them differently. To show you what I mean, some periodization should help. Just as historians often refer to the First New Deal and the Second New Deal as distinct legislative packages, I think it is useful to conceive of and label the legislation passed by the 116th Congress and signed by President Trump as the First Pandemic Deal and the legislation passed by the 117th Congress and signed by President Biden as the Second Pandemic Deal.^[39]

By the First Pandemic Deal, I mean the four bills that passed in the first few months of the pandemic (the Coronavirus Preparedness and Response Supplemental Appropriations Act, the Families First Act, the Coronavirus Aid, Relief, and Economic Security (CARES) Act, and the Paycheck Protection Act), as well as the appropriations bill from the 2020 lame duck session.^[40] These five bills that Congress passed, and President Trump signed, were, to my mind, old-fashioned Keynesianism: countercyclical government spending to prop up an economy that looked like it was in freefall.^[41] In addition to mobilizing resources for the public health response, the First Pandemic Deal’s primary purpose was to stabilize aggregate demand: checks in the mail, expanded UI, and forgivable loans through the Paycheck Protection Program.^[42] These macroeconomic policies were intended to stabilize consumer confidence and demand, which would, in turn, prop up corporations, financial markets, and the global economy.^[43] I do not disagree with Bagenstos that they were extraordinary in scale,^[44] but I see them not as incipiently universalist, but rather conventional in aim. Members of Congress were not focused on, say, the persistence of child poverty, although the policies did lead to a historic drop in the latter.^[45] Rather, the 116th Congress was bent on not repeating the mistakes of the 111th in its response to the most recent previous financial crisis in 2009. The 111th Congress’s American Recovery and Reinvestment Act (ARRA) was too small to provide the macroeconomic boost to pull the country out of the worst economic downturn since the Great Depression.^[46] Members on both sides of the aisle, and President Trump were not going to make the same mistake.

The difference in legislative posture also stemmed from electoral timing. The 111th Congress took its seats while the financial crisis was spreading, worsening, and deepening.^[47] But members of Congress knew that none of them would face the electorate for nearly two years.^[48] By contrast, the 116th Congress was wrapping up its session when the pandemic hit, meaning it would legislate in the shadow of the 2020 presidential election.^[49] More Americans lost their jobs in 2020 than in any year since 1939, and members of Congress were worried they would lose theirs if they did not act swiftly and forcefully.^[50]

The Second Pandemic Deal—the American Rescue Plan Act (ARPA), the appropriations bills of the 117th Congress, and the relevant provisions of the Inflation Reduction Act—was admittedly more ambitious. Professor Bagenstos, in particular, is focused on ARPA,^[51] but I think we risk misreading the policy agenda of a different President and Congress embedded in the Second Pandemic Deal back into the purposes of the First Pandemic Deal. The previous Congress was willing to fund a massive stimulus package that included significant direct support to American households, but they were not intending to build a more generous welfare state. They just sought to forestall another Great Recession or Depression, for that matter.^[52]

When Congress had a chance to take a more permanent step towards a universal welfare state, they balked. Much of the pandemic-related spending was stripped from the Consolidated Appropriations Act of 2022.^[53] The next appropriations bill did provide additional funding for food assistance and delinked the Medicaid continuous coverage requirement from the COVID-19 public health emergency, required all states to provide twelve months of continuous eligibility for children in both Medicaid and Children’s Health Insurance Program (CHIP), extended federal funding for CHIP, and significantly increased Federal Medical Assistance Percentage for the territories.^[54]

But those pitched appropriations battles also need to be considered in light of the failure of the Biden Administration to pass its Build Back Better Act. That legislation passed in the House by seven votes in late November 2021, but it languished in the Senate. Eventually, the Senate passed the Inflation Reduction Act of 2022, but that legislation, while significant for its climate and energy provisions, was stripped of all the safety net proposals.^[55] As a result, the Biden Administration failed to secure in the Senate, various provisions the House had approved including universal preschool for all three and four-year-olds, a child-care cost cap of 7% of income for low-income and middle class households, four weeks of paid family and medical leave, and extensions of the expanded CTC and Affordable Care Act (ACA) subsidies.^[56]

By disaggregating the COVID-era expansions into two different legislative deals, I concede that my reframing is a better fit for some of the expansions on Bagenstos’s list, including the expansions of the CTC, the EITC, and UI, than others, such as paid family and medical leave and health insurance. But I take the examples of paid Family Medical Leave Act and Medicaid’s continuous coverage provisions to be as much a part of the public health response as they were part of a more generous welfare state. We would expect these kinds of health policies given the realities of the COVID crisis, especially in contrast to the Great Recession. And looking back at the COVID legislation that Professor Bagenstos highlights, it is hard not to be reminded (perhaps wistfully) of what looked like a coordinated, national public health response to try to both stem the transmission of a novel virus and make sure that people who did get sick from COVID also got care.

