As it has done with increasing frequency in recent years, the Supreme Court under Chief Justice John Roberts has again demonstrated its willingness to enforce the Constitution’s original checks and balances, even if means overturning long-standing precedent. In Loper Bright v. Raimondo, the Court detonated a decades-old pillar of the administrative state: its decision in Chevron v. National Resources Defense Council (1984). Chevron required courts to defer to agency interpretations of law whenever statutory text was ambiguous. For 40 years agencies have routinely abused that deference to expand their authority, rewriting federal law in ways Congress never intended.

To understand the significance of Loper Bright, start with the presidency of Franklin D. Roosevelt. In the midst of the Great Depression, Roosevelt’s New Deal aimed to give Americans renewed hope. Unfortunately, his key proposals entailed raising wages and agricultural prices,  through economic controls that were wholly incompatible with the Constitution as it had been understood since ratification.

First, what the New Dealers proposed far exceeded the federal government’s limited power to regulate commerce “among the several states,” a power previously restricted to transportation and communication across state lines. Just as destructively, they aimed to subordinate the key economic liberties enshrined in the Constitution—private property and the freedom of contract—to a new federal Leviathan bent on reallocating wealth through monopoly and cartel arrangements. With this vast expansion of federal authority, the Constitution’s very separation of powers, on which its checks and balances depended, would be imperiled.

Initially, Roosevelt’s “scorpions” of the Supreme Court struck down one New Deal program after another. But after his landslide reelection in 1936, he threatened to “pack the Court” with additional justices if it did not adopt his radical constitutional views. Many fellow Democrats, including the vice president, were appalled, but the Supreme Court got the message.

>>> Judicial Humility & Reticence in Administrative Law

Starting with the stormy 1936–37 term, the Supreme Court rubber-stamped the New Deal, abdicating its role as guardian of the limits on federal power. The Constitution’s bounded use of the administrative state to keep deed registries, taxes, and voter rolls in order was replaced by a Leviathan empowered to accomplish the New Deal’s grand design: creeping socialism through technocratic regulation rather than outright state ownership of the means of production. “In this theory the individual has no rights as against the majority, for constitutional checks and bills of rights exist only by consent of the majority,” the great progressive columnist Walter Lippmann ruefully wrote of the New Deal in his 1937 book The Good Society.

World War II muted the political consequences of the New Deal’s constitutional transformation. But after the war, the increasingly arbitrary excesses of federal agencies led Congress to bipartisan passage of the Administrative Procedure Act (APA) of 1946. The APA was a sort of constitution for the new administrative state, so defining the scope of judicial review of agency action assumed central importance.

In Section 706 of the APA, Congress created a sustainable arrangement. Analogizing agencies to trial courts, it provided for review of final agency action by federal courts of appeal, and set up essentially the same three standards of appellate review, which differed in the amount of deference they granted to agencies. Agency findings of fact could be set aside only if clearly erroneous, a very deferential standard. Agency conclusions on mixed questions of fact and law (for example, whether a particular drug was safe and effective) could be set aside only if not supported by substantial evidence—intermediate deference.

Finally, pure questions of law were subject to “de novo” review, with no deference other than that due to the inherent persuasiveness of the agency position: “To the extent necessary to decision and when presented,” says Section 706, “the reviewing court shall decide all relevant questions of law, interpret constitutional and statutory provisions, and determine the meaning or applicability of the terms of an agency action.”

In the 1970s, the powerful D.C. Circuit Court of Appeals started to turn this simple scheme into a hodgepodge. Then a pair of mid-1980s Supreme Court cases turned everything upside down. First came Motor Vehicles Manufacturers Association v. State Farm in 1983, which held that an agency’s factual and policy determinations had to be deemed “arbitrary and capricious” if the agency had failed to take a “hard look” at all relevant facts, missing none. This capacious standard lets courts block any agency action that rubs them the wrong way, by pointing to something—anything—that the agency had arguably overlooked. Overnight, Section 706’s substantial deference to agency findings receded as courts started micromanaging minute aspects of executive operations.

The following year, in Chevron, the Court then eviscerated de novo review of questions of law. The project to confound the APA’s deference regime—and an important element of the Constitution’s separation of powers—was complete.

The issue in Chevron was whether the EPA could treat all smokestacks within the same power plant as parts of a single pollution “source” under the Clean Air Act. Under the Carter administration, anti-industry groups had successfully opposed that “bubble” approach, insisting that each smokestack be treated as a separate source. Then the Reagan EPA adopted the “bubble rule” to encourage higher output, with offsets inside the facility to limit emissions increases.

