Loading case…
United States Court of Appeals for the Fifth Circuit
Appeal from the United States District Court for the Southern District of Mississippi USDC No. 3:22-CV-589
Before Southwick, Oldham, and Ramirez, Circuit Judges .
Leslie H. Southwick, Circuit Judge :
The validity of the arbitration agreement at issue in this appeal turns on the interpretation of a Mississippi statute. The statute is ambiguous, and this question has not been addressed by Mississippi courts. Therefore, we respectfully request that the Mississippi Supreme Court respond to the question we certify and allow us to resolve this case without a risk of misinterpretation of this state statute.
Page 2The style of this case is Stanford v. Brandon Nursing and Rehab. Ctr., L.L.C. , No. 24-60509, in the United States Court of Appeals for the Fifth Circuit. The case is on appeal from a judgment of the United States District Court for the Southern District of Mississippi. Federal jurisdiction is based on diversity of citizenship.
In 2022, Mark Stanford was a resident of Brandon Nursing and Rehabilitation Center. While in his room, he flicked a lighter near his oxygen concentrator, starting a fire that caused him to suffer second and third degree burns. The Mississippi State Department of Health found that Brandon Nursing “failed to provide [Stanford] with the supervision that he required to prevent accidents” or “hazards of a facility fire.” The State Department of Health cited Brandon Nursing under federal regulations for long-term care facilities for failing to maintain a safe environment and give each resident the required supervision to prevent accidents.
Through his conservator and guardian, Stanford filed a lawsuit against Brandon Nursing and related entities, alleging negligence and medical malpractice. Brandon Nursing moved to compel arbitration based on a 2017 arbitration agreement that was part of Stanford’s admittance paperwork and was signed by Stanford's brother, Russell Phillips, acting as his health surrogate. Stanford responded to the motion to compel arbitration, arguing that the arbitration agreement was invalid because Phillips was not authorized to sign it on Stanford's behalf. It is undisputed that Stanford lacked capacity to make his own health decisions, and that Stanford had not designated a health surrogate to make decisions for him.
The district court held that Phillips was not a proper statutory surrogate under the relevant Mississippi statute. Accordingly, it denied the Page 3motion to compel arbitration. Brandon Nursing timely appealed. We have jurisdiction over this interlocutory appeal under 9 U.S.C. § 16(a)(1).
We review the district court’s denial of a motion to compel arbitration de novo , its interpretation of state law de novo , and its findings of fact for clear error. Gross v. GGNSC Southaven, L.L.C. , 817 F.3d 169, 175 (5th Cir. 2016). Mississippi law governs this diversity action.
The Federal Arbitration Act (' FAA ') applies to 'nursing-home admissions agreements that contain an arbitration clause.' Adams Cmty. Care Ctr., LLC v. Reed , 37 So. 3d 1155, 1158 (Miss. 2010). “Under the FAA , courts employ a two-pronged inquiry. ‘ The first prong has two considerations: (1) whether there is a valid arbitration agreement and (2) whether the parties’ dispute is within the scope of the arbitration agreement.'' Id. (quoting East Ford, Inc. v. Taylor , 826 So. 2d 709, 713 (Miss. 2002)). Only the first part of the first prong, the validity of the arbitration agreement, is at issue here. Under Mississippi law, the validity of an arbitration agreement is determined by applying ordinary contract law. Reed , 37 So. 3d at 1158. Among the elements of a contract is the requirement that both parties have legal capacity to enter into it. Id.
Mississippi’s Uniform Health-Care Decisions Act governs when and how healthcare surrogates can make decisions on behalf of incapacitated adults. Miss. Code Ann. § 41-41-211. Under the Act, '[a] surrogate may make a health-care decision for a patient who is an adult or emancipated minor if the patient has been determined by the primary physician to lack capacity and no agent or guardian has been appointed or the agent or guardian is not reasonably available.' § 41-41-211(1). If no surrogate has been designated, “any member of the following classes of the patient’s family who is reasonably available, in descending order of priority, may act as surrogate:
Stanford argues that Phillips did not have statutory authority to bind Stanford to the arbitration agreement because Mark Stanford's son was reasonably available and willing to act as surrogate. Stanford argues that the existence of a higher-priority family member who is willing to serve precludes any lower-priority family member from acting as surrogate under Section 4141-211(2). Brandon Nursing argues the statute empowered Phillips, as Stanford’s adult brother, to act as his surrogate.
This case requires the interpretation of Section 41-41-211(2), specifically the phrase “any member of the following classes of the patient’s family who is reasonably available, in descending order of priority, may act as surrogate.” In Mississippi, “the primary rule of [statutory] construction is to ascertain the intent of the legislature from the statute as a whole and from the language used therein.” DePriest v. Barber , 798 So. 2d 456, 458 (Miss. 2001) (quoting Clark v. State ex. rel. Miss. State Med. Ass’n , 381 So. 2d 1046, 1048 (Miss. 1980)). The Mississippi Supreme Court “resorts to the canons of statutory interpretation only where a statute is ambiguous or silent on a specific issue.” Lutz Homes, Inc. v. Weston , 19 So. 3d 60, 62 (Miss. 2009). Mississippi courts require 'a strict interpretation of the Health-Care Decisions Act,” that “follow[s] the plain and unequivocal language of [the Act].' Belhaven Senior Care, LLC v. Smith , 359 So. 3d 612, 618 (Miss. 2023) (second alteration in original) (quotation marks omitted) (quoting Tarvin v. CLC of Jackson, LLC , 193 So. 3d 633, 637–38 (Miss. 2016)).
Both parties assert that the plain language of the statute supports their position and that public policy concerns weigh in their favor. According to Stanford, health care providers are obligated under the statute to ensure “that a resident is receiving the proper medical care authorized by the proper Page 5party.' Stanford argues '[p]riority is unequivocally a precondition to surrogacy.' Therefore, a lower-priority family member cannot serve as surrogate if a higher-priority family member is 'reasonably available.' Brandon Nursing argues “the plain language of the Act requires the opposite conclusion.” It emphasizes the need to read the Act as a whole, citing two provisions as instructive: the provision that allows any individual to disqualify another and the provision that disqualifies all members of an evenly divided class and all members of classes with lower priority from making a decision. § 41-41-211(5), (8). According to Brandon Nursing, any individual in the enumerated list can serve as the surrogate, and priority only matters if a disagreement arises among multiple surrogates. At oral argument, Brandon Nursing acknowledged that its position was not based on the express language of the statute, but instead 'taking the statute as a whole and applying a strict interpretation to the language itself.”
The statute is silent as to this precise point, and we have discovered no Mississippi caselaw addressing whether the existence of a reasonably available, higher-priority family member precludes a lower-priority family member from serving as a surrogate. The 1998 Uniform Health-Care Decisions Act was taken by the Mississippi Legislature from a 1993 uniform law by the same name prepared by the Uniform Law Commission. See 1998 Miss. Laws ch. 542; Unif. Health-Care Decisions Act (Unif. L. Comm’n 1993). Mississippi courts allow for consideration of the official comments to such laws if a provision is ambiguous. See State ex rel. Fitch v. Yazaki N. Am., Inc. , 294 So. 3d 1178, 1185 n.8 (Miss. 2020). On this issue, we found no instructive comments.
The validity of the arbitration agreement turns on whether Phillips was permitted to serve as surrogate even though Stanford had a reasonably available adult son with higher priority under the statute. This question is 'determinative of all or part of [this] cause[,] and there are no clear Page 6controlling precedents in the decisions of the Mississippi Supreme Court” to guide our analysis. Miss. R. App. P. 20(a). 'Therefore, it is appropriate under state law to pose the question[] for certification.” Swindol v. Aurora Flight Scis. Corp. , 805 F.3d 516, 522 (5th Cir. 2015). In addition to state law considerations, our circuit precedents require us to consider three factors when deciding whether to certify a question:
(1) 'the closeness of the question and the existence of sufficient sources of state law'; (2) 'the degree to which considerations of comity are relevant in light of the particular issue and case to be decided”; and (3) “practical limitations of the certification process: significant delay and possible inability to frame the issue so as to produce a helpful response on the part of the state court.”
Id. (quoting Williamson v. Elf Aquitaine, Inc. , 138 F.3d 546, 549 (5th Cir. 1998)).
All three factors favor certifying the question. This is a close question. We have no state caselaw interpreting the provision, and none of the comments to the uniform law address this question. “An Erie guess in these circumstances would be a leap into the dark.” Johnson v. Miller , 98 F.4th 580, 586 (5th Cir. 2024). Though the legal issue has not so far been addressed in Mississippi judicial precedents, its controlling significance in this case, the fact that our answer would be the first from a court for this frequently relevant (if not frequently litigated) question, and the statutory uncertainty cause us to seek the state court's controlling interpretation. Should the Mississippi Supreme Court accept our request, its construction of its own law will allow us to conclude this case in a manner that is not at risk of inconsistency with any later determination by that court.
We respectfully request an answer from that court on this state-law issue of first impression.
Page 7We certify the following question of state law to the Supreme Court of Mississippi:
Does Mississippi’s Uniform Health-Care Decisions Act require health care providers, in the absence of a designated surrogate, to identify qualifying family members and ensure that no such family member with higher priority is reasonably available under the circumstances before relying on a health care decision made by a lower-priority family member?
