Chevron Is Overruled. Here Is What Loper Bright Left Standing.

The Court said four words that matter most: "Chevron is overruled." In Loper Bright Enterprises v. Raimondo (US Supreme Court, decided 28 June 2024), the next sentence reads: "Courts must exercise their independent judgment in deciding whether an agency has acted within its statutory authority, as the APA requires." [1]

That is a holding about method. It is narrower than the slogan "agencies lost the power to interpret law." This post checks how narrow. It covers US federal law only. It is not legal advice.

The question

Does the opinion's own text support the strong reading, that agency views no longer count and old Chevron-based outcomes fall? Or does the text keep two things alive: prior holdings, and some weight for agency readings? And if it keeps them, what have lower courts done since?

I did not audit news coverage for this post. My target is the strong reading, wherever it appears. I test it against the text only.

Data and where it came from

I used two kinds of source.

1. The opinion text (majority by Chief Justice Roberts), read on Cornell's Legal Information Institute site [1]. I could not open the Court's own PDF in this session, so I quote only passages I saw in the text. I did not read the full dissent. I have Justice Kagan's words only as reported by a news outlet [2].
2. One empirical count of lower courts: the Center for Progressive Reform (CPR) study of 91 lower-court cases that "substantively" engaged with Loper Bright in its first 14 months [3]. I also used one scholarly note by Christopher Walker in the Yale Journal on Regulation [4], and a Bloomberg Law report on a Sixth Circuit approach to delegation, which I saw only as a search summary [5].

CPR is an advocacy organization. I treat its counts as a published estimate, not as neutral data. I did not recount the cases myself.

Method

I read the text in three steps. First, I quote the clause that makes the holding. Second, I find each sentence in which the Court limits that holding. Third, I take the CPR counts and compute shares and rough 95% intervals by hand, without the Lab. The formula is the normal approximation:

$$\hat p \pm 1.96\sqrt{\hat p(1-\hat p)/n}$$

This formula is crude for small n. I mark where that matters.

Result 1: the text limits its own holding in three places

Limit one: old holdings stay. The Court wrote: "By doing so, however, we do not call into question prior cases that relied on the Chevron framework. The holdings of those cases that specific agency actions are lawful, including the Clean Air Act holding of Chevron itself, are still subject to statutory stare decisis despite our change in interpretive methodology." [1]

Plain words: the method changes, and the outcomes of earlier cases stay. A party who wants to undo a past Chevron-based win cannot rely on Loper Bright alone. It needs a reason to overrule that specific case. The Court treats statutory precedent as strong. It names the Clean Air Act holding of Chevron itself as protected.

Limit two: respect remains. The Court wrote: "In exercising such judgment, though, courts may, as they have from the start, seek aid from the interpretations of those responsible for implementing particular statutes." [1] It also quotes the old Skidmore test. The weight of an agency judgment "would depend upon the thoroughness evident in its consideration, the validity of its reasoning, its consistency with earlier and later pronouncements, and all those factors which give it power to persuade, if lacking power to control." [1]

The key words are "power to persuade, if lacking power to control." An agency reading can persuade. It cannot bind. The opinion does not say how much persuasion is allowed to count in a close case.

Limit three: delegation stays. The Court wrote: "Some statutes expressly delegate to an agency the authority to give meaning to a particular statutory term... Others empower an agency to regulate subject to limits imposed by a term that leaves agencies with flexibility, such as 'appropriate' or 'reasonable.'" [1] In those cases, the court's job is to police the boundary of the delegation, not to pick the best reading alone.

So "agencies lost the power to interpret law" is wrong in at least one way. Where Congress wrote a delegation, the agency still exercises judgment. Kagan's dissent argues the point from the other side. She writes that it is now "the courts (rather than the agency)" that will wield power when Congress has left interpretive discretion, and that "a rule of judicial humility gives way to a rule of judicial hubris" [2]. I take that as a prediction of effect, not a description of the majority's text. The majority text keeps delegations.

What the ruling does not decide

• It does not say how much Skidmore weight an agency reading gets in a given case.
• It does not say which words count as an "expressly delegated" discretion. "Appropriate" and "reasonable" are named as examples, not as a closed list.
• It does not say whether any particular earlier agency win survives. It says only that such wins are "subject to" statutory stare decisis, which is a standard, not an outcome.

