Claims That Arise Under Federal Law

Claims That Arise Under Federal Law

The well-pleaded complaint rule settles where a court looks, confining the inquiry to the plaintiff's own statement of the claim. It does not say what the court is looking for once it gets there. A complaint may mention federal law in a dozen places and still not arise under it, and a complaint that never uses the word "federal" may arise under it all the same. The test the Court applies to the pleading asks the pleading to establish one of two things: either that federal law creates the cause of action the plaintiff asserts, or that the plaintiff's right to relief depends on resolving a question of federal law.¹ Those two branches, plus one narrow doctrine that overrides the plaintiff's own characterization of the claim, are the whole of the inquiry.

The first branch does nearly all the work, and it does so because it is cheap to administer and because it tracks the reason Congress created the grant. The familiar formulation is Justice Holmes's: "A suit arises under the law that creates the cause of action."² When Congress has both declared a right and authorized a private suit to enforce it, the dispute is federal in substance as well as in form, and no judgment about importance is needed to see it. An author who sues a publisher for an injunction against unauthorized reproduction of his play is suing on a right the federal copyright statute creates and a remedy the same statute supplies, so the case is federal without further analysis.³

What makes the doctrine harder than that is the status of the creation test as a rule of inclusion rather than exclusion. A case that satisfies it arises under federal law; a case that fails it is not thereby excluded, because the Court has repeatedly found arising-under jurisdiction where the vindication of a state-created right turned on the construction of federal law.⁴ The presence of a federal ingredient in a state-law claim has sometimes been enough and sometimes not, which is why this branch has generated most of the litigation and most of the exam questions.⁵ The reason for the hedging is a genuine tension. A purely mechanical creation test would push disputes whose entire contested content is federal — the validity of a federal agency's securities, the adequacy of notice under a federal tax-sale statute — into state court merely because the plaintiff's remedy carries a state-law label.⁶ An open-ended ingredient test would do the opposite damage, federalizing routine tort and property litigation whenever a federal regulation happened to supply the standard of care, and creating a federal forum for claims Congress deliberately declined to make federal at all.⁷ The doctrine's shape is an attempt to catch the first group without sweeping in the second.

The first two are branches of the same statutory question and are analyzed on the face of the complaint as pleaded. The third operates differently, which is why it is taken last and kept separate.

Federally Created Causes of Action

The first and easiest route into § 1331 asks a single question about the source of the plaintiff's right to sue. A suit arises under the law that creates the cause of action, which means that if Congress both declared the right and authorized this plaintiff to enforce it in court, the inquiry is over and the district court has jurisdiction.⁹ The most reliable way to run the test is to imagine the judge asking the plaintiff, before any evidence is taken, why he is entitled to anything at all from this defendant. If the honest answer is "because a federal statute gives me this right and lets me sue on it," the creation test is satisfied; if the honest answer names a state's contract, tort, or property law, it is not. That framing matters because the question is about the source of the entitlement, not about how prominently federal material appears in the pleading or how fiercely the parties expect to argue about federal law later.

Three words that students treat as interchangeable do different work here, and the creation test keys on only one of them. A right is a judicially enforceable obligation — the substantive entitlement the law says one person owes another. A cause of action is membership in the class of persons the law empowers to go to court and enforce that right. A claim is the set of facts which, if true, would entitle this plaintiff to relief. The creation test asks who supplied the cause of action, because that is the question about which sovereign has opened its courts and authorized this category of private enforcement.¹⁰ In the ordinary federal statutory case the right and the cause of action travel together, so the distinction stays invisible: Congress declares the entitlement in one section and authorizes the private suit in the next, and both are federal. The distinction becomes visible only when the two come apart, which is where the hard cases in the following subsections live.

The copyright statute shows the test working at its simplest. Suppose a playwright learns that a publishing house has announced an edition of his play without permission, and he sues for an injunction; the federal copyright statute both creates the author's exclusive right to reproduce the work and authorizes suit to enjoin anyone who uses it without permission, so the case plainly arises under federal law.¹¹ Employment statutes work the same way. A worker discharged at sixty-one who sues under the federal age discrimination statute can point to a provision creating his right to be free of age discrimination and a provision expressly permitting employees who suffer it to sue for damages, and he relies on that statute as the source of his right to relief.¹² Notice what the second example reveals about the test's reach. The parties in that suit may never dispute a word of the statute's meaning, arguing only about whether the employer fired him for his age or for poor performance, and the case still arises under federal law. Jurisdiction follows the source of the cause of action, not the presence of a contested federal issue.