Our framing of the certified question should not be construed as limiting the scope of the Mississippi Supreme Court’s response to the precise form of the question.
QUESTION CERTIFIED TO THE MISSISSIPPI SUPREME COURT.
Page 8Andrew S. Oldham , Circuit Judge , dissenting:
I am troubled by the Fifth Circuit’s standards for certifying questions to state courts. Certification barely existed at all before the Supreme Court’s decision in Clay v. Sun Insurance Office Ltd. , 363 U.S. 207 (1960). But today it looms large over our diversity docket. And I worry that overuse of the certified-question procedure contravenes history and tradition, undermines federalism, diminishes the parity of state and federal courts, and patronizes our colleagues in the judicial branches of the several States. Our purported “test” for certifying questions is theoretically bankrupt. And the procedure’s much-celebrated efficiency is overstated. We should reconsider these errors, swear off the drink of overcertification, and bring our practices in line with Article III and the jurisdictional statutes enacted by Congress.
This dissent proceeds in four parts. Part I includes a brief history of diversity jurisdiction, abstention, and certified questions. Part II identifies eight theoretical problems with modern certification practices. Part III explains the exceptional circumstances that could, in accordance with Supreme Court precedent, warrant certification. Finally, Part IV explains that my court’s certification practices, including in this case, burden rather than promote federalism.
History first. I (A) begin with the Founding. Then I (B) explain the historical premises behind diversity jurisdiction. Then I (C) turn to Swift , Erie , and federal courts' historical approaches to diversity cases.
At the Founding, almost the entirety of the inferior federal courts' jurisdiction arose in diversity cases.
Page 9The 'source of diversity jurisdiction, like that of so many other features of the new government, is to be found in the famous Virginia plan” presented to the Convention by Edmund Randolph. Henry J. Friendly, The Historic Basis of Diversity Jurisdiction , 41 Harv. L. Rev. 483, 485 (1928). After a short-lived attempt by John Rutledge and Roger Sherman 'to eliminate all lower federal court jurisdiction” in favor of state courts, Scott Dodson, Beyond Bias in Diversity Jurisdiction , 69 Duke L.J. 267, 272 (2019), the Convention instead adopted James Madison’s compromise position. The Madisonian Compromise accepted the Virginia plan’s jurisdictional grants, including diversity. But it left for Congress the decision whether to institute inferior federal courts at all-and all logically subsequent decisions about how much jurisdiction (if any) to vest in those inferior courts. See William Baude et al. , Hart & Wechsler’s The Federal Courts and the Federal System 9–10 (8th ed. 2025) (“ Hart & Wechsler ”).
The Anti-Federalists ardently opposed diversity jurisdiction from the jump. See Friendly, Historic Basis , supra , at 487. For example, Federal Farmer saw no “need of opening a new jurisdiction” so “citizens of different states” could “drag each other many hundred miles into the federal courts.” Federal Farmer III , Letter of Oct. 10, 1787, in 2 The Complete AntiFederalist 234, 243 (Herbert Storing ed., 1981) (“ Storing '). And George Mason thought it ridiculous to create new, inconvenient federal courts to adjudicate claims that extant, convenient state courts already adjudicated: “Can we not trust our State Courts with the decision of [suits between citizens of different states]? . . . Is it suspected that [state courts] would enforce the payment if unjust, or refuse to enforce it if just?” 3 The Debates in the Several State Conventions, on the Adoption of the Federal Constitution, as Recommended by the General Convention at Philadelphia, in 1787 526 (Jonathan Elliot ed., 2d. ed. 1836) (“ Elliot’s Debates ”).
Page 10But the bulk of Anti-Federalist criticism was directed the Madisonian Compromise and the entire notion of inferior federal courts. Mason, for instance, argued that federal courts would “absorb and destroy the judiciaries of the several States,” 1 Elliot’s Debates , supra , at 495, and that the “effect and operation” of the federal courts themselves “[would] be utterly to destroy the state governments,” 3 Elliot’s Debates , supra , at 521. Patrick Henry agreed, proclaiming that “from the extensive jurisdiction of these paramount courts, the state courts must soon be annihilated.” Id. at 542. Brutus developed a similar theme, claiming that federal courts “totally independent of the states, deriving their authority from the United States, and receiving from [the federal government] fixed salaries” would inevitably 'swallow up all the powers of the courts in the respective states' and “eclipse the dignity . . . [and] respectability” of the same. Brutus I , Letter of Oct. 18, 1787, in 2 Storing , supra , at 367. Brutus tied this subversive influence not to any particular grant of jurisdiction, but to the judicial power’s “exalted [station] above all other power in the government.” Brutus XIV , Letter of March 6, 1788, in 2 Storing , supra , at 437. Centinel likewise predicted that “the state courts of justice, like the barony and hundred courts of England, [would] be eclipsed and gradually fall into disuse'-a development he connected to federal question jurisdiction rather than diversity. Centinel II , Letter of Oct. 24, 1787, in 2 Storing , supra , at 148.
The First Congress largely implemented the Madisonian Compromise in the Judiciary Act of 1789. See Friendly, Historic Basis , supra , at 500 (“The first Congress of the United States wasted but little time before taking up the organization of the judiciary.”); David P. Currie, The Constitution in Congress: The Federalist Period 1789– 1801 , at 47 (1997) (The first Judiciary Act 'itself established a number of significant constitutional precedents.”). The Act extended diversity jurisdiction, in relevant part, to cases with an amount in controversy abovePage 11 $500 1 and 'between a citizen of the state where the action was brought and a citizen of another state.” Hart & Wechsler , supra , at 1071. But in a nod to Anti-Federalist opposition, the Act did not grant federal question jurisdiction to federal courts—leaving the decision of most federal claims to the state courts. Id. at 993; see also Tyler S. Moore, Trimming the Least Dangerous Branch: The Anti-Federalists and the Implementation of Article III , 56 Tulsa L. Rev. 1, 33–37 (2020). As a result, in what looks like a complete inversion from our modern perspective, nearly all federal court business consisted of state-law diversity claims, while nearly all federal claims began and ended in state court. See Moore, supra , at 33-37; see also Diego Zambrano, Federal Expansion and the Decay of State Courts , 86 U. Chi. L. Rev. 2101, 2113-15 (2019). It was not until 1875 that Congress gave federal courts general federal question jurisdiction. See Act of March 3, 1875, ch. 137, § 1, 18 Stat. 470, 470 (codified as amended at 28 U.S.C. § 1331).
Given that inferior federal courts were raised on a one-food diet— diversity cases—you might reasonably wonder why the Founders cared so much about creating federal forums for state-law claims. “Why is it that a United States court is given this duty of administering the law of another jurisdiction? Why did the States allow it? Why was it important that the United States should have it?” James Bradley Thayer, The Case of Gelpcke v. Dubuque, 4 Harv. L. Rev. 311, 316 (1891). The available historical materials are short on definitive answers. See 13E Wright & Miller’s Page 12Federal Practice & Procedure § 3601 (3d. ed. 1998) (noting the lack of “any substantial [historical] light on why diversity jurisdiction was granted to the federal courts”).
The traditional explanation is that the Framers were worried that state courts would be biased against out-of-state litigants. See Thayer, supra , at 316. The earliest expression of that view is probably in The Federalist , where Alexander Hamilton supposed that 'the inviolable maintenance of that equality of privileges and immunities to which the citizens of the Union will be entitled, the national judiciary ought to preside in all cases in which one State or its citizens are opposed to another State or its citizens,” because those were 'cases in which the State tribunals cannot be supposed to be impartial.” Federalist No. 80 (Alexander Hamilton), in The Federalist Papers 475, 478 (Clinton Rossiter ed., 1961). And right or wrong, that view was enthusiastically adopted by the Arch-Federalist Chief Justice Marshall:
However, true the fact may be, that the tribunals of the states will administer justice as impartially as those of the nation, to parties of every description, it is not less true that the constitution itself either entertains apprehensions on this subject, or views with such indulgence the possible fears and apprehensions of suitors, that it has established national tribunals for the decision of controversies between aliens and a citizen, or between citizens of different states.
Bank of United States v. Deveaux , 9 U.S. (5 Cranch) 61, 87 (1809); accord Martin v. Hunter's Lessee , 14 U.S. (1 Wheat.) 304, 347 (1816) ('The constitution has presumed (whether rightly or wrongly we do not inquire) that state attachments, state prejudices, state jealousies, and state interests, might sometimes obstruct, or control, or be supposed to obstruct or control, the regular administration of justice.”).
Page 13The evidence supporting that traditional view is thin at best. True, Madison argued in Virginia’s ratifying convention that diversity jurisdiction was necessary because “[i]t may happen that a strong prejudice may arise, in some states, against the citizens of others, who may have claims against them.” 3 Elliot’s Debates , supra , at 533. But as Friendly pointed out in his seminal article, “Madison does not point out any specific examples of prejudice,' nor does he 'allege that any exist.' Friendly, Historic Basis , supra , at 493. George Mason mocked 'the very idea' of such bias as “ridiculous.” 3 Elliot’s Debates , supra , at 526. And Friendly observed that the early state reporters 'entirely fail[] to show the existence of prejudice on the part of the state judges.” Friendly, Historic Basis , supra , at 493. In fact, he concluded that “none have been found which indicate undue prejudice on the part of the local tribunal.” Id. at 494.
