Constitutional Limits Do Not Cure an Agency Record

Constitutional litigation maps.

Institutional-design diagrams.

A separation-of-powers graphic where each branch has been assigned a different shade of expensive blue.

Somebody will announce that the action implicates Article II, Article III, the Tenth Amendment, the Fourteenth Amendment, due process, federalism, and possibly the part of the Constitution that stops the vending machine taking a crumpled transit token.

Then the agency produces a decision memorandum with no stated reason for rejecting the material it received.

The memorandum is unimpressed by your constitutional keywords.

It has a smaller, meaner question.

Where is the reason?

Not a grand reason.

Not a national-theory reason.

The reason the actual decision-maker gave at the time, in the record for this action.

A structural limit can matter enormously. It doesn’t reach backward through the file and complete an omitted notice.

It can’t invent an enabling statute.

It can’t put a missing explanation into the mouth of an administrator who had already gone home.

A very large shadow

The constitutional label attracts your attention, because it comes with architecture.

Article I, Section 7 requires a federal bill to pass 2 houses and be presented to the President.

Article II has 4 sections covering the principal textual starting points for executive office, powers, duties and impeachment.

Article III, Section 2 identifies 9 categories of Cases and Controversies.

The Tenth Amendment has 1 reservation clause.

The Constitution Annotated identifies 8 interpretive modes.

Those are real divisions of public power.

They’re also remarkably good at making your modest agency record look like a symposium.

For a challenged agency action, constitutional analysis asks you a bounded set of things.

Which institution acted.

What constitutional provision, structural allocation, right or federalism limit is said to matter.

Whether there’s a justiciable case.

What the procedural posture is.

What remedy is requested.

Youngstown, at 343 U.S. 579, is useful because an asserted policy objective did not replace an available constitutional or statutory source of seizure authority.

That’s an actor-and-power question.

It’s not a reason to stop reading the agency file early.

Your remedy question does its own administrative work.

A successful constitutional claim may lead a court to refuse enforcement, enjoin an official, invalidate an act as applied, reverse a judgment, exclude evidence, or order a new process.

Marbury occupies 44 PDF pages in the Library of Congress scan.

The constitutional point in that report concerns a statutory grant inconsistent with Article III.

The result in your particular case still depends on jurisdiction, claim, defect and remedy.

A facial ruling, an as-applied ruling, damages and a prospective injunction don’t share a single office or a single job description.

That’s your first gate.

It tests whether public power crossed a constitutional boundary.

It doesn’t decide whether the administrator used the correct statutory path on the way to that boundary.

Your agency can face a constitutional process question and a separate programme-specific procedure at once.

A statute or regulation can promise more than the constitutional floor.

The two rules may be friends.

They don’t become the same rule because they arrived at the same building.

Classify before you slogan

Administrative law asks you for a less glamorous act of classification first.

Is the thing a rule, an individual order, a sanction, or non-enforcement?

That distinction isn’t decorative.

It changes your possible trigger, your participation rule, your record, your standard of review and your remedy.

Start with “arbitrary and capricious” before naming the action type and you’ve brought a smoke detector to a filing cabinet.

It may eventually make noise.

It isn’t opening the right drawer.

For a rule of general future effect, the default APA route under 5 U.S.C. 553 has three stages.

Locate authority.

Publish notice and give interested persons an opportunity to submit written data, views or arguments.

Then give the final rule a concise general statement of basis and purpose.

The comments are not a referendum with bad coffee.

They’re your record material.

A technical submission may test a cost assumption.

A comment may identify a missing statutory factor.

The final explanation shows what the agency did with the problem it was required to consider.

An individual benefits, licence, permit, discipline or sanction decision has a different machine.

A fair system ordinarily answers four questions.

What action is proposed.

What facts or rules support it.

How you can respond.

Who decides after your response.

Mathews v. Eldridge treats process as context-sensitive, rather than as a universal pre-deprivation hearing rule.

Formal US federal adjudication gets its particular APA rules only when the enabling statute triggers them under 5 U.S.C. 554.

Every stern letter from a government office is not, despite appearances, a miniature tribunal.

Non-enforcement is different again.

Two challenges get introduced as one grievance far too often.

A regulated party can contest authority, procedure or reasons in its own enforcement case.

You can also try to make the agency pursue somebody else.

The second often meets Heckler v. Chaney and 5 U.S.C. 701.

Non-enforcement is presumptively unreviewable where the statute supplies no meaningful standard.

A mandatory command, a binding rule, discriminatory treatment or a jurisdictional mistake changes the question.

It doesn’t change the fact that the question was different.

The record won’t wear a cape

Your constitutional chart asks who may act.

Your administrative chart asks what this actor actually did with the authority it had.

A review petition or brief ordinarily separates five questions.

Reviewability.

Finality.

Source of law.

Standard of review.

Requested relief.

That sequence can be unreasonably effective.

Your brilliant view of statutory meaning may never be reached, if the action isn’t final or you missed an exclusive review route.

