Who Gets to Undo the Administrative State?

California sued and obtained preliminary relief. But the dispute matters well beyond vehicle emissions. It asks a basic question with increasingly broad consequences: who gets to reopen a government decision after the agency has acted, reliance has formed, and ordinary channels of review appear to have closed?

Administrative law often assumes a sequence: Congress writes the statute, the agency acts, courts review the result, and regulated parties adjust. The California waiver fight shows why that sequence is less settled than it looks. Even after an agency acts and reliance builds, another institution may try to reopen the decision by changing the legal label attached to it. That is the durability problem. The question is not just whether a decision was lawful when made but whether it will remain reliable after businesses, states, lenders, and counterparties have organized around it.

From Validity to Durability

For a long time, companies shared an understanding of administrative finality and how it affected their operations. The usual questions were familiar: what does the rule require, did the agency have authority, and did the action survive review? Those questions still matter. They just may not be enough. A decision can be valid today and exposed tomorrow if a later actor can recast the action, reopen the process, or invoke a different procedural route.

Administrative decisions can be reopened in several ways, such as: 

• Congress can use the Congressional Review Act (CRA) or appropriations process. That matters because treating an agency action as a CRA “rule” may expose it to expedited congressional disapproval. (For a deep-dive on the CRA, see here.)

• A new administration can reconsider or rescind a predecessor’s action. (For examples, see here and here.) 

• Grant programs can be cancelled or funding withheld. 

• Leases or permits can be delayed or withheld.

• Following challenges, courts can vacate rules or apply new doctrine to old assumptions. (See, e.g., recent actions under the Clean Water Act.) 

None of these tools are new. The change is that regulated parties increasingly have to account for several of them occurring at once, often after they already have made business decisions in reliance on the government’s earlier action.

That changes the advice. Counsel cannot stop at whether the agency acted lawfully. The harder questions are who else can disturb the result, whether the attack will be substantive or procedural, and what happens to investments already made if the government later recharacterizes the action.

Final Decisions Still Matter

Durability is not just a problem when government decisions are reopened. It also matters because some decisions, once final, have consequences that businesses cannot simply ignore. In Chamber of Commerce v. EPA, the DC Circuit upheld EPA’s designation of perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid (PFOS) as hazardous substances under the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA). (For more on this case, see here.) The court did not reopen the agency’s decision. It sustained it. That result matters because a final hazardous-substance designation can reshape legal and commercial obligations well beyond the rulemaking record.

A CERCLA hazardous-substance designation does not stay confined to the administrative process. It can affect cleanup authority, reporting, cost recovery, transactions, insurance disputes, and allocation proceedings. For businesses with historic per- and polyfluoroalkyl substance (PFAS) use or exposure, the designation can connect current liability to old operations, old properties, and old supply chains. The consequences may be prospective in form, but they often reach backward in practical effect.

That is why the durability frame matters. The solution to uncertainty cannot be to treat administrative decisions as if they do not matter until every possible avenue of revision has closed. Once final, agency decisions can become the baseline for regulators, counterparties, insurers, plaintiffs, lenders, and deal teams. Companies are then placed in a difficult position: reliance may be expensive if the decision later changes but refusing to account for the decision may be more expensive if it survives.

The PFAS case shows one side of the problem. A consequential agency decision can survive review and become embedded in the legal landscape, with effects that reach across legacy operations and future transactions. The California waiver fight shows the other side: an already-effective decision may later be targeted through recharacterization. Together, the cases show why the operative question is not simply whether an agency action is valid at a particular moment. It is how much practical reliance the decision can bear, and what happens if institutions later act to unsettle it.

The California Waiver Fight

The California litigation presents the fragility problem in a particularly clean form. After decades in which California waivers were treated as outside the CRA process, EPA reclassified several already-effective waiver orders as CRA “rules” to allow CRA-based challenges. That move created a path to congressional disapproval that did not exist when the waivers were issued. California challenged the move on the grounds that EPA had converted settled waiver decisions into targets for a different institutional process.