If not bias, what then? Friendly argued that “the real fear was not of state courts so much as of state legislatures.' Id. at 495. More specifically, 'a principal reason for the grant of diversity jurisdiction” was “the desire to protect creditors against legislation favorable to debtors.' Id. at 496-97. And unlike Madison’s speculative concerns about bias, there is real evidence to support the protect-all-creditors thesis.2 Friendly also hypothesized that the Page 14Federalists distrusted state courts for additional other reasons: the lack of lifetime tenure; procedures that allowed sometimes-hasty appointments, removal, and recusal; and the existence of legislative review of judicial decisions. None of these “invite[d] confidence” in the quality of state courts. Id. at 497. One can only guess whether Friendly's reading of the history was influenced by his well-known disdain for diversity jurisdiction.3 See Henry Friendly, Federal Jurisdiction: A General View 141 (1973) (criticizing “the diversion of judge-power urgently needed for tasks which only federal courts can handle or which, because of their expertise, they can handle significantly better than the courts of a state”).
Whatever the true reason at the Founding, experience has put paid to the bias myth. On the contrary, empirical evidence suggests that out-of-state bias is “rare and isolated,” Dodson, supra , at 294, and fear of such bias “is only a weak motivation for invoking diversity jurisdiction” in federal court, id. at 296. This hard data confirms what federal courts have long presumed Page 15to be true: that “federal and state courts alike are competent to apply federal and state law.” Mckesson v. Doe , 592 U.S. 1, 5 (2020) (per curiam) (citing Lehman Bros. v. Schein , 416 U.S. 386, 391 (1974)).
The final piece of historical background concerns federal court practice. I explain (1) the federal courts’ approach to general law from the Founding to the 1930s. Then I explain (2) the big bang in Erie . Finally, I discuss (3) postErie developments, including (4) certification.
With only trifling exceptions, inferior federal courts at the Founding heard one and only one type of case—the diversity case. See Part I.A , supra . From the Founding until the 1930s, federal courts adjudicated diversity cases according to the distinction between local law and general law. See, e.g. , Swift v. Tyson , 41 U.S. (16 Pet.)1 (1842); Caleb Nelson, A Critical Guide to Erie Railroad Co. v. Tompkins, 54 Wm. & Mary L. Rev. 921, 925 (2013). A State's local law 'included both its written laws (such as the state constitution and statutes enacted by the state legislature) and at least a portion of its unwritten law,' Nelson, supra , at 925, such as 'longestablished local customs having the force of laws,” Swift , 41 U.S. (16 Pet.) at 18. On questions of local law, federal courts deferred to state courts, “even if they themselves would have read the state statute differently.' Nelson, supra , at 926.
In contrast to local law, 'general law was not attached to any particular sovereign; rather, it existed by common practice and consent among a number of sovereigns.” William A. Fletcher, The General Common Law and Section 34 of the Judiciary Act of 1789: The Example of Marine Insurance , 97 Harv. L. Rev. 1513, 1517 (1984). When answering 'questions of general commercial law,' Swift , 41 U.S. (16 Pet.) at 19, federal Page 16courts did not defer to state court precedent—“even when the highest court of that state had repeatedly expressed its understanding of the proper answer,” Nelson, supra , at 926. State court decisions about the general law were “entitled to . . . the most deliberate attention and respect,” but they did not “furnish positive rules, or conclusive authority.” Swift , 41 U.S. (16 Pet.) at 19. This accorded with the then-universal belief that judicial decisions “are, at most, only evidence of what the laws are, and are not, of themselves, laws.” Id. at 18; see also Andrew S. Oldham, The 25th Annual Joseph Story Lecture: Precedent as a Scatterplot , https://www.youtube.com/watch?v= NL2r97Qrstw (Oct. 22, 2025); Stephen E. Sachs, Finding Law , 107 Calif. L. Rev. 528, 573 (2019).4 So Swift treated judges, whether state or federal, as finders rather than declarers of law. State or federal judges were equally equipped to decipher general law questions, and hence did not defer to one another. Sachs, supra , at 556–59. And this is key: federal courts deferred to state courts on questions of unwritten local law not because those courts created the relevant rules, but because their opinions represented particularly persuasive (even decisive) evidence of what those rules were. Cf. ibid ; 1 James Kent, Commentaries on American Law 475 (3d. ed.
Sir Matthew Hale, The History of the Common Law of England 45 (Charles M. Gray ed. 1971, 3d ed. 1739).
Page 171836) (“A solemn decision upon a point of law . . . becomes an authority in a like case, because it is the highest evidence which we can have of the law applicable to the subject.” (emphasis added)) .
The Big Bang in the federal courts’ universe came in 1938. See Erie R.R. Co. v. Tompkins , 304 U.S. 64 (1938). Erie changed everything. See, e.g. , Abbe R. Gluck, Intersystemic Statutory Interpretation: Methodology as “Law” and the Erie Doctrine , 120 Yale L.J. 1898, 1902 (2011) (describing Erie as “a sea change in how judges view law”). It collapsed the distinction between local and general law, announcing: 'There is no federal general common law.” Erie , 304 U.S. at 78. Erie jettisoned the idea that judges find law—as judges had done for centuries—and replaced it with the Holmesian-realist idea that judges make law. Oldham, supra , at 3:38-5:05; Sachs, supra , at 570– 72. And Erie purported to declare Swift ’s general law approach unconstitutional, 5 holding that Article III recognized only two bodies of law, federal and state. See Erie , 304 U.S. at 78 ('Except in matters governed by the Federal Constitution or by acts of Congress, the law to be applied in any Page 18case is the law of the State.”). And state law is state law, no matter whether it was 'declared by its Legislature in a statute or by its highest court in a decision.” Ibid. So if federal law does not apply, a federal court sitting in diversity must apply state law. And by 'state law,' Erie meant state statutes, constitutions, and common-law doctrines as construed by the relevant State’s courts. Id. at 78-80. That ensures 'the highest court of the [S]tate is the final arbiter of what is state law.” West v. AT&T Co. , 311 U.S. 223, 236 (1940).
Erie ’s approach to law had radical implications for federal courts, the rule of law, and our federal system. The preErie legal order was premised on judicial equality: All judges, state and federal, had equal access to and equal responsibility for finding the general law. Holmes mocked that judicial endeavor and derided the general law as a “brooding omnipresence in the sky.” S. Pac. Co. v. Jensen , 244 U.S. 205, 222 (1917) ( Holmes , J. , dissenting). But unquestionably, that was the job that all judges had performed since at least the sixteenth century. See Oldham, supra , at 14:00– 15:15; Sachs, supra , at 557–58. After Erie , by contrast, the federal court first must assume that its work is to declare, not find, the law. But because it cannot declare state law, see Erie , 304 U.S. at 78, the federal court first pretends that it is not a federal court at all, and is instead “the highest court of the [S]tate [that] would decide' the relevant question. Hart & Wechsler , supra , at 772 (quoting Wright & Kane, Law of Federal Courts § 58, at 356 (8th ed. 2017)).
' Erie 's Austinian legal theory is nowadays thought to be wrongobviously, laughably, friendlessly wrong.' Sachs, supra , at 571 (quotation omitted). But it governs numerous dimensions of federal court practice. And it requires us to make so-called ' Erie guess[es]” in diversity cases. See, e.g. , Temple v. McCall , 720 F.3d 301, 307 (5th Cir. 2013).
Almost immediately, Erie created practical problems. One of the most obvious is this: What should a federal judge do when state law is so unclear that an Erie guess is akin to an Erie dart-throw? One postErie answer was abstention.
Take, for example, Railroad Commission of Texas v. Pullman Co. , 312 U.S. 496 (1941). In that case, the Supreme Court approved “abstaining”— or declining to render a judgment—where a federal constitutional challenge turned on uncertain issues of state law. Id. at 501-02. Pullman abstention spared federal judges from guessing at state law and unnecessarily deciding constitutional questions, while also minimizing the risk of federal decisions upsetting state policies.
Shortly after Pullman came Burford v. Sun Oil Co. , 319 U.S. 315 (1943). In that case, the Court approved abstaining “on grounds of comity with the States when the exercise of jurisdiction by the federal court would disrupt a state administrative process.' 17A Wright & Miller , supra , § 4244 (emphasis added). No state law confusion was necessary; it was enough that Texas law provided bodies for administrative review that specialized in the complicated law of oil and gas. See Burford , 319 U.S. at 334.
Then came Louisiana Power & Light Co. v. City of Thibodaux , 360 U.S. 25 (1959). The case hinged on neither the federal constitution nor a state administrative process, but on the scope of the eminent domain power of a municipality under Louisiana state law. Id. at 25. According to the Court, the eminent domain power was so 'intimately involved with [the] sovereign prerogative” that abstention was necessary to avoid “serious disruption by federal courts of state government or needless friction between state and federal authorities.” Id. at 28.