The test earns its place because it is nearly self-applying, and a jurisdictional rule that is nearly self-applying keeps the parties litigating the merits instead of litigating the forum. Anything requiring a judgment about how important the federal element is invites a preliminary fight that costs more than it resolves, and the overwhelming majority of § 1331 cases are plaintiffs pleading a cause of action created by federal law — so a rule that disposes of them mechanically is doing most of the statute's work at almost no administrative cost.¹³ There is also an institutional reason the test is drawn this way. When Congress defines a right and simultaneously authorizes private enforcement, it has already made the judgment that the right matters enough to be vindicated in court, and routing those suits to federal judges is what produces a coherent body of interpretation of the statute across the country.

The other pole is where federal law is present in the surroundings but absent from the cause of action. A driver injured in a collision on an interstate highway, suing a trucking company engaged in interstate commerce, has a federal aura about her case and no federal cause of action in it: her complaint alleges duty, breach, and causation under state tort law, and she will not have to prove anything about federal law to recover.¹⁴ The same is true of a contractor suing over unpaid work on a federally funded research project. Federal money paid for the building, but the right the contractor asserts was created by the state law of contracts, and that is the law under which the suit arises. The contrast with the copyright and discrimination examples is a single variable — whether a federal enactment is the reason this plaintiff may sue at all — and nothing about the visibility of federal law in the background changes the answer. The free-pass litigation in Louisville & Nashville R. Co. v. Mottley is the classic instance: the Mottleys could sue because state contract law permits a court to order specific performance of a broken agreement, so their suit arose under state law despite the federal statute everyone knew was coming.¹⁵

Two refinements should be filed away rather than fully unpacked here. The first is that the creation test does not require the plaintiff to have a winning federal cause of action, only a colorable assertion of one; a defendant who answers that the plaintiff is not actually covered by the statute is denying the claim on the merits, not challenging the court's power, and that distinction is developed in the section on jurisdiction versus the merits below.¹⁶ The second is the rare statute that authorizes a federal suit but directs the court to decide it entirely under local substantive standards. Such a case clears the creation test in form while generating no federal law at all, which is why it sits uneasily with the grant: federal dockets would fill with disputes whose governing rules are wholly state-made, and no uniformity of federal doctrine would result. Keep it in mind as an oddity; the ordinary case is the one where the right and the cause of action are both federal and travel together.

Embedded Federal Issues

The creation test includes cases without excluding them, and this subsection takes up the cases it leaves out: a plaintiff pleads a cause of action that state law created, yet the plaintiff cannot win without establishing a proposition of federal law. The Court has recognized for roughly a century that federal question jurisdiction will sometimes lie over such claims, because a federal court ought to be able to hear a state-law claim that turns on a substantial question of federal law and thereby offer the experience and the hope of uniformity that a federal forum supplies on federal issues.¹⁸ Two paradigm cases fix the category. In Smith v. Kansas City Title & Trust Co. a shareholder sued under state corporate law to enjoin his company from buying bonds issued by federal land banks, and the only reason the purchase could be unlawful was that the federal statute authorizing the bonds was unconstitutional; Missouri law gave the right of action, but the principal issue in the case was federal, and the Court sustained jurisdiction.¹⁹ In Grable & Sons Metal Products, Inc. v. Darue Engineering & Manufacturing the Internal Revenue Service seized Michigan land for unpaid taxes and sold it, and the former owner later brought a state quiet-title action contending that the record title was void because the agency had mailed the seizure notice by certified mail rather than serving it personally as the federal tax statute required.²⁰ The Court held that the national interest in a federal forum for federal tax litigation was substantial enough to support jurisdiction over the disputed federal issue on removal, and that taking the case would not distort the division of labor between state and federal courts that Congress had provided or assumed.²¹