Judge Beryl Howell described the agency’s position as requiring “Orwellian doublethink”: EPA argued both that it had determined the waivers should be treated as rules and that it had merely sent a report to Congress, with no separate reclassification at all. That is not just a pleading problem; it is a practical problem for regulated parties. Finality becomes unstable when the government can benefit from a new characterization while denying that the characterization changed anything.

The threshold issue is whether the waivers are “rules” for CRA purposes. The court rejected the suggestion that EPA could make that designation unreviewable simply by invoking the CRA, quoting the DC Circuit’s warning that a jurisdictional bar “does not turn upon” an agency’s “say-so” and that a “bald assertion of power by an agency cannot legitimize it.” The opinion then put the point more directly: an agency cannot use a statutory no-review provision to cloak what the court viewed as an otherwise reviewable power grab.

That language matters because it moves the dispute beyond a technical CRA question. The court saw a procedural move with real institutional consequences: an agency telling Congress one thing, taking a narrower position in court, and asking the judiciary to accept both at once. The opinion called EPA’s explanation “mind-numbing word calisthenics,” observed that “such gamesmanship may seem like a clever policy move,” and warned that it “undermines honest compliance with the law.” For businesses, that is the warning. Policy can change. But so can the legal categories that are used to make policy change predictable.

The comparison is useful because the cases point in opposite directions. PFAS shows what happens when a consequential decision survives and becomes embedded in the legal landscape. The California dispute asks what happens when the government tries to reopen an older decision through recharacterization. In one setting, durability extends costs into legacy operations and future transactions. In the other, reliance turns out to be less secure than it appeared — not because the original merits changed, but because later actors found a new threshold theory.

Durability and Investment Risk 

Revision risk is especially acute where the investment horizon is longer than the political cycle. Data centers, transmission lines, pipelines, manufacturing facilities, and remediation projects all depend on government decisions that may remain exposed long after financing closes or construction begins.

A permit may be valid but vulnerable to reconsideration. A grant may be announced but still exposed to a funding pause. A regulatory classification may survive review and create liabilities tied to decades of prior operations. The valid-or-invalid frame misses the business question: whether the decision is reliable enough to support capital commitments, contracts, financing, and long-term compliance planning.

Simple, concrete questions can provide some certainty.

• What kind of agency action is involved?

• Was it adopted through notice and comment, or does it rest on guidance or an informal position?

• Has a court reviewed it? Could the CRA, appropriations process, or state litigation supply another path of attack?

• Which institution has both the authority and the incentive to act?

• What happens to the project if the answer changes after commitments are made?

Those questions belong in diligence, contracting, financing, and litigation strategy. The protections will vary by project, but the parties who must live with the risk — developers, manufacturers, sponsors, lenders, insurers, and counterparties — may need change-in-law provisions, staged investment, alternative compliance paths, or a record built with later reconsideration in mind. The point is not to assume every decision will be reopened. It is to identify which decisions are important enough that durability should be tested before reliance hardens around them.

Who Gets the Last Word?

The administrative state has never offered absolute finality. Agencies reconsider policy. Congress changes statutes. Courts correct legal errors. Elections have consequences. But the current fights are exposing how many institutions may claim revision authority over the same decision, sometimes on different theories and different timelines.

That fragmentation increases the value of procedural leverage. A party that cannot win the policy fight may still challenge the form of the agency action, the source of authority, the timing of submission, the availability of review, or the remedy. Those threshold questions can decide whether the merits are ever reached. The California opinion makes the point vividly because the court treated the characterization fight itself as consequential, not as a harmless label.

The California waiver litigation is a clean example. A fight presented publicly as a dispute over vehicle policy turns on characterization, procedure, and institutional power. The PFAS case supplies the corollary. Once a consequential designation survives challenge, its practical impact can expand well beyond the immediate rulemaking. Together, the cases suggest that the operative question is not merely whether a rule is valid. It is whether the legal settlement around it is durable enough to support reliance.

For the regulated community, compliance remains necessary, but it is no longer sufficient. The harder question is who can still disturb the decision, on what theory, and after how much reliance has formed. More examples will test the point — permitting, grants, infrastructure approvals, and legacy-liability designations all present versions of the same problem — but the lesson is already visible. In an era of repeated regulatory revision, administrative finality is not something to assume. It is something to studiously evaluate.