Page 20Then came Younger v. Harris , 401 U.S. 37 (1971). In that case, Harris was charged in state court for violating the California Criminal Syndicalism Act. Then Harris filed suit under 42 U.S.C. § 1983 in federal court, seeking an injunction to stay the prosecution as violative of the First Amendment. Younger , 401 U.S. at 38–39. Such an injunction would flatly violate the AntiInjunction Act, 28 U.S.C. § 2283 (“ AIA ”), so the Supreme Court refused it. But in explaining that judgment, the Supreme Court invoked the 'longstanding public policy against federal court interference with state court proceedings.” Younger , 401 U.S. at 43. And Justice Black extolled the principles of “Our Federalism,” as reflected in the AIA . Id. at 44. The very next Term, confusingly, the Supreme Court reversed course and held that federal courts could in fact enjoin state court proceedings under § 1983. See Mitchum v. Foster , 407 U.S. 225, 242-43 (1972).6
Finally, the last of the Big Five Abstention Doctrines: Colorado River Water Conservation District v. United States , 424 U.S. 800 (1976). Colorado River turned on a water dispute between the United States and Colorado landowners. Its resulting abstention doctrine “rest[s] on considerations of wise judicial administration, giving regard to conservation of judicial resources and comprehensive disposition of litigation.” Id. at 817 (quotation omitted). And it creates a multi-factor balancing test for abstention, under which no factor is necessary and none is sufficient. See id. at 818-19; Moses H. Cone Mem’l Hosp. v. Mercury Constr. Corp. , 460 U.S. 1, 15–16 (1983).
Page 21While abstention was born of an effort to manage Erie 's fallout, the abstention doctrines created their own problems. For example:
After Meredith , federal judges remained confused by how to decide state-law questions in the face of uncertainty. Chief Justice Warren publicly requested the American Law Institute’s assistance in “achiev[ing] a proper jurisdictional balance between the Federal and State court systems, assigning to each system those cases most appropriate.' Charles Alan Wright, Restructuring Federal Jurisdiction: The American Law Institute Proposals , 26 Wash. & Lee L. Rev. 185, 185 (1969). The ALI proposed, among other things, abolishing diversity jurisdiction entirely. Richard D. Freer, The Political Reality of Diversity Jurisdiction , 94 S. Cal. L. Rev. 1083, 1084 n.2 (2021). A bill to abolish diversity later sailed through the House of Representatives in 1978 but died in the Senate. See S. Res. 1613, 95th Cong., 124 Cong. Rec. 33546 (1978) (proposing to 'strik[e] out' 28 U.S.C. § 1332(a)(1)); H.R. Res. 9622, 95th Cong., 124 Cong. Rec. 5008–09 (1978) (same). Apparently, the problems created by Erie were so bad that at least some folks were ready to revisit the First Judiciary Act and eliminate a class of cases that occupied well-nigh 100 percent of federal dockets at the Founding.
Page 234
Certified questions emerged from this desperation. Florida was the first State to authorize certified questions of state law from federal courts. See Act of June 11, 1945, ch. 23098, 1945 Fla. Laws 1291 (codified at Fla. Stat. § 25.031). But that statute was virtually dormant for two decades. Then, in 1960, the Supreme Court decided Clay . The facts were this: Clay sued the Illinois-based Sun Insurance in a Florida district court after Sun refused to pay benefits, invoking diversity jurisdiction. Sun claimed that Clay’s suit was time-barred, but the district court agreed with Clay that a Florida statute kept his claim alive. On appeal, the Fifth Circuit set aside whether the Florida statute applied and instead held that applying the statute to a contract made in Illinois was unconstitutional. 363 U.S. at 208–09.
The Supreme Court vacated and remanded. Id. at 212. Echoing Burford , Justice Frankfurter rebuked the Fifth Circuit for deciding a complex constitutional choice of law question that could have been avoided by interpreting the Florida statute. Id. at 209-10. And in a first for the Court, Justice Frankfurter instructed the Fifth Circuit on remand to make use of Florida's then-long-dormant certification statute-commending the 'rare foresight' of the Florida legislature, so that the 'federal constitutional question might be mooted.' Id. at 212. On remand, the Fifth Circuit certified the question, the Florida Supreme Court answered, the Fifth Circuit rendered its constitutional holding again, and the Supreme Court reversed— “seven years after suit was begun and four years after certification had first been ordered.” 17A Wright & Miller , supra , § 4248.
And it was off to the races. Almost overnight, the Supreme Court became a cheerleader for certification. See Rebecca A. Cochran, Federal Court Certification of Questions of State Law to State Courts: A Theoretical and Empirical Study , 29 J. Legis. 157, 166 (2003). In Lehman Brothers v. Schein ,Page 24 416 U.S. 386 (1974), for instance, the Court declared that certification “save[s] time, energy, and resources and helps build a cooperative judicial federalism,” id. at 391; see also Bellotti v. Baird , 428 U.S. 132, 150–52 (1976) (quoting Lehman Brothers before ordering certification). By 1985, certification was so firmly ensconced that one member of the Court described reaching state law questions as 'gratuitous,' where the procedure was available. Brockett v. Spokane Arcades, Inc. , 472 U.S. 491, 510 (1985) ( O’Connor, J ., concurring). In 1997, the Court outright declared that “[c]ertification today covers territory once dominated by a deferral device called ‘ Pullman abstention.'' Arizonans for Off. Eng. v. Arizona , 520 U.S. 43, 75 (1997). And the Supreme Court's practice of reversing inferior federal courts for failing to certify-even in the absence of a request from the parties—remains vibrant. See Mckesson , 592 U.S. at 5–6.
The inferior federal courts certainly got the message. By 2020, fortynine states (all but North Carolina), D.C., Puerto Rico, Guam, the Virgin Islands, and the Mariana Islands authorized their courts to answer certified questions. Jason A. Cantone & Carly Giffin, Fed. Jud. Ctr., Certified Questions of State Law: An Examination of State and Territorial Authorizing Statutes 1 (2020). And every federal circuit uses the procedure. See John Macy, Give and Take: State Courts Should Be Able to Certify Questions of Federal Law to Federal Courts , 71 Duke L.J. 907, 922 (2022).
And thus it came to be that federal judges see state-law questions and then say: “Let’s certify!”
While certification was imagined to fix the problems created by Erie guesses and abstention, certification has not proved the 'panacea for resolution of [all] complex or difficult state law questions.” Williamson v. Elf Page 25Aquitaine, Inc. , 138 F.3d 546, 549 (5th Cir. 1998) (quotation omitted). Eight problems (A)–(H) merit emphasis here.
First , certification of state law questions inverts the history and tradition of federal courts. At the Founding, virtually all the inferior courts’ dockets consisted of diversity cases. See Part I.A , supra ; cf. Zambrano, supra , at 2113-15. Contrariwise, the federal courts did not have general federal question jurisdiction—which meant federal cases were heard (overwhelmingly) in state courts. No one considered this problematic in any way. In fact, the most obvious potential Founding-era critics-the AntiFederalists— wanted state courts to decide all federal questions. See Moore, supra , at 36–37. Nor was there any reason to worry because all judges, state and federal, were engaged in the same project of finding law. The federal judge could find state law in a diversity case just as easily as the state judge could find federal law in a federal question case.
Certification, of course, turns this upside down. The procedure relies on Erie 's deeply ahistorical premise that federal judges are custodians of federal law, while state judges are custodians of state law-and never the twain shall meet. See Erie , 304 U.S. at 79 ('[The] only authority is the State . . . and . . . the voice adopted by the State as its own . . . should utter the last word.”). That concern would have been alien to the Founders, who crafted the largely unaltered grant of diversity jurisdiction in 28 U.S.C. § 1332(a) with a different vision for how all judges, state and federal, jointly find law.
Second , and relatedly, certification claims the mantle of 'cooperative federalism” because it leaves state law in state judges’ hands. But it is a poor substitute for truly reckoning with the awesome power federal courts hold Page 26over the States, which we exert largely through our federal question jurisdiction. See 28 U.S.C. § 1331.
The Anti-Federalists understood this, of course. They had plenty to say against Article III , and even against diversity jurisdiction in particular. See Part I.A , supra . Yet for all these grievances, the Anti-Federalists registered zero concerns about federal courts interpreting and applying state law. The closest one gets is Patrick Henry’s orthogonal concern that diversity jurisdiction would require judges residing near state borders to 'be acquainted with all the laws of the different States.'3 Elliot’s Debates , supra , at 542. By contrast, the Anti-Federalists had plenty to say about the existence of federal courts, sitting in diversity or otherwise. They feared that federal courts would crowd out state courts, see 2 Storing , supra , at 148, that a ballooning federal government would expand federal question jurisdiction, see id. at 367, and that the growing national government would then subordinate state governments, see 3 Elliot’s Debates , supra , at 521. Of course, much of that doomsaying has come to pass. See Zambrano, supra , at 2125–51 (empirically linking federal expansion to the decay of state courts). Brutus and Federal Farmer would find hollow solace in the assurance: “Yes, your worst dreams will come true, but federal courts will (sometimes) avoid interpreting state law!”
Third , certification is difficult to reconcile with our fundamental obligation to decide cases. Federal courts must decide validly presented questions of law, rather than avoid them. When Congress created a federal forum for diversity cases, it expected those cases to be heard in our courts. By shifting part of the decisional task to a state tribunal, certification undermines this congressionally conferred responsibility and “thwart[s] the purpose of the jurisdictional act.” Meredith , 320 U.S. at 235. In that sense, Page 27certified questions perpetuate all the problems of abstention. See Burford , 319 U.S. at 336 ( Frankfurter , J ., dissenting) (arguing that federal court refusal to decide state law questions 'disregard[ed] a duty enjoined by Congress and made manifest by the whole history of the jurisdiction of the United States courts based upon diversity of citizenship between parties”); Thibodaux , 360 U.S. at 32 ( Brennan , J ., dissenting) (arguing that federal court refusal to decide state law questions “passe[d] beyond disrespect for the diversity jurisdiction to plain disregard of this imperative duty”).