What unites those two suits is the position the federal issue occupies in the plaintiff's own case. The shareholder had to prove the federal statute invalid to show that the investment exceeded the corporation's powers, and Grable had to prove that certified mail did not satisfy the notice statute to show that the buyer's title failed. Neither plaintiff could have reached judgment by proving state-law elements alone and letting federal law enter later through the defendant's answer. That is why these decisions sit comfortably alongside the well-pleaded complaint rule rather than in tension with it: the court still looks only at what the plaintiff must establish as part of the plaintiff's case, and it still ignores the defendant's response.²² Make sure you hold that point separately from the rest of this subsection, because it is where most readers go wrong. The embedded-issue doctrine is an exception to the creation test and to nothing else.

The first two steps usually take care of themselves, and the last two decide the close cases. A plaintiff who has pleaded a federal ingredient at all has normally pleaded one the defendant disputes. Substantiality and balance are different in kind, because they ask not about this lawsuit but about the class of lawsuits that would follow if jurisdiction were upheld.

Substantiality asks whether deciding the federal question matters beyond the parties in front of the court. A pure question of federal statutory or constitutional meaning, of the sort that will govern the government's own future conduct or the validity of a federal enactment, is substantial because the answer becomes law for everyone. The notice question in Grable qualified on exactly that footing: whether the Internal Revenue Service must serve a delinquent taxpayer personally before selling his land determines how the federal government may exercise its collection power everywhere, and the government has a direct interest in a federal court saying so.²⁴ Compare a state legal malpractice suit in which a client alleges that his lawyer botched a patent case, so that the malpractice court must decide how the patent argument would have come out had it been made. The embedded patent question decides that suit completely, yet the court's answer binds no one, sets no precedent for the patent bar, and cannot change the validity of the patent itself, which has already been adjudicated. An issue that is dispositive between these parties but inert everywhere else produces no uniformity benefit, so it is not substantial in the sense the doctrine means.²⁵

The fourth factor is a veto, and it is the reason the doctrine stays small. Even a contested and substantial federal question yields no jurisdiction unless hearing the case is consistent with congressional judgment about the sound division of labor between the two court systems, because arising-under jurisdiction over a state-law claim always risks upsetting a line Congress drew or assumed.²⁶ The mechanism is arithmetic as much as principle. Federal regulation now touches nearly every commercial transaction and most consumer products, so a rule admitting any state claim that borrowed a federal standard would move ordinary tort and contract litigation wholesale into federal court; read too narrowly, the same category would leave the meaning of federal regulatory statutes to uncoordinated state decisions, which balkanizes the federal scheme.²⁷ The balance factor is the valve that holds the middle position, and Merrell Dow Pharmaceuticals Inc. v. Thompson shows it closing. Plaintiffs there pleaded a state negligence per se claim alleging that a drug maker had marketed Bendectin without the warnings federal law required, and the Court held the federal reference insufficiently substantial to confer jurisdiction.²⁸ Two things drove that result. Congress had created no private federal cause of action for the labeling violation, and treating a borrowed federal standard as a jurisdictional key would have attracted a horde of original filings and removals raising other state claims with embedded federal issues.²⁹

The missing cause of action deserves one further word, because students routinely overstate it. A federal private right of action is evidence relevant to the sensitive judgments about congressional intent that § 1331 requires, and it is not dispositive; Grable refused to read Merrell Dow as converting a federal cause of action from a sufficient condition for jurisdiction into a necessary one.³⁰ Congress had created no federal quiet-title action either, and the Court gave that omission little weight, reasoning that Congress may simply have assumed the adequacy of the state remedial schemes under which most quiet-title suits proceed.³¹ The difference between the two cases is volume. Negligence per se complaints borrow federal standards with monotonous regularity, so jurisdiction there would have federalized a tremendous number of personal injury suits; quiet-title actions resting on the validity of a federal tax sale are a thin and self-limiting category.³²

One consequence of resting the test on a disputed question is that the category can shift over time. A federal issue of first impression may be both substantial and genuinely contested, while the same issue, once authoritatively settled, may support no jurisdiction at all — so a later plaintiff raising the identical certified-mail objection in Michigan might well fail the test that Grable passed.³³ That flexibility buys accuracy at the cost of certainty, and it is why this branch generates far more jurisdictional litigation than its share of the docket would suggest.³⁴