Seen this way, the pandemic expansions look less like a coherent move toward universalism than a pair of crisis-driven bargains, each reflecting the political and economic conditions of its moment. The First Pandemic Deal was stimulus under electoral pressure; the Second Pandemic Deal was public health protection with a renewed focus on the costs of care. But fiscal rules and politics on Capitol Hill constrained both bargains. Some members of Congress who voted for the Second Pandemic Deal hoped to build a durable, expanded welfare state, but I doubt we can ascribe such intentions to those who voted for the First. That disaggregation, in turn, may help explain why the expansions dissipated once the pandemic ebbed, and why it may be a mistake to read such aspirations into this spate of legislation.

IV. Making Sense of How the United States Went from ARPA to OBBBA

As I explained above, the pandemic expansions were less a step toward a new welfare state than a series of stopgap measures tethered to the rhythms of macroeconomic stabilization. They revealed our fiscal state’s capacity in crisis, but they did not remake the political economy of work or the structure of the safety net. Even if they did, as I suggest in this Part, it is not clear that they could have created the political circumstances necessary for four more years of a Democrat-led federal government. Nor do I think we should be surprised by the programmatic retrenchment of OBBBA, given the substantive similarities of those cuts to attempts by previous Republican party caucuses in Congress.

I agree with Professor Bagenstos that the social meaning of these expansions probably got in the way of making them permanent.^[57] If the policies were framed as part of a public health emergency, then the policies’ proponents would have a tough time making the case to the American people that the policies should outlast the emergency.^[58] I part ways, however, with Professor Bagenstos on his analysis that the failure to expand these benefits stems, in large part, from an employer backlash.^[59] I am always uneasy when we try to tie national electoral results to a particular policy domain, especially in a country as large as ours—perhaps even more so at a time of bitter and extreme partisan polarization and a splintered media landscape. But since I am responding to Professor Bagenstos’s lecture, I will throw in my two cents. Here’s a penny’s worth of thoughts on the 2024 election plus another penny on the second Trump Administration’s social policy agenda so far.

My first thought is to narrow the timeframe in the 2024 election, albeit with a historical analog. To better understand the first national election when the Second Pandemic Deal was arguably on the ballot, it’s worth considering parallels to the first national election when the New Deal was effectively on the ballot.^[60] The Democratic Party’s fate in 2024 was markedly different than that party’s fate in 1936. In 2024, Democrats lost the White House and seats in both chambers of Congress.^[61] In 1936, President Roosevelt was reelected with the highest share of the popular vote and the largest share of the Electoral College since 1820, to this day, the largest ever for a Democrat.^[62] Democrats won the largest House majority since 1866. Indeed, it was the last time either party held more than three-quarters of all the seats in that chamber. Meanwhile, the Democrats in the Senate held seventy-six of the ninety-six seats, still the largest majority in the history of the party.^[63]

It is understandable, then, why historians might see 1936 as a ringing endorsement of President Roosevelt’s first term in office, and particularly his New Deal policies.^[64] It is certainly true that the results of the 1936 election allowed FDR and his Congressional coalition to consolidate and entrench their domestic agenda.^[65] But new research has come to light about whether we should see the 1936 election as a referendum on a national policy platform and actually as a pocketbook election more reminiscent of the 2024 race.

Indeed, the work of Christopher Achen and Larry Bartels on how FDR and the Democrats were able to string together repeated victories in national elections is particularly instructive for our inquiry here. Achen and Bartels admit that “[t]he most significant electoral legacy of the New Deal era was the establishment of durable Democratic majorities in Congress,” pointing out that following the 1932 election, “the House of Representatives remained in Democratic hands for 58 of the next 62 years.”^[66] But Achen and Bartels show that “the partisan realignment of the 1930s resulted in significant part from the accumulation of myopic short-term assessments of economic conditions in each successive election year.”^[67] Achen and Bartels show that voter attitudes on their own household finances (specifically, income growth) improved just before the nationwide election in 1936.^[68] In other words, the fate of the New Deal, according to their research, hinged not on new political constituencies but ordinary voters looking at their wallets and deciding they had more money than they had been before FDR took office, and, as a result, decided to reward him for it in 1936.

