RECOMMENDED FOR PUBLICATION Pursuant to Sixth Circuit I.O.P. 32.1(b)
File Name: 26a0244p.06
UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT
┐
DENNIS O’CONNOR, and all those similarly situated, │
Plaintiff-Appellant, │ > No. 25-2104 │
v. │ │
RACHAEL EUBANKS and TERRY STANTON, in their │ personal capacities; STATE OF MICHIGAN, │ Defendants-Appellees. │ ┘
Appeal from the United States District Court for the Eastern District of Michigan at Bay City.
No. 1:21-cv-12837—Nancy G. Edmunds, District Judge.
Argued: April 22, 2026
Decided and Filed: August 26, 2026
Before: MOORE, THAPAR, and NALBANDIAN, Circuit Judges.
_________________
COUNSEL
ARGUED: Philip L. Ellison, OUTSIDE LEGAL COUNSEL, PLC, Hemlock, Michigan, for Appellant. James A. Ziehmer, OFFICE OF THE MICHIGAN ATTORNEY GENERAL, Lansing, Michigan, for Appellees. ON BRIEF: Philip L. Ellison, OUTSIDE LEGAL COUNSEL, PLC, Hemlock, Michigan, Matthew E. Gronda, GRONDA PLC, Saginaw, Michigan, for Appellant. James A. Ziehmer, B. Thomas Golden, Brian K. McLaughlin, OFFICE OF THE MICHIGAN ATTORNEY GENERAL, Lansing, Michigan, for Appellees.
THAPAR, J., delivered the opinion of the court in which MOORE and NALBANDIAN, JJ., concurred. NALBANDIAN, J. (pp. 12–17), delivered a separate concurring opinion.
No. 25-2104 O’Connor v. Eubanks, et al. Page 2
_________________
OPINION
_________________
THAPAR, Circuit Judge. Michigan took custody of Dennis O’Connor’s unclaimed financial accounts under the state’s Uniform Unclaimed Property Act. And it didn’t give O’Connor the interest those accounts produced while the state held them. So O’Connor sued and argued that Michigan’s actions violated his due-process rights. The district court rejected his claims after finding that Michigan had taken ownership of his accounts. But under Michigan law, the original owners of presumptively abandoned property continue to own that property— even after the state takes custody of it. We thus vacate the district court’s order and remand for further proceedings.
I.
Michigan’s Uniform Unclaimed Property Act dictates what happens to unclaimed property in the state. It provides that “all property, including any income or increment derived from the property,” left unclaimed for a set period is “presumed abandoned.” Mich. Comp. Laws § 567.223(1); id. §§ 567.225–567.237. After the holder of the unclaimed property delivers it to the state, the state “assumes custody and responsibility for [its] safekeeping.” Id. § 567.241(1). At that time, the state must issue public notices about the property to alert potential owners to its status. Id. § 567.238(1), (2); id. § 567.239(1), (3). But owners don’t always respond to those notices.
“[F]unds received under this act” are deposited in Michigan’s “general fund.” Id.
§ 567.244(1). For other types of property, such as physical property or securities, the state treasurer publishes a notice of sale and liquidates the assets if no one claims the property. She then deposits the “proceeds from the sale” in the general fund. Id. In either case, the treasurer then subtracts any administrative expenses upon deposit into the general fund. Id. § 567.244(1), (2). And as the money sits in the general fund, it may earn interest for the state.
If the original owner asks for his property back, the state must return it—but not necessarily any interest the property earned. The state pays the owner interest only if the
No. 25-2104 O’Connor v. Eubanks, et al. Page 3
property was interest bearing when it was delivered to the state. Id. § 567.245(3) (requiring the state to pay “interest at a rate of 6% a year or any lesser rate the property earned while in the possession of the holder”). But the state doesn’t pay interest if the assets weren’t interest bearing when Michigan took custody. Id. § 567.243(3); O’Connor v. Dep’t of Treasury, 9 N.W.3d 351, 357 (Mich. Ct. App. 2023). Instead, the state keeps any interest the assets earned while they sat in its custody.
Dennis O’Connor had non-interest-bearing financial accounts with two private companies. One account was worth between $100 and $250, while the other was worth less than $100. After the companies lost touch with O’Connor, they turned over his accounts to Michigan under the Act. The state took possession of both accounts and eventually deposited the money into its general fund.
O’Connor believed those actions violated his constitutional rights. So he brought suit against Michigan and two state officials in federal court. He sued the officials under § 1983, alleging violations of the Fourteenth Amendment’s Due Process Clause and the Fifth Amendment’s Takings Clause. See 42 U.S.C. § 1983. And he sued Michigan directly for alleged violations of the Takings Clause. He brought these claims individually and on behalf of a putative class.