If the term “source” really was textually ambiguous, judges had a straightforward remedy. Given the Clean Air Act’s explicit policy of controlling pollution while promoting innovation and efficiency, the Reagan EPA’s interpretation of “source” was clearly more consonant with the statute. Instead, the Supreme Court skipped over the substantive argument and let the Reagan administration win under a new two-part analysis. Chevron “step one” asks whether the statutory provision is clear; if it is, that settles the matter. But if it is ambiguous, then “step two” defers to the agency’s interpretation if at all “permissible,” even if the Court thinks some other interpretation better.

It was the right result, but the reasoning was disastrously wrong. Section 706 contains not a hint of deference to agencies on pure questions of law. So why Chevron? The answer is partly that in the 1970s the pendulum had swung too far in the opposite direction, with the D.C. Circuit Court of Appeals striking down or altering agency rules on the flimsiest of legal grounds to achieve its invariably progressive policy preferences. For example, in Calvert Cliffs v. Atomic Energy Commission (1971), the D.C. Circuit expanded judicial review over agency permits for infrastructure projects subject to the National Environmental Policy Act, despite Congress’s refusal to provide a private right of action in the original law. That single decision allowed small pockets of determined opponents to tie up badly needed infrastructure projects, such as airports, roadways, and power plants, in years of bureaucratic delay and litigation.

Hence Chevron was initially celebrated as a win for embattled agencies. But the ink had hardly dried on it when its erstwhile supporters started doubting its wisdom. Confusion about how much ambiguity was needed to trigger the second step led multiple courts to create a slew of exceptions, preconditions, and clarifications. Chevron couldn’t explain why the same statutory provision was subject to court interpretation in a private lawsuit but suddenly meant whatever an agency happened to prefer whenever an agency was involved in the litigation. The doctrine led to notorious flip-flops. If a future president reverted to Carter’s aggressive smokestack approach, or decided that the Clean Air Act required “environmental justice,” courts would have to defer. Chevron’s deference to bad legal interpretations turned statutory terms into ping-pong balls between clashing administrations. Judges began to fret that administrative deference under Chevron could let prosecutors undo the rule of lenity, under which courts have long interpreted ambiguous provisions in favor of the accused, not the prosecutor.

The increasing abuse of Chevron reached new heights under President Obama, who routinely bragged that he “rewrote the law.” That led to a stiff judicial reaction in the form of the “major-questions doctrine,” under which courts presume that the agency’s interpretation of an ambiguous statutory provision is not correct when it claims vast new powers. Thus West Virginia v. EPA (2022), in which the Supreme Court struck down Obama’s Clean Power Plan, essentially turned the presumption of Chevron in the opposite direction—against the agency—in cases of great political and economic significance.

After a thousand cuts, the final blow to Chevron was modest enough. A family fishing company, Loper Bright Enterprises, was hemorrhaging revenue because the National Marine Fisheries Service required it, without fresh congressional authorization, to spend perhaps $710 per day to carry observers aboard their vessels as part of an industry-monitoring program. Applicable law authorized the NMFS to require certain categories of fishing vessels—but not Atlantic herring fishermen such as Loper Bright—to pay for observers. Despite this specifically limited authority, in 2020 the agency adopted a rule requiring Loper Bright vessels to hire certified observers, arguing that, under Chevron, any statutory ambiguity must be resolved in the agency’s favor.

The majority opinion in Loper Bright, by Chief Justice John Roberts, is a resounding vindication of checks and balances. He first states the obvious fact that Chevron contradicts the plain text of Section 706 of the APA, which, inexplicably, Chevron did not even mention. He then hits on the real problem: “Perhaps most fundamentally, Chevron’s presumption is misguided because agencies have no special competence in resolving statutory ambiguities. Courts do.” He rejects the view that resolving ambiguous delegations of rulemaking authority involves policy-making best left to political actors: “By forcing courts to . . . pretend that ambiguities are necessarily delegations, Chevron does not prevent judges from making policy. It prevents them from judging.”

The flaw in Chevron becomes clearer when one considers rule-of-law principles along with separation of powers. As Judge Raymond Kethledge of the Sixth Circuit Court of Appeals has noted, to comply with basic rule-of-law principles, every government action must rest on a duly enacted law of Congress, properly enforced by an impartial executive branch, under the supervision of an independent judiciary. The government’s proper exercise of coercive power therefore requires the concurrence of all three branches of government. But under Chevron, any unclear or sweeping delegation whose supposed meaning depends on agency discretion rests ultimately on the consent of only one branch exercising all the functions of government, and not three, as our Constitution requires.