Complete Preemption

Preemption looks at first like the one place where the well-pleaded complaint rule must give way, and for the ordinary case it does not. Under the Supremacy Clause a federal statute displaces state law that conflicts with it, but that displacement is something the defendant raises in answer to the plaintiff's state-law claim, which means it enters the case exactly where Mottley said jurisdiction may not be found.³⁶ A plaintiff who pleads a state tort and a defendant who answers that federal law occupies the field have produced a case with a genuine federal question in it and no federal question on the face of the complaint, so the suit neither arises under § 1331 nor becomes removable.³⁷ Call that ordinary preemption. It is a merits defense wearing federal clothing, and the jurisdictional answer is the same one the anticipated-defense rule gives everywhere else.³⁸

Complete preemption is a different animal, and the difference is not one of degree. The doctrine holds that a federal statute's preemptive force may be so extraordinary and all-encompassing that it converts an ordinary state-common-law complaint into one stating a federal claim for purposes of the well-pleaded complaint rule.³⁹ Notice what that does to the pleading. The court is not excusing the plaintiff from the well-pleaded complaint rule; it is deciding what the well-pleaded complaint actually says, and concluding that the state-law label on its face describes a cause of action that federal law has abolished and replaced. Once the claim is reread as the federal claim Congress supplied, it satisfies the creation test in the ordinary way, because federal law now both grants the right and authorizes this plaintiff to sue on it. That is why the doctrine supports original jurisdiction and removal alike, while ordinary preemption supports neither.

The reason the category exists is that a contrary rule would let the plaintiff's choice of words defeat a decision Congress has already made. When Congress builds an exclusive enforcement scheme, it has determined that this class of disputes will be governed by one body of law and remedied on one set of terms; a plaintiff who pleads the same grievance as a state breach of contract and stays in state court undoes that determination by artful drafting. The reason the category stays tiny is the mirror image. Federal supremacy ordinarily produces a defense and nothing more, so a rule that treated any serious preemption argument as jurisdictional would reverse the anticipated-defense rule across most of federal regulatory law and move a vast quantity of state litigation into federal court on the defendant's say-so. Courts therefore confine complete preemption to statutes whose text and structure show that Congress meant its remedy to be the only one available.

The recognized examples are few, and two of them carry nearly all the weight. Employee benefit litigation is the first: the federal courts have exclusive jurisdiction over ERISA causes of action apart from a participant's suit to recover plan benefits, which may be brought in either system, and a benefits claim filed in state court falls within federal question jurisdiction and may be removed on the normal timing and joinder conditions.⁴⁰ Labor contract litigation is the second: a state-law claim is preempted by section 301 of the Labor Management Relations Act when its resolution is substantially dependent on analysis of the terms of a collective bargaining agreement, so an employee's state tort suit for bad faith refusal to pay disability benefits owed under such an agreement was preempted even though state law called it a tort, the duties allegedly breached having come from the contract.⁴¹

The two poles sit very close together, and the variable that separates them is whether the plaintiff's own claim lives inside the federal scheme or merely runs into it. Suppose a plan participant sues his employer in state court for breach of contract, alleging that the plan document promised a surgery the administrator refused to pay for. He has pleaded no federal law, yet the only thing he is suing about is a benefit owed under an ERISA plan, so the claim is in reality the federal benefits claim and the defendant may remove it.⁴² Change the claim rather than the label and the answer flips. An employee who sues under state law for retaliatory discharge after filing a workers' compensation claim asserts a right the state confers independently, and because that claim can be resolved without interpreting the collective bargaining agreement it is not preempted at all, let alone completely.⁴³ The employer in that second case may still argue federal displacement, but the argument is a defense, and a defense keeps the suit in state court.⁴⁴

One housekeeping point will save confusion later. Complete preemption must be kept analytically separate from the declaratory judgment cases, where a court unwinds the pleading to ask what the coercive suit would have looked like, and from the line of decisions permitting suits for injunctive and declaratory relief against allegedly preempted state regulatory schemes.⁴⁵ Those doctrines reach federal court by different routes and do not recharacterize anyone's claim; only complete preemption changes what the complaint is held to say.