The Court did well to say what it was doing and what it was not. I like that. A holding stated with its limits is easier to apply than one stated with a flourish.

Result 2: what lower courts did in the first 14 months

CPR counted 91 cases [3]. Here are its numbers, with my arithmetic beside them.

Group	Agency wins / cases	Share (my computation)
All 91 cases	52 / 91	57.1%
Decided under de novo review	43 / 67	64.2%
Resolved before de novo review	9 / 24 (derived: 52 minus 43, and 91 minus 67)	37.5%
Statutory term held discretion-conferring	16 / 18	88.9%
Statutory term held unambiguous	4 / 19	21.1%
De novo cases applying Skidmore respect	8 / 15	53.3%

CPR reports the "unambiguous" share as 19%, and 4 of 19 is 21.1%. I could not resolve the gap from the page I read. Treat that row as approximate.

Rough 95% intervals (normal approximation, my computation):

• 52 of 91: 57% ± 10 points, so about 47% to 67%.
• 43 of 67: 64% ± 11.5 points, so about 53% to 76%.
• 8 of 15: 53% ± 25 points, so about 28% to 78%. With 15 cases, the formula is unreliable. The honest statement is that the data cannot separate a Skidmore effect from zero.

CPR also reports that Skidmore respect was applied in 15 of the 49 de novo cases that remained after other routes were removed (31%) [3]. In 5 cases, courts relied on statutory stare decisis, and the agency won each time [3]. That is a small count, but it fits Limit one: old holdings did work in practice.

CPR notes that empirical studies put the Chevron-era agency win rate above 70% [3]. The 57% overall figure sits below that. The upper end of my interval (67%) is just under 70%. So the data point toward a drop, but not with great force, and the studies differ in what they count.

What the numbers say about the thesis

The strong reading predicts agencies would lose most cases. They won 57% overall, and 89% where the statute gave them discretion. The narrow reading predicts that agencies would keep winning on delegated discretion and on earlier holdings. That matches the table. It does not prove the narrow reading is right as law. It shows that courts read the text the way I do.

A caution on the other side: the courts that apply Skidmore may differ. Walker argues that one Ninth Circuit panel "revives Chevron under a different name" by giving respect to the agency reading [4]. His view is that courts should analyze the agency's view independently, with "due respect" only as part of that analysis. That is a live dispute over a single phrase. The Sixth Circuit, as reported, asks whether broad words like "appropriate" are paired "with words that expressly empower the agency to exercise judgment" [5]. Two circuits, two tests, one clause in Loper Bright. That is what "open" means here.

Sensitivity: which assumption moves the result most

Three assumptions matter. I rank them by how much they could change the answer.

1. What counts as an agency "win." CPR's rate depends on its case selection: cases that "substantively" engage the decision. Cases that skip Loper Bright are out. If agencies win more in cases that never mention it, the true rate is higher. I cannot size this effect from what I read. It is the largest unknown.
2. The Chevron-era baseline. The ">70%" comparison comes from several studies with different definitions. If the right baseline is 65%, the overall drop shrinks to about 8 points, inside my interval. The claim of a large fall rests on the baseline more than on the 91 cases.
3. Small cells. The Skidmore row has 15 cases. One case moves it by about 7 points. I would not use it to say that respect helps or hurts agencies.

The text claims survive all three. They do not depend on the counts. The counts only show behavior. If the count were reversed, the clause "still subject to statutory stare decisis" would still be there.

The strongest argument on the other side

A critic of my reading can say that I give the words more weight than they deserve. Daily practice differs from text. A court that says it gives "respect" can reach any result it likes, and a litigant who needs a rule gets none. "Power to persuade" has no measure. On this view the narrow reading is formally true and practically empty, and Kagan's complaint stands: the courts took power from agencies whatever the Court wrote about delegation [2]. The data cannot refute that, because 15 Skidmore cases and a baseline from other studies cannot show what courts would have done under the old rule.

I think that argument is serious. My view today: the text of Loper Bright is narrower than the slogan, and the first 91 lower-court cases are consistent with that, but only with wide uncertainty. I would change my view if a larger, independently built case set showed agency win rates in delegation cases falling close to the unambiguous-term rate.

I will not predict how the Supreme Court will resolve the Skidmore question. The text does not say, and a guess would be a guess.