In 2024, President Biden and his political coalition suffered a different fate. Inflation surged soon after President Biden took office, driven by pandemic disruptions, supply-chain breakdowns, stimulus spending, and Russia’s invasion of Ukraine.^[69] Even as inflation eased after 2022, wages failed to keep pace with prices: from 2021 to 2024, real hourly earnings declined. For many households on the eve of the 2024 election, they did not feel like they were experiencing a slowdown in inflation but rather, continued diminished purchasing power.^[70] That gap between nominal growth and lived economic reality helps explain why voters punished President Biden’s party at the polls in 2024, despite the administration’s claims of recovery.^[71]

Furthermore, the Biden Administration and the 117th Congress were working against fifty years of accumulated economic stagnation for most Americans. No presidential administration could pull American workers and low-income households out of the hole dug by a half-century of flat wages. Even if the COVID expansion had persisted, four years of incremental relief could not resolve affordability crises in housing, education, and healthcare that have accumulated over decades. The stagnation of American wages for half a century is not a problem that a handful of reconciliation and appropriations bills can solve.^[72]

But that might not even be the right question for why the policy response of the Biden Administration did not translate into electoral victory for his Vice President Kamala Harris or create Democratic majorities in Congress. That coalition might merely have been punished for the very real economic hardships American households were experiencing when they went to the polls—or at least that’s what Achen and Bartels might predict.^[73] The ways in which those hardships coincided with the timing of the 2024 election mattered, just as they did in 1936, but with dramatically different results.

Here’s my other reaction to tying the pandemic legislation to the trajectory of President Trump’s social policy agenda. I question whether we should see the policies of the Second Pandemic Deal as paving the way for OBBBA. I do not see OBBBA as a reaction to COVID-era policies, but rather as another chapter in insecure Congressional majorities contesting the three interlinked legacies of the New Deal, the Great Society, and the ACA.^[74] There has been a remarkable amount of continuity in the social policy planks of both parties’ national platforms. The Democratic Party has championed health care expansion since its inclusion in its 1948 platform.^[75] The Republican Party, since at least the 1980s, has been committed to undoing the welfare state the Democrat-dominated Congresses began to build in 1933 and into the 1970s.^[76] As I have explored in earlier work, President Trump tried mightily in his first term to undermine Medicaid and SNAP, with the help of a committed and self-proclaimed welfare reform ideologue in Speaker Paul Ryan.^[77] The first Trump Administration did not have the votes to do it legislatively, so they tried administratively, only to be stymied in the federal courts.^[78] Once President Trump returned to the White House, the Republican coalition in Washington ran the same retrenchment playbook, but this time with a more cohesive party caucus in both chambers of Congress.^[79] So I take (hopefully not inconsistently) a short-term view of the 2024 election—albeit on the basis of a useful historical analog—and a longer view on the reconciliation bill that followed. OBBBA is not a repudiation of ARPA and President Biden’s agenda. Instead, it is President Trump and his allies in Congress in 2025 picking up where they left off in 2020.

On this reading, OBBBA is the latest chapter in the Republican Party’s four-decade fight to undo the welfare state the Democratic coalition in Congress built. Recall that since Congress passed the ACA, the U.S. House of Representatives voted to repeal or amend the ACA more than fifty times in a four-year span.^[80] Now, the House GOP has willing partners in the Senate and the White House.^[81] As Professor Bagenstos rightly points out, with OBBBA, the Republican-led Congress did not technically repeal the ACA, but they have set our society on a course that will wipe out the coverage gains of the ACA in the next few years.^[82] That chilling prospect is where I turn next.

V. More Reasons Why OBBBA Is Era-Defining Retrenchment

My research and perspective overlap most with Bagenstos’s lecture, where he argues that OBBBA is more than routine welfare cutbacks. It is a legislative package with the potential to define an era.^[83] Here are two more reasons why Professor Bagenstos is right about OBBBA. The first is that the Trump Administration has achieved in statute what it tried in administrative practice during the first term. Second, OBBBA is also a statute that pursues systemic, in addition to programmatic retrenchment.

A. Legislating Devolved, Disaggregated Conditionality

In earlier work, I argued that the first Trump Administration pursued an “interlocking strategy” of what I called “devolved, disaggregated conditionality” to undermine Medicaid and SNAP.^[84] “Devolved” because the administration used ideological allies in states to implement its preferred policies, especially through waivers; “disaggregated” because eligibility and deservingness was sliced into ever-narrower categories administered through siloed systems; and “conditionality” because access to benefits was made contingent on continuous proof (including work reporting, verification, recertifications, and compliance with appointments) such that missing a step functioned as a sanction.^[85] The strategy was to cut welfare programs by raising an individual’s price to participate in these programs so that enrollment would fall without an outright cut.^[86] The first Trump Administration’s efforts in this area were largely blocked by the federal courts.^[87]