Moreover, think about the bias thesis for granting diversity jurisdiction to inferior federal courts in the first place. As noted above, there is scant evidence (at the Founding or after it) to suggest state courts are biased against out-of-State parties. See Part I.B , supra . But suppose for a moment that Madison and Chief Justice Marshall were right that such bias exists and poses a threat to out-of-State litigants. If a litigant seeks to avoid bias in state courts, why subject them to the 'state attachments, state prejudices, state jealousies, and state interests” they fled by filing in federal court? Hunter’s Lessee , 14 U.S. (1 Wheat.) at 347. Now suppose the bias thesis is false and that state courts can be trusted just as much as federal courts. If a litigant can trust the state court’s impartiality in deciding a certified question, why do we need a federal forum in the first place? Put simply, the certification device dilutes the assurance of an unflagging, independent federal forum that diversity jurisdiction was designed to provide.
Fourth , certification can disrespect state sovereignty rather than promote it. State courts are our 'equal partners, not junior partners' in deciding cases. McMillan v. Amazon.com, Inc. , 983 F.3d 194, 202 (5th Cir. 2020). If 'equality' means anything, it means not 'abdicat[ing]' our obligation to decide questions of state law because they are somehow better Page 28relegated to state courts. Barnes v. Atl. & Pac. Life Ins. Co. of Am. , 514 F.2d 704, 705 n.4 (5th Cir. 1975). On the contrary, “equal partners” means that “federal and state courts alike are competent to apply federal and state law.” Mckesson , 592 U.S. at 5 (quotation omitted). Indeed, even Erie assumes “that the bases of state law are . . . communicable by the parties to a federal judge no less than to a state judge.” Salve Regina Coll. v. Russell , 499 U.S. 225, 239 (1991). And our “consistent history of hospitable acceptance of concurrent jurisdiction” likewise shows that state courts are fully competent to decide questions of federal law. Tafflin v. Levitt , 493 U.S. 455, 466 (1990) (quotation omitted).
Certifying questions to state courts patronizes them. It “draws from a limited reservoir of comity' by 'shift[ing] the difficult work of deciding [cases] to the state court, which is often so busy keeping its own house in order that it scarcely has time for our overflow laundry.” Kremen v. Cohen , 325 F.3d 1035, 1043-44 (9th Cir. 2003) ( Kozinski , J ., dissenting). A certified question is not a transfer of a case to state court. The state court does not assume jurisdiction over the case. Instead, the state court is asked to give “an advisory opinion on an abstract question of law” and send it back to help us decide the case. Perez v. City of San Antonio , 715 S.W.3d 709, 731 (Tex. 2025) ( Sullivan , J ., dissenting) (quotation omitted). It is only the federal court that 'will resolve the parties' controversy and enter the final judgment.” Bruce M. Selya, Certified Madness: Ask A Silly Question . . . , 29 Suffolk U. L. Rev. 677, 682 (1995). The state courts are, in effect, treated as adjuncts of the federal courts. Cf. Crowell v. Benson , 285 U.S. 22, 51 (1932) (noting federal judges can sometimes employ adjuncts without violating Article III ).
Perez illustrates this condescension. The question in that case was whether San Antonio's plan to redevelop a public park impermissibly burdened the religious exercise of Native Americans. Perez v. City of San Page 29Antonio , 98 F.4th 586, 591–93 (5th Cir. 2024). Our court initially said no, id. at 591, before withdrawing that opinion upon a motion for rehearing, Perez v. City of San Antonio , 115 F.4th 422, 423 (5th Cir. 2024). The panel then certified a broad question about the “scope” of Article I, § 6-a of the Texas Constitution to the Texas Supreme Court.7 Section 6-a provides that the State “may not enact, adopt, or issue a statute, order . . . or rule that prohibits or limits religious service.” Tex. Const. art. I, § 6-a. To avoid deciding the host of questions pregnant in the certification order (What constitutes a policy? A prohibition? A religious service?), the Texas Supreme Court contoured their answer to the facts of the case: whatever else it means, § 6-a 'does not extend to governmental actions for the preservation and management of public lands.” Perez , 715 S.W.3d at 730. Back in this court, the panel again sided with San Antonio, largely for the same reasons it supplied the first go-round. Compare Perez v. City of San Antonio , 150 F.4th 430, 453–54 (5th Cir. 2025), with Perez , 98 F.4th at 612–13.
Certification benefited no one. The litigants spent a year arguing over whether § 6-a was essentially unlimited in scope—a question we could have answered with ordinary tools of interpretation. See Perez , 715 S.W.3d at 718– 19 (citing, inter alia , Bryan A. Garner , Garner’s Dictionary of Legal Usage 568, 954 (3d. ed. 2011)). And the receipt of a helpful answer turned on the Texas Supreme Court’s good grace to not “really explore the studio space” with every imaginable application of the four-year-old, neverbefore-interpreted constitutional provision. Id. at 731 ( Sullivan, J ., Page 30dissenting) (quoting Saturday Night Live: More Cowbell (NBC television broadcast Apr. 8, 2000)).
Diversity cases are not an opportunity to ask state courts, like our law clerks, to 'go about researching a point of state law.' Lehman Bros. , 416 U.S. at 394 ( Rehnquist , J ., concurring). But as distinguished jurists have pointed out, federal courts sometimes let that attitude bubble to the surface. See Guido Calabresi, Federal and State Courts: Restoring a Workable Balance , 78 N.Y.U. L. Rev. 1293, 1302 (2003) ('[W]hen the New York Court of Appeals declines certification, some federal judges walk around saying, ‘What did they do to us? After all, we are the Second Circuit, they should listen to us!’”). And it belies the idea that state and federal judges are “equal partners' in finding law. Think about what would happen if a state court certified a federal question to us : We'd need fainting couches and therapy dogs in the John Minor Wisdom Courthouse. Moreover, it belies the idea that state and federal judges are “equal partners” when inferior federal judges can certify questions to the State’s highest court: Our intermediate federal courts are not equal to the State’s Supreme Court. Compare, e.g. , U.S. Const. art. III § 1 (not creating inferior federal courts), with Miss. Const. art. 6, § 144 (creating State Supreme Court). So if we really believed in 'equal partners,' only the Supreme Court of the United States would certify questions-as that Court has ordered. See, e.g. , Clay , 363 U.S. at 212; Mckesson , 592 U.S. at 5-6; see also Justin R. Long, Against Certification , 78 Geo. Wash. L. Rev. 114, 127 (2009) ('If the federal and state judiciaries are to be treated as parallel, the U.S. Supreme Court (which has final authority over federal law) is logically the only analog to state high courts (which have final authority over state law).').8
Fifth , it is unclear that our certification practice complies with Article III of the federal Constitution.
Article III vests federal courts with the judicial power-and only the judicial power. In exercising that power, federal courts must act, well, judicially . The Supreme Court has told us the sorts of ingredients that, at a minimum, make something “judicial”: procedural protections and valid final judgments with res judicata effect. Ortiz v. United States , 585 U.S. 427, 438 (2018). And the Supreme Court has told us the sorts of things that make something non-judicial. We know, for example, that administering pension benefits to veterans is not judicial. See Hayburn’s Case , 2 U.S. (2 Dall.) 408, 409 (1792). We know that rendering opinions that are subject to legislative revision is not judicial. See Plaut v. Spendthrift Farm, Inc. , 514 U.S. 211, 226– 27 (1995). We know that it is not judicial to answer political questions. See Nixon v. United States , 506 U.S. 224, 228–29 (1993). We know that it is not judicial for a court to merely certify an opinion on monetary claims against the United States to an executive officer. See Gordon v. United States , 69 U.S. 561, 561 (1864). And, as most relevant here, we know that issuing advisory opinions is not judicial. See Letter of Aug. 8, 1793, from John Jay, C.J. & Assoc. JJ., U.S. Sup. Ct., to George Washington, President, in 3 The Correspondence and Public Papers of John Jay 488, 488-89 (Henry P. Johnston ed., 1891). Thus, the Court has set out the various attributes that make a proceeding judicial within the meaning of Article III .
Page 32State courts, of course, are not governed by Article III . That's because state courts derive their power from their respective States' constitutions. And States are free to structure their constitutions and their judiciaries however their citizens deem proper. The People of Texas, for example, have vested their State Supreme Court with inherent “administrative powers.” Webster v. Comm’n for Law. Discipline , 704 S.W.3d 478, 489–90 (Tex. 2024). State courts can answer political questions. See, e.g. , Nat Stern, Don’t Answer That: Revisiting the Political Question Doctrine in State Courts , 21 J. Const. L. 153 (2018). (States can be so comfortable with their judges answering political questions that they can make state judges answer them outside the courthouse, in partisan political elections.) And a number of States authorize their courts to issue advisory opinions. See Hart & Wechsler , supra , at 75. Indeed, as a matter of state law, a State could authorize its courts to do all sort of non-judicial things that federal courts cannot—like administering veterans’ benefits. Cf. Hayburn’s Case , 2 U.S. (2 Dall.) at 409 (prohibiting Article III courts from such non-judicial functions).