As James Madison wrote in Federalist No. 47, “the accumulation of all powers, legislative, executive, and judiciary, in the same hands, whether of one, a few, or many . . . may justly be pronounced the very definition of tyranny.”

The newly emboldened Roberts Court cannot rest on its laurels, for much work remains to restore the Constitution’s separation of powers. The other side of the Chevron coin is State Farm, and its fatal embrace of “hard look” review of core executive functions.

State Farm asked whether the Reagan-era National Highway Traffic Safety Administration (NHTSA) had properly rescinded a regulation requiring passive-restraint systems, particularly automatic seat belts, in new cars. The Court ruled that the NHTSA’s rescission of the rule was “arbitrary and capricious” under the APA because the agency had failed to provide an adequately reasoned basis for its decision; in particular, said the Court, the agency had not considered alternatives to rescission and had not explained its reversal from the Carter-era policy.

State Farm held that agencies must take a “hard look” at all relevant facts, factors, and alternatives, under a definition so exacting that an agency decision can fail “if the agency has relied on factors which Congress has not intended it to consider [or] failed to consider an important part of the problem.” Now any rulemaking or environmental-impact statement, even those that took years of meticulous preparation and run hundreds of pages long, could be overturned because it failed to include one paragraph the Court might have liked to see.

Concerns about judicial usurpation of agency expertise, so misplaced in Chevron, were fully justified in State Farm. Ministerial fact-finding in the implementation of a statutory scheme is a core executive function. A reviewing court’s inquiry should be limited to whether the agency has properly exercised that function, either as part of delegation of rulemaking authority or as an exercise of inherent executive authority. On these mixed questions of law and fact, the agency should ordinarily receive ample running room. Whereas Chevron confused the separation of powers by letting agencies usurp legislative and judicial functions, State Farm represented judicial usurpation of agency functions.

>>> The Tyranny of the Administrative State

State Farm has been cited nearly 15,000 times since 1983, usually by federal courts second-guessing agency actions they don’t like. However, in an earlier 1983 case, Baltimore Gas & Electric Co. v. NRDC, the Court took a far better approach to the “arbitrary and capricious” standard by upholding the Nuclear Regulatory Commission’s “generic” procedure for nuclear-plant approval, emphasizing the respective roles of Congress and agencies in resolving fundamental policy questions. The Baltimore Gas court got the separation of powers right, recognizing that agency procedures are chiefly for agencies to decide.

In practice, State Farm’s “hard look” doctrine erects a double standard. Any infrastructure permit can be vacated under “hard look” review, to great acclaim from environmentalists. But an agency’s refusal to grant a permit, or its imposition of vast paperwork burdens, is routinely accorded the sweeping deference of “rational basis” review, in which the governmental action is upheld if the relevant court can conceive of any rational basis for it. That runs counter to the sensible approach of Baltimore Gas, which leaves administration to the administration, and is a big reason why American infrastructure is the costliest, most time-consuming, and riskiest to build in the industrial world.

Alas, the correct systematic approach to these mixed questions of fact and law is a harder nut to crack than Chevron, because the highly variable quality of these agencies makes it difficult to balance the twin errors of intervening too quickly and waiting too long. But now that the Supreme Court has rededicated its attention to undoing the structural errors of any earlier generation, it needs to rethink the problem of expertise and bias, which are often present in uneven proportions in administrative decisions.

The time to insist that an agency take a “hard look” at relevant factors is when the agency is opining on matters outside its jurisdiction and expertise. For example, EPA’s rumination on the feasibility of building a vast network of carbon dioxide pipelines to support its new mandate that fossil-fuel plants adopt carbon-capture technologies should receive scant deference: Pipelines are outside its jurisdiction, beyond the control of any entity subject to its regulations, and quite beyond the expertise of anybody who works for the EPA. Agency technical judgments shouldn’t get deference when the agency doesn’t know what it’s talking about.

The issue of court deference to agency action may seem narrow. But it implicates the Constitution’s crucial separation of powers, which has been dissolving since the New Deal and must be shored up before it is too late. We think that, with the most originalist majority in a hundred years, the Supreme Court is well positioned to undertake that task of constitutional restoration. The chief justice is starting to give the distinct impression that he thinks so, too.