The district court dismissed all of O’Connor’s claims. After O’Connor appealed, we affirmed the dismissal of O’Connor’s takings claims. O’Connor v. Eubanks (O’Connor I), 83 F.4th 1018, 1022, 1024 (6th Cir. 2023). But we vacated the dismissal of his due-process claims against the state officials and remanded for further proceedings. Id.
On remand, the district court again dismissed O’Connor’s due-process claims. The district court reasoned that while Michigan had taken ownership of O’Connor’s accounts, it had provided him sufficient process. And because the state had ownership of those accounts, the court concluded that Michigan also owned the interest that the accounts generated. With all of O’Connor’s claims dismissed, the district court denied class discovery.
O’Connor timely appealed. He argues the district court erred by rejecting his due-process claims against the state officials and denying class discovery.
No. 25-2104 O’Connor v. Eubanks, et al. Page 4
II.
We review the district court’s dismissal of O’Connor’s claims de novo. Wheaton v.
McCarthy, 800 F.3d 282, 285 (6th Cir. 2015). And we review the district court’s denial of class discovery for abuse of discretion. Pub. Int. Legal Found. v. Benson, 136 F.4th 613, 622 (6th Cir. 2025).
Under the Fourteenth Amendment, states must provide citizens with “due process of law”
before depriving them of property. U.S. Const. amend. XIV, § 1. O’Connor contends that Michigan’s officials deprived him of two types of property without sufficient process: (1) the original funds in his accounts (the principal), and (2) the interest that those accounts generated while they were in Michigan’s custody. Both arguments turn on whether Michigan actually took ownership of the interest or the principal while holding it in custody under the state’s Unclaimed Property Act.
A.
Start with the principal. To determine what counts as someone’s property, we often look to “existing rules or understandings that stem from an independent source such as state law.” Bd. of Regents of State Colls. v. Roth, 408 U.S. 564, 577 (1972). But state law isn’t the only source of law defining property rights. Tyler v. Hennepin County, 598 U.S. 631, 638 (2023). That’s because allowing a state to “simply exclude from its definition of property any interest that [it] wished to take” would render constitutional property protections a “dead letter.” Id. (quoting Hall v. Meisner, 51 F.4th 185, 190 (6th Cir. 2022)). So courts must consider existing state law alongside other sources such as “traditional property law principles,” historical practice, and Supreme Court precedent in order to define property interests. Id. (quotation omitted).
Based on the Act’s text and the history of Michigan property law, O’Connor kept ownership of the principal even after Michigan took custody of it. The statutory text repeatedly refers to the state’s control over unclaimed items as “custody,” not ownership. Mich. Comp. Laws §§ 567.224, 567.238, 567.241, 567.246. While the property is in the state’s custody, Michigan “assumes . . . responsibility for [its] safekeeping.” Id. § 567.241(1). And, tellingly, the statute consistently refers to the rightful holders of the unclaimed property as “owners,” not
No. 25-2104 O’Connor v. Eubanks, et al. Page 5
“former” or “previous” owners. Id. §§ 567.223, 567.224, 567.238, 567.242, 567.243(4), 567.244(1), 567.245(3)–(4), 567.252(1), 567.254(5), 567.256. That language suggests the original owner retains title to the property.1 Consistent with that understanding, the Act provides that property is “presumed abandoned” because the owner is unknown to the State—not actually abandoned because the owner has relinquished title. Id. § 567.223(1) (emphasis added); see Cerajeski v. Zoeller, 735 F.3d 577, 581 (7th Cir. 2013). All told, this text strongly implies that O’Connor retained ownership of his accounts while they were in the state’s custody.
The Act’s text also accords with Michigan’s common-law presumption: Owners retain their property rights even when the state holds their property and doesn’t know their identity. The Act recognizes the difference between an “escheat” (state ownership) and a “custodial taking” (no state ownership). Mich. Comp. Laws § 567.224(d). And Michigan’s courts frequently interpreted “escheat” laws dating back to the nineteenth century that allowed the state to take possession of property owned by an individual who died without known heirs. See Evans Prods. Co. v. Fry, 12 N.W.2d 448, 452 (Mich. 1943). Before the state took ownership of such property, Michigan escheat laws provided a period in which the state merely took custody “in the capacity of a conservator for the benefit of any person lawfully entitled to it.” Braun v. McPherson, 269 N.W. 211, 213 (Mich. 1936); accord Evans, 12 N.W.2d at 452. Thus, any heirs of the deceased owner had time to claim their property before the state took title. Braun, 269 N.W. at 213. This regime didn’t deprive owners “of property without due process of law” because “any one who establishe[d] his right [could] claim and receive his property or the net proceeds” thereon from the state. Evans, 12 N.W.2d at 458–59 (emphasis added).2 So state custody didn’t mean state ownership.