For a state proceeding to enter the federal court system, however, it must be “judicial” in the Article III sense. Consider Article III ’s limitations on the Supreme Court's appellate jurisdiction to review 'other Cases.' U.S. Const. art. III, § 2, cl.2. That jurisdiction extends to state-court “cases” that are “within the purview of the Constitution.” Hunter’s Lessee , 14 U.S. (1 Wheat.) at 342. This standard limits what the Supreme Court can review: Yes to judicial disputes, and no to everything else. For example, in Gordon , because the Court of Claims could only 'certify its opinion to the Secretary of the Treasury,” the Supreme Court concluded that the claims court's judgments were not 'final and conclusive' and thus outside of Article III 's 'judicial power.' 69 U.S. at 702. Therefore, the Supreme Court had no appellate jurisdiction over the non-judicial decisions. See ibid.
Page 33('Congress cannot extend the appellate power of [the Supreme] Court beyond the limits prescribed by the Constitution.”). So a state proceeding can get into federal court only if it's judicial in the Article III sense.9
It is unclear how certification fits within this framework. When a federal court asks a state court to answer a certified question, it is requesting that the state court act extra-judicially. That’s because the federal court is asking the state court to issue “an advisory opinion on an abstract question of law”; the state court’s answer is neither binding on the federal court nor the parties. Perez , 715 S.W.3d at 731 ( Sullivan , J ., dissenting) (quotation Page 34omitted). Thus, the certification process features an Article III court— which can engage only in judicial acts-asking a state court to do a nonjudicial act by rendering an advisory opinion. The Supreme Court has even found it reversible error for a federal court not to ask a state supreme court to engage in such extra-judicial decisionmaking. See Mckesson , 592 U.S. at 5. But I cannot think of another example of Article III allowing (and Supreme Court precedent requiring) federal judges to participate in non-judicial decisionmaking. Imagine the most extreme form of non-judicial decisionmaking—flipping a coin to decide whether to affirm or reverse. It would not matter whether the coin happened to land on the right answer: Throwing the case to chance would itself abdicate the judicial duty and constitute non-judicial decisionmaking. Cf. Ortiz , 585 U.S. at 438 (holding procedural protections are part of what makes a proceeding judicial in the Article III sense). So if some forms of non-judicial decisionmaking (like coin flips) make a proceeding non-judicial, why are other forms of non-judicial decisionmaking (like advisory opinions answering certified questions) ok?
Further, the mere fact that a state court might call its answer to a certified question “judicial” does not solve this problem. Many thoughtful state jurists take pride in responding to our certified questions. And their courts put considerable effort into answering them: they request briefing, hold argument, and write thoughtful opinions. They might even publish those opinions in leather-bound case reporters. But such processes cannot make the opinion a judicial act in the Article III sense. Cf. Hunter’s Lessee , 14 U.S. (1 Wheat.) at 338 (“It is the case , then, and not the court , that gives the jurisdiction.'). Instead, the meaning of Article III is a question of federal law independent from state-court process. And the Article III problem in Hayburn’s Case was not that the judge failed to take the pensioners’ claims seriously; that he didn’t hear the evidence fairly; that he didn’t adjudicate the claims while wearing a robe; that he didn’t explain his decisions; &c. SeePage 35 2 U.S. (2 Dall.) at 409. The problem was that the act of administering a veterans’ benefit program is not judicial in the Article III sense-just as the act of issuing an advisory opinion is not judicial in the Article III sense. See ibid. ; see also Gordon , 69 U.S. at 561; Ortiz , 585 U.S. at 438.
One possible answer is that the Article III problem vanishes when the state-court advisory opinion becomes part of a federal-court judgment. But this answer proves too much. Erie explains that state law can be made by either a state supreme court 10 or a state legislature. 304 U.S. at 78. In fact, it says that the distinction between the two bodies is “not a matter of federal concern.” Ibid. So can a federal appellate court certify a legal question for non-judicial resolution in the state legislature ? The federal-court judgment would then cleanse that decision, too. Or suppose that the People of a State pass a constitutional amendment giving the Governor, State Attorney General, or the State Bar President the power to answer certified questions. Cf. Tex. Const. art. IV, § 22 (authorizing the State Attorney General to render legal opinions). If the point of Erie is to identify the content of state law on the terms that the State and its People direct, cf. 304 U.S . at 78, could a federal court certify a question to the Texas attorney general? If the answer is no, why can a federal court certify questions for state judges to just-asclearly and just-as-extra-judicially 'give legal advice'?11
Page 36All to say, the certified-question practice raises a host of Article III questions. And we should at least try to answer them before asking our statecourt colleagues to issue advisory opinions.
F
Sixth , certification interferes with the normal state court appellate review process by bypassing the lower state courts. When a federal court certifies a question, it plucks issues out of the state judicial hierarchy and presents them to the state’s highest court in the abstract, without the benefit of intermediate review or a fully developed factual record, as in Perez . Worse, in States like Texas, it circumvents the state high court's discretionary review process. The Supreme Court of Texas “has accepted more certified questions from the Fifth Circuit” in some years “than it has granted petitions for review from all but one of the 14 intermediate appellate courts.” Michelle Casady, A Question on Certified Questions: Analyzing the Recent Increase in Queries from the Fifth Circuit , The Texas Lawbook (Aug. 4, 2023), https://perma.cc/87NF-K3BY. The disparity is so great that one appellate lawyer in Texas remarked, because '[i]t's very difficult to get one of the special spots on the Texas Supreme Court’s plenary docket, . . . [l]itigants have to take notice of [certified questions] as a potential pathway of getting to the Texas Supreme Court.' Ibid. And this dynamic is not limited to Texas. See, e.g. , W. Helicopter Servs., Inc. v. Rogerson Aircraft Corp. , 811 P.2d 627, 633 (Or. 1991) (“Because of considerations of comity, . . . [w]e may, on occasion, Page 37accept certification of questions that, were they tendered to us in a traditional petition for review, we would decline to address.'); Leiter Mins., Inc. v. California Co. , 132 So. 2d 845, 849–50 (La. 1961) (“Therefore, out of respect for, and as a courtesy to, that court, we proceed to [a decision], in the hope that our opinion will be of some assistance to the United States Supreme Court.”).
By transforming the noble goal of comity into a Fast Pass for litigants, certification distorts state court dockets and priorities. It favors out-of-State parties (in federal court) over local litigants who must climb the normal ladder of state appeals. It cuts out state trial judges. And it creates perverse incentives for forum shopping, as defendants can remove into federal court, test the proverbial waters, and then seek certification to the state high court for appellate review they would otherwise rarely get.
Seventh , federal courts’ use of certification is hopelessly procedurally irregular. The Supreme “Court has not laid down a clear test for when federal courts should avail themselves of the certification option, and patterns appear to vary among the circuits.” Hart & Wechsler , supra , at 1429; see also 17A Wright & Miller , supra , § 4248. One panel may find an issue clear enough to make an Erie guess; another may disagree. Who or what governs whether a state-law question needs clarification? After all, “ambiguity is in the eye of the beholder and cannot be readily determined on an objective basis.' Wooden v. United States , 595 U.S. 360, 378 (2022) ( Kavanaugh , J ., concurring). As a result, whether to certify a state-law issue is “more a question of the considerable discretion of the federal court in going about the decisionmaking process' than a decision to maintain federal-state comity. Lehman Bros. , 416 U.S. at 394 ( Rehnquist , J ., concurring). Such an arbitrary trigger generates inconsistent and Page 38unpredictable answers-a cruel irony given Erie and its progeny’s promise to avoid “accident[s] . . . disturb[ing] equal administration of justice.” Klaxon Co. v. Stentor Elec. Mfg. Co. , 313 U.S. 487, 496 (1941). Worse, it invites federal courts to certify questions that just feel “icky,” beneath their station, easy to delegate, or some combination of all three. So, we end up with unwritten nonrules about when to certify. That is not only patronizing to state courts, but simply unfair to litigants.
Eighth , and finally, even if I were to cede the conceptual minefield above to certification's defenders, the device still fails to deliver its one practical promise: efficiency. Instead of streamlining litigation, certification 'prolong[s] the dispute and increase[s] the expenses incurred by the parties.” Mckesson , 592 U.S. at 5. The procedure splits a single controversy into two separate proceedings in two court systems, creating piecemeal litigation and the need for more briefing and argument. See Selya, supra , at 690. It therefore inherently “entails more delay and expense than would an ordinary decision of the state question on the merits by the federal court.” Lehman Bros. , 416 U.S. at 394 ( Rehnquist , J ., concurring); see also Clay , 363 U.S. at 224 ( Black , J ., dissenting) (noting in certification context that litigants “have a right to have their lawsuits decided without unreasonable and unnecessary delay or expense”); Thibodaux , 360 U.S. at 43 ( Brennan , J ., dissenting) (“I cannot escape the conclusion in these cases that delay in the reaching of a decision is more important to those parties than the tribunal which ultimately renders the decision.”). As one scholar noted in 1969, “[i]n federal courts it takes as long to determine that a case should be certified as it does to decide a case on the merits.” Brian Mattis, Certification of Questions of State Law: An Impractical Tool in the Hands of the Federal Courts , 23 U. Miami L. Rev. 717, 726 (1969).