Law is fond of a threshold.

It saves everybody from having to admire your merits.

For reasons review, your record holds the unglamorous evidence.

Studies.

Comment responses.

Factual submissions.

The decision memorandum.

In Overton Park, review had to engage the administrative record.

Not litigation affidavits and post hoc rationalizations.

The agency’s explanation has to be contemporaneous.

Government counsel cannot later place a different motivational engine under the hood and call it a tune-up.

The ordinary arbitrary-and-capricious inquiry is not a demand for prose with perfect hair.

It asks whether the stated explanation connects relevant considerations to the result you got.

You get four common explanation failures.

An impermissible factor.

An important aspect left unconsidered.

A factual premise contrary to the record.

An explanation too implausible to attribute to expertise or a difference in view.

Department of Commerce v. New York describes review as deferential and narrow, while still requiring relevant data and a satisfactory explanation.

Deferential is not a synonym for telepathic.

The record can be more exacting still.

In a covered formal proceeding, 5 U.S.C. 556 makes testimony, exhibits and papers on which findings rest part of the exclusive record for decision.

A party denied access to decisive material has a process question, even where you think the agency reached a sensible result.

Your correction depends on three variables.

The protected interest.

The source of procedure.

The practical value of new process.

Which is why a missing reason isn’t repaired by saying the big constitutional issue somewhere else is very important.

It may be important.

It’s still somewhere else.

Put the questions in columns

Your durable check is smaller than an institutional theory deck, and more useful than a pile of constitutional nouns.

Take one agency act.

Keep the document announcing it, and the material explaining it.

Then make two columns with the same action at the top.

The constitutional column doesn’t absorb the administrative column.

It sits beside it.

Which is a decent arrangement for most things in public law.

ONE AGENCY ACTION

CONSTITUTIONAL COLUMN                 ADMINISTRATIVE COLUMN
actor and constitutional source       decision type: rule / order / sanction / non-enforcement
limit, right, structure, federalism   statutory grant and applicable text
justiciability and posture            procedural trigger and participation supplied
requested constitutional remedy       reasons stated when the action was taken
                                      record materials and review ground

FOUR QUESTION LABELS
constitutional | statutory | procedural | remedial

Those four labels prevent an innocent-looking category error.

Constitutional asks whether the institution or mechanism may act this way.

Statutory asks whether this agency received this authority.

Procedural asks which step, notice, participation opportunity, hearing protection or decisional safeguard applies.

Remedial asks what a reviewing court is being asked to do, and who has to do the next thing.

One factual gap may appear in more than one column.

That doesn’t turn your columns into a group chat.

The statute supplies you particular review grounds where the APA applies.

Section 706(2) names six.

Contrary to law.

Contrary to constitutional right.

Beyond statutory authority.

Procedurally deficient.

Short of substantial evidence, where that standard applies.

Or unwarranted by the facts, where trial de novo is authorized.

Constitutional right is on that list.

So are authority, procedure, evidence and facts.

The list is not a scavenger hunt where finding one dramatic item dissolves the other five.

Current statutory interpretation has its own allocation, and it changed recently.

Loper Bright requires courts in US federal statutory-review cases to exercise independent judgment about the best reading of the statute.

Rather than defer because ambiguity appeared.

Three practical consequences for you.

Courts do their own interpretive work.

Agency expertise may remain persuasive.

Expertise doesn’t end the inquiry.

Kisor supplies a different constrained inquiry for an agency reading its own regulation, beginning with ordinary interpretive tools and genuine ambiguity.

“Agency deference” is therefore a bad umbrella.

It leaks different doctrines on different people.

The court reviews the fence

When your administrative ground succeeds, the usual result has two steps.

Set aside the defective action.

Remand the matter to the agency.

Set-aside removes the action’s legal support.

Remand returns the authorized decision to the authorized administrator, with the missing legal work to do.

That can mean considering a neglected factor, explaining a changed policy, reopening a limited factual issue, or using the correct interpretation.

It doesn’t guarantee you a different result.

And it doesn’t ordinarily ask a court to select the policy in the agency’s place.

This is where a constitutional victory and an administrative remand look alike to you from a distance.

Both can halt an action or require a new process.

Their grounds, records and constraints stay distinct.

New York v. United States, at 505 U.S. 144, held that Congress could not compel states to enact or administer the disputed take-title mechanism, while leaving other routes available.

A notice-and-comment remand is not an anti-commandeering ruling wearing sensible shoes.

The court is enforcing a legal boundary and returning the decision to the institution the law authorizes to make it.

Two gates, in order

Call it the two-gate review.

The first gate asks your constitutional actor-and-remedy questions.

The second asks whether your particular agency action has statutory authority, required procedure, contemporaneous reasons, a record, and a review ground.

It doesn’t decide constitutional validity, statutory meaning, or the final remedy.

It marks the place where you would have to test each decision.

Less majestic than announcing that a matter has constitutional implications.

It’s also how the missing reason stays missing.