1For what it’s worth, the Act’s legislative background reinforces this understanding. The Uniform Law Commission’s own notes on the 1995 Act, which Michigan adopted, are clear that “the State does not take title to unclaimed property, but takes custody only, and holds the property in perpetuity for the owner.” Unif. Unclaimed Prop. Act, Refs. & Annos. (1995); see also Commonwealth Edison Co. v. Vega, 174 F.3d 870, 872 (7th Cir. 1999) (noting that under the predecessor “Uniform Unclaimed Property Act of 1981 . . . [t]he state . . . is merely a custodian”).
2Michigan points out that, under the state’s historical escheat regime, the term “escheated” was defined as the state taking title to property. See Evans, 12 N.W.2d at 453. It’s true that the state would eventually take ownership of property by “escheat” if it was never claimed by any heirs. See id. But in the meantime, the property was only “alleged[ly] escheated”: held in the state’s custody so the “person lawfully entitled to it” had the chance to assert his claim. Id. at 459. And if an heir did assert his claim, then “there [was] no escheat.” Braun, 269 N.W. at
No. 25-2104 O’Connor v. Eubanks, et al. Page 6
And that common-law reasoning extended to other types of unclaimed or abandoned property. Michigan courts noted that the state would take “custody” of “unclaimed, uncalled-for, abandoned property, or that of missing, disappeared persons” in order “to conserve” it “for the benefit of the actual owner if he claims it at some future time.” Id. (emphasis added). The state’s focus on the rights of the actual owner shows that Michigan viewed itself as assuming custody but not ownership over that property.
When Michigan established a statutory program to handle unclaimed property, it didn’t displace this common-law presumption. In 1947, the state legislature entitled Michigan to “take charge of all matters pertaining to lands or other property” abandoned by its owner. 1947 Mich. Pub. Acts 329, § 1. The code specified that “[a]bandoned property” would be held by the state until a claim for the property was “filed by the owner.” Id. § 33. Crucially, as the Michigan Supreme Court explained, the statutory relationship between the state and owner was “custodial in nature,” so the code effectuated “no deprivation of property rights of an owner.” Schoener v. Cont’l Motors Corp., 106 N.W.2d 774, 777 (Mich. 1961). Thus, under the Act’s predecessor statute, the owner of presumptively abandoned property kept ownership of it.
Michigan courts have found that statutes don’t displace common-law principles unless the state legislature has spoken “in no uncertain terms.” Yang v. Everest Nat’l Ins. Co., 968 N.W.2d 390, 396 (Mich. 2021) (cleaned up). Far from clearly displacing the common law, the Act and its predecessor statute accord with the common-law rule that the original owners retain ownership of property while it’s in Michigan’s custody. That means O’Connor continued to own the principal while his accounts were in state custody.
B.
Given that O’Connor owned the principal, the next question is whether he owned the interest that Michigan generated from his accounts. In O’Connor I, we explained, “When the government takes custody of private property and earns interest on it, that interest belongs to the owner.” 83 F.4th at 1023. That’s because of the longstanding common-law rule that “interest
212. So prospective claimants continued to own the property while it sat in state custody—at least, until the time period for an escheat ended.
No. 25-2104 O’Connor v. Eubanks, et al. Page 7
follows principal.” Phillips v. Wash. Legal Found., 524 U.S. 156, 165 (1998) (“The rule that ‘interest follows principal’ has been established under English common law since at least the mid-1700’s.”). And the Supreme Court has repeatedly reaffirmed that interest generated by private property in state custody belongs to the original owner. See City of New Orleans v. Fisher, 180 U.S. 185, 197 (1901); Webb’s Fabulous Pharmacies, Inc. v. Beckwith, 449 U.S. 155, 161–62 (1980). We reaffirm that principle again here.
The rule that interest follows principal applies regardless of whether the principal was interest bearing at the time the state took custody. See O’Connor I, 83 F.4th at 1023 (citing Webb’s Fabulous Pharmacies, 449 U.S. at 156–57); Phillips, 524 U.S. at 172. After all, the interest is generated by property that still belongs to the original owner—even if the property didn’t start off producing interest. So the interest can’t be taken as a “forced contribution” to the “general governmental