Page 39That remains true today.
The Supreme Court has instructed that certification can be appropriate-but only '[i]n exceptional instances.' Mckesson , 592 U.S. at 5. What circumstances are exceptional? The Supreme Court has given us two examples: (1) avoidance of a substantial question of federal constitutional law, see Pullman , 312 U.S. at 500; Clay , 363 U.S. at 212; Mckesson , 592 U.S. at 5; and (2) when answering a state law question 'hazards . . . serious disruption by federal courts of state government,” Thibodaux , 360 U.S. at 28, see also Burford , 319 U.S. at 333–34. If answering a state law question poses either of these risks, then certification may be necessary. But beyond these circumstances, mere 'difficult[y]' in determining state law should not 'afford a sufficient ground for a federal court to decline to exercise its jurisdiction.” Meredith , 320 U.S. at 234. Answering difficult legal questions is our job—just as much as it is the job of our colleagues in the several States’ judiciaries. A few recent cases certifying questions illustrate how these standards work in practice.
First, take Mckesson . There, a badly injured police officer sought damages from DeRay Mckesson under Louisiana law on the theory that Mckesson negligently directed a protest. Mckesson , 592 U.S. at 2. The question before this court was whether holding Mckesson liable would violate the First Amendment if Louisiana law otherwise imposed liability for it. Id. at 3. A panel of our court held that Louisiana law did impose liability and that the First Amendment did not stand in the way. Id. at 3-4. The Supreme Court vacated and remanded so we could certify the state-law liability question to the Louisiana Supreme Court. Id. at 6. The Court reasoned that this court need not answer the First Amendment issue if the question of Louisiana law came out the other way. See id. at 5. Thus, certification served Page 40the same purposes of avoidance that abstention served in Pullman . See ibid. On remand, we certified the question accordingly. See Doe v. Mckesson , 2 F.4th 502, 504 (5th Cir. 2021) (per curiam).
Second, take Whole Woman’s Health v. Jackson , 23 F.4th 380 (5th Cir. 2022). On remand from the Supreme Court, we certified a question to the Supreme Court of Texas about whether Texas law authorizes various state officials “to take disciplinary or adverse action of any sort against individuals or entities that violate the Texas Heartbeat Act.” Id. at 389. The answer to that question would determine whether plaintiffs could seek an injunction under Ex parte Young against state officials enforcing the Act. Id. at 383. Federal court injunctions against an array of state officials would undoubtedly 'hazard[] . . . serious disruption by federal courts of state government.” Id. at 389; Thibodaux , 360 U.S. at 28; Valentine v. Collier , 993 F.3d 270, 293 (5th Cir. 2021) ( Oldham , J ., concurring in judgment) (emphasizing federalism costs associated with Ex parte Young actions). So it was essential to clarify the scope of state law before aggravating the federalism costs and allowing a suit to proceed in federal court.
Finally, we recently certified a question to the Supreme Court of Texas to avoid abstention. In Umphress v. Hall , 133 F.4th 455 (5th Cir. 2025) (per curiam), a county judge who refused to officiate same-sex marriages challenged the application of Canon 4A(1) of the Texas Code of Judicial Conduct, which “requires Texas state judges to conduct their extra-judicial activities in a manner that does not call into question their impartiality.” Id. at 460. The panel certified the question of whether refusing to conduct samesex weddings on religious grounds violates Texas’s Canon 4A(1). Id. at 461. That certification served two goals. First, to sidestep the First Amendment question of whether the Canon could constitutionally apply to Umphress. Second, to give Texas a chance to review its own state administrative law before a federal court enjoined a state commission.
Page 41This case falls nowhere near the standard for appropriate certification. I begin (A) with the Fifth Circuit’s certification “standard.” Then I (B) turn to this case.
Our “standard” for certifying questions to state courts is standardless. To wit, the Fifth Circuit applies three “factors” when deciding whether to certify a question:
(1) the closeness of the question and the existence of sufficient sources of state law; (2) the degree to which considerations of comity are relevant in light of the particular issue and case to be decided; and (3) practical limitations of the certification process: significant delay and possible inability to frame the issue so as to produce a helpful response on the part of the state court.
McMillan , 983 F.3d at 202.
On the first prong, the “closeness” of a question is not “a sufficient ground for a federal court to decline to exercise its jurisdiction.” Meredith , 320 U.S. at 234. What was true of abstention in Meredith is just as true for certification-whatever their differences, both devices run afoul of our jurisdictional duty, so the same strictures should apply. Moreover, answering 'close' questions is not just the job of an appellate court; it is the entire point of an appellate court. It is why we have three-judge panels, life tenure, salary protection, and the opportunity to devote our entire lives to the law. Our job is not to pawn questions off to state courts when we would rather not expend the energy answering them.
Further, 'considerations of comity' do not counsel in favor of certifying questions. See Part II , supra . That's because state courts do not Page 42owe us a “helpful response.” McMillan , 983 F.3d at 202. The only relevant comity consideration is whether federal court resolution of a state law question risks interfering with a State’s governance, as it could have in Whole Woman's Health and Umphress . And no matter the efficiency of our state court colleagues, it is hard to imagine a case where certification would not 'prolong the dispute and increase the expenses incurred by the parties.' Mckesson , 592 U.S. at 5. Common sense dictates that there will always be significant “delay” when certifying a question, compared with “an ordinary decision of the state question on the merits by the federal court.” Lehman Bros. , 416 U.S. at 394 ( Rehnquist , J ., concurring).
In sum, our certification standard bears no resemblance to a defensible standard for certifying questions to our colleagues in the state courts.
Rather than meeting the proper standard for certification, this case perfectly illustrates my concerns with the whole enterprise.
The question presented is how to apply Mississippi surrogacy law to a private dispute between Brandon Nursing and Mark Stanford. There’s no hint of a constitutional question. Cf. Pullman , 312 U.S. at 500; Mckesson , 592 U.S. at 4. And there's no chance of interfering with Mississippi’s sovereignty, whether in an administrative proceeding or some other core state power. Cf. Thibodaux , 360 U.S. at 28; Burford , 319 U.S. at 333–34. If we interpret “health surrogate” under the statute one way, the parties will go to arbitration. If we interpret it the other way, the parties will litigate the plaintiff’s tort claims in federal court. In so describing the case, I do not mean to belittle it—on the contrary, I would treat Mr. Stanford’s invocation of our jurisdiction with respect by exercising it. I only emphasize that the case presents no exceptional reason to certify anything to anyone.
Page 43In addition to not meeting the exceptional-circumstances criterion for certification, there are a host of other reasons not to certify anything to the Mississippi Supreme Court.
First, no one is asking us to do it. And it’s easy to understand why. The delay will be prejudicial to both parties. The Mississippi Supreme Court could tell us to pound sand—which will bring us right back where we started. Or maybe the State Supreme Court will tell us, however long from now, whether Stanford’s brother was a legitimate surrogate under Miss. Code § 41-41-211(2). Even armed with an answer to that question, we would still have to determine whether Stanford’s adult son was “reasonably available.” Ibid. If Stanford's brother was a legitimate surrogate or Stanford's adult son was not reasonably available, we would hold the arbitration agreement valid; if not, we would hold it invalid. If the agreement is valid, the parties will go to arbitration; if the agreement is invalid, the parties will return to litigating in the district court. Stanford suffered his burns in January 2022. It is already the end of 2025. When will Stanford hope to get to judgment—2030?
Moreover, while I maintain that the “closeness” of a question should not matter, the legal issues in this case are not “close” in any relevant sense. The absence of 'state caselaw' or 'comments to the uniform law address[ing],” ante , at 6, this exact situation does not render the legal knot Gordian, or else any uncertain question of law would be close enough to certify. The parties have fully briefed this case, relying on analogous Mississippi court precedents, cases from other states with similar Uniform Act statutes, and traditional textual arguments. We also have a well-reasoned district court opinion below. The tools to decide this case are here, and we should use them.
Finally, put yourselves in the shoes of the Mississippi Supreme Court justices. What will they think when they get today's certification order? This Page 44case implicates a purportedly “uniform” state law proposed by the National Conference of Commissioners on Uniform State Laws in 1993. Over the last 32 years, that purportedly 'uniform' law has been adopted by only five States: Alaska, Maine, Mississippi, New Mexico, and Wyoming. See Uniform Law Commission, Health-Care Decisions Act 2023 (Map), https://perma.cc/5B3S-6ERJ. It has generated precisely zero appellate decisions in Mississippi, which is why both parties have litigated the case by invoking or distinguishing a fifteen-year-old intermediate appellate decision from New Mexico. See Corum v. Roswell Senior Living, LLC , 248 P.3d 329 (N.M. Ct. App. 2010). If I were a statewide elected official in Mississippi, and I was told to drop whatever else the people of that good State tasked me to do, to focus instead on interpreting a “uniform” law that apparently has fewer adopters than Bluesky, and to decide whether to follow an intermediate appellate decision from New Mexico at the behest of federal judges in New Orleans, I can tell you this much: I would not think the exercise beneficial to federalism.
I respectfully dissent.