OBBBA codifies the Trump Administration’s playbook into the statutory architecture of SNAP and Medicaid. Where the prior strategy relied on waivers, guidance, and state-level experimentation, OBBBA mandates more eligibility determinations and strengthens time limits and work requirements to encourage faster and more frequent benefit termination.^[88] For instance, by implementing national work requirements for Medicaid recipients, federal law now requires all states to verify either a certain number of work hours or an appropriate exemption at least every six months.^[89] By design, these provisions create a programmatic default from continued enrollment to programmatic churn. As a result, OBBBA institutionalizes administrative burdens as a core method of retrenchment. By burying would-be beneficiaries in paperwork, the legislation shifts compliance costs to households and insulates lawmakers from direct accountability. Moreover, in ways that rhyme with Bagenstos’s analysis, the bureaucratic innovations to reduce administrative burdens for American households during the pandemic have been deliberately reversed.

OBBBA also hardens longstanding lines of deservingness. It intensifies time limits for so-called ABAWD Able-Bodied Adults Without Dependents (ABAWD) in SNAP.^[90] It excludes more categories of immigrants from cash, food, and health supports.^[91] These measures codify the idea that welfare assistance is suspect and contingent. In short, OBBBA hardwires into the Medicaid Act and the Food & Nutrition Act the very policies and strategies of the first Trump Administration: it shifts costs to households, shifts discretion to state agencies, and disaggregates beneficiaries in ways that target working-age adults and immigrant families.

B. OBBBA as Systemic Retrenchment

Moreover, OBBBA should be understood not just as a case of programmatic, but also systemic retrenchment, to use Pierson’s terms.^[92] Indeed, these efforts, as I suggested above, pre-date OBBBA.^[93] Pierson identifies four ways government can engage in systemic retrenchment of the welfare state. First, the government can “defund the welfare state by constraining the flow of revenues to future administrations.”^[94] By making permanent the individual and corporate tax cuts from the 2017 Tax Cuts and Jobs Act and layering on new preferences for wealthy households and large firms, OBBBA is Robin Hood in reverse.^[95] President Trump’s success in his second administration to make his first administration’s tax cuts permanent erodes the fiscal capacity of the American welfare state.^[96]

Second, policymakers can engage in efforts to reshape public opinion about the welfare state.^[97] As Bagenstos suggests, the failure of the Biden Administration and its congressional allies to contest the social meaning of COVID-era expansions may have created openings for President Trump and the new Congress to reshape our understanding of Medicaid as something certain adults need to work for. Relatedly, I hope Paul Pierson would admit as a friendly amendment to his systemic retrenchment typology including those efforts that undermine the data by which we assess the welfare state’s efficacy. The Trump Administration’s efforts, albeit not through OBBBA, to undermine reporting of the unemployment rate or food insecurity, should be added to the catalog of systemic retrenchment efforts, especially as it relates to public understanding of need and government effort.^[98]

Third, retrenchers can “modify political institutions, changing the way decision making about the welfare state is carried out and thus potentially changing policy outcomes.”^[99] OBBBA entrenches federalism in ways that deepen inequality. Programs that once operated with national floors are now fractured by state choice, inviting divergence and exclusion.^[100] In that sense, my earlier point about devolved conditionality could be recast as not just programmatic, but also systemic retrenchment.

Fourth and finally, “an administration might weaken pro-welfare state interest groups.”^[101] OBBBA weakens pro-welfare constituencies because many of the impacts on Medicaid will reverberate for health care providers in ways that could undermine the sector for future expansions.^[102]

OBBBA defunds the tax base that undergirds the welfare state through tax policy, making future expansions harder to finance. It reframes public opinion by valorizing work requirements. It alters institutional rules by devolving authority downward. OBBBA is thus both programmatic and systemic in its ambitions to undermine the American welfare state.

Taken together, these features of OBBBA suggest that the legislation is an inflection point in our country’s social policy. OBBBA, as Bagenstos relates, crystallizes the fragility of the pandemic-era expansions while embedding inequality into the statutory structure and fiscal federalism of the safety net.^[103] It does not merely reverse the last four years. It sets the terms for the next several years to come.

VI. Conclusion

The last five years of American social policy, from the pandemic expansions to OBBBA’s recent retrenchment, underscore how fragile and contingent welfare politics remain in the United States. Crises still generate moments of possibility, but those openings are quickly narrowed by the constraints on Capitol Hill and in our reliance on state and local implementation. Retrenchment, in turn, comes not only through programmatic cuts but through systemic strategies in social policy. In responding to Professor Bagenstos, I have tried to show that while expansion and retrenchment are best understood together, we should not necessarily read causation into that relationship. That recognition should not make us complacent. It should remind us that the work of building and sustaining social provision will always be contested, incomplete, and shaped by law.