1 The '(then) sizable amount in controversy requirement' was a direct response to Anti-Federalist concerns about federal courts forcing citizens “to travel great distances to access the courts,” because it ensured that only expensive controversies would require such effort. Tyler S. Moore, Trimming the Least Dangerous Branch: The Anti-Federalists and the Implementation of Article III , 56 Tulsa L. Rev. 1 , 34 (2020) (quotation omitted).
2 For example, the Anti-Federalists worried that the point of the new federal courts was to make southern States amenable to suit by creditors in northern States over Revolutionary War debts. See Brutus XIII , Letter of Feb. 21, 1788, in 2 Storing , supra , at 429 (predicting the clause would “subject[] a state to answer in a court of law, to the suit of an individual.”); Federal Farmer III, Letter of Oct. 10, 1787, in 2 Storing , supra , at 245 (predicting that 'this new jurisdiction will subject the states' to suits based on 'promises made during the war.'). The Federalists responded that the fears were overblown. See Federalist No. 81 (Alexander Hamilton), in The Federalist Papers , supra , at 487. And of course, the Anti-Federalists were right: It took a New York minute for a creditor to sue a southern State over a war debt. See Chisholm v. Georgia , 2 U.S. (2 Dall.) 419 (1793). And another New York minute for the People to vindicate the AntiFederalists by passing the Eleventh Amendment. See Cutrer v. Tarrant Cnty. Loc. Workforce
Dev. Bd. , 943 F.3d 265, 269 (5th Cir. 2019), as revised (Nov. 25, 2019) (“That’s how we got the Eleventh Amendment.”).
3 Friendly was not alone: contempt for diversity jurisdiction was common among federal judges in the early-to-mid twentieth century. See, e.g. , Robert H. Jackson, The Supreme Court in the American System of Government 37 (1955) ('In my judgment the greatest contribution that Congress could make to the orderly administration of justice in the United States would be to abolish the jurisdiction of the federal courts which is based solely on the ground that the litigants are citizens of different states.”); Lumbermen’s Mut. Cas. Co. v. Elbert , 348 U.S. 48, 54 (1954) ( Frankfurter, J. , concurring) (decrying “the mounting mischief inflicted on the federal judicial system by the unjustifiable continuance of diversity jurisdiction”); id . at 56 (“A legal device like that of federal diversity jurisdiction which is inherently, as I believe it to be, not founded in reason, offers constant temptation to new abuses.”); Earl Warren, Address by Chief Justice of the United States , 36 A.L.I. Proc . 27–33 (1959) (opining that balanced federalism “may very well require further restrictions on diversity jurisdiction'); Warren E. Burger, Annual Report on the State of the Judiciary , 62 A.B.A. J. 443, 444 (1976) ('[D]iversity cases have no more place in the federal courts . . . than overtime parking tickets or speeding on the highways.”).
4 This point derives from Hale's observations on the common law of England: It is true, the Decisions of the Courts of Justice, tho’ by Virtue of the Laws of this Real they do bind, as a Law between the Parties thereto, as to the particular Case in Question, ’till revers’d by Error or Attaint, yet they do not make a Law properly so called, (for that only the King and Parliament can do); yet they have a great weight and Authority in Expounding, Declaring, and Publishing what the Law of this Kingdom is, especially when such Decisions hold a Consonancy and Congruity with Resolutions and Decisions of former Times; and tho’ such Decisions are less than a Law, yet they are a greater Evidence thereof than the Opinion of any private Persons, as such, whatsoever.
5 Erie grounded its constitutional conclusion in legal positivism, the 'mandate that all law-including the common law'does not exist without some definite authority behind it.’” Michael Steven Green, Erie ’s Suppressed Premise , 95 Minn. L. Rev. 1111, 1113 (2011) (quotation omitted). But “[p]ositivism is compatible with the Swift ian notion that . . . common-law standards are a . . . question about which . . . courts of other sovereigns, including federal courts, may come to their own judgment.” Ibid. ; see also Jack Goldsmith & Steven Walt, Erie and the Irrelevance of Legal Positivism , 84 Va. L. Rev. 673, 685 (1998) ('Swift supporters . . . believed the general common law applied by federal courts was . . . authorized by the Constitution . . . They too were legal positivists.”); Craig Green, Repressing Erie ’s Myth , 96 Cal. L. Rev. 595, 606 (2008) ('Simply put, positivism does not require Erie 's rule that common-law substantive authority be transferred from federal judges to state judges.”); cf. H.L.A. Hart, The Concept of Law 101 (2d. ed. 1994) (“Custom and precedent . . . owe their status of law . . . to the acceptance of a rule of recognition which accords them this independent though subordinate place.').
6 On 'the awkward pas de deux of Mitchum and Younger ,” see James E. Pfander & Nassim Nazemi, The Anti-Injunction Act and the Problem of Federal-State Jurisdictional Overlap , 92 Tex. L. Rev. 1, 46 (2013). And on what could have driven the Court to reach such incommensurate holdings a year apart, compare Paul M. Bator, The State Courts and Federal Constitutional Litigation , 22 Wm. & Mary L. Rev. 605, 620-21 (1981), with Richard H. Fallon, Jr., The Ideologies of Federal Courts Law , 74 Va. L. Rev. 1141, 1164-72 (1988).
7 Specifically, it asked whether 'the ‘ Religious Service Protections’ provision of the Constitution of the State of Texas—as expressed in Article 1, Section 6-a—impose a categorical bar on any limitation of any religious service, regardless of the sort of limitation and the government's interest in that limitation?' Perez , 115 F.4th at 428.
8 Professor Long points out that 'one of the insurance cases that the Fifth Circuit felt was important enough to be conclusively and authoritatively determined by the state
high court was not important enough to garner a published, precedential decision from the federal court itself.” Long, supra , at 130 (citing XL Specialty Ins. Co. v. Fin. Indus. Corp. , No. 06-51683, 2009 WL 1532047, at *1 (5th Cir. June 1, 2009)).
9 ASARCO Inc. v. Kadish , 490 U.S. 605 (1989), is not to the contrary. The plaintiffs in that case did not have Article III standing to sue in federal court. Id. at 612. They brought suit instead in state court. Id. at 610. The state trial court then rendered judgment for the plaintiffs, as it was entitled to do: “[T]he constraints of Article III do not apply to state courts, and accordingly the state courts are not bound by the limitations of a case or controversy or other federal rules of justiciability even when they address issues of federal law . . . .” Id. at 617. The Supreme Court then held that it had Article III appellate jurisdiction to review the state court's judgment even if the state court rendered that judgment without satisfying the prerequisites of Article III (like the standing rules). Id. at 618. Why? Because the party invoking federal jurisdiction—the defendant who lost in state court—was injured by the state court's judgment. Ibid. (noting the state court's judgment invalidated the defendant's valuable mineral leases). And the defendant's injury was sufficient to satisfy Article III ’s injury-in-fact requirement, even if the plaintiff in state court did not have Article III standing when he brought suit. See ibid.
So, you might ask, doesn't ASARCO obviate any Article III concerns with certification of questions to state courts because it shows that state-court business can enter federal courts even when the state courts operated free from Article III 's constraints? That reading of ASARCO proves too much and too little. It’s too much because if injury from a non-judicial order was enough to satisfy Article III , then Article III would also allow parties injured by administrative agencies to petition the Supreme Court for relief from them. And no one thinks you can do that. See, e.g. , Marbury v. Madison , 5 U.S. (1 Cranch) 137, 175 (1803) (holding the Supreme Court has no Article III jurisdiction to review the decision of an executive official, namely the Secretary of State); cf. Chandler v. Jud. Council of the Tenth Cir. of U.S. , 398 U.S. 74, 88-89 (1970) (noting Article III probably does not allow Supreme Court to review decision of administrative judicial tribunal, even when it unquestionably hurt the petitioner). ASARCO also proves too little because the essential fact in that case is that the state court rendered a judgment that hurt the defendant. See 490 U.S. at 618; see also Ortiz , 585 U.S. at 438 (one essential attribute of a 'judicial' act in the Article III sense is that the court renders a judgment that carries res judicata and rights-altering effects). Obviously, advisory opinions like the state courts’ answers to our certified questions are unaccompanied by judgments.
10 In directing federal courts to consider decisions from the State's highest court, Erie presumably meant that federal courts should consider state courts' judicial decisions. After all, certification did not become a phenomenon until 22 years after Erie . See Clay , 363 U.S. at 212. So there is no reason to think that Erie itself contemplated that federal courts could ask state courts for non-judicial advisory opinions.
11 Consider how this plays out practically in a case like Mckesson . That case implicated both a federal claim (First Amendment) and a state-law dispute (negligence). If Mckesson had arisen from the Louisiana state courts, the Supreme Court of the United States would obviously have statutory appellate jurisdiction to review both the federal and the state-law claims. See 12 U.S.C. § 1257. But would the result be different if a federal
court asked the state court, via certified question, to render a non-judicial advisory opinion on how state negligence law interacts with the First Amendment? Cf. Doe v. Mckesson , 339 So. 3d 524, 548 (La. 2022) ( Griffin, J. , dissenting) (dissenting from the State Supreme Court’s answers to the Fifth Circuit’s certified questions because they pit state negligence law against the First Amendment). The Supreme Court of the United States presumably would still have statutory jurisdiction to review the state court’s advisory opinion under § 1257—but would lack constitutional jurisdiction to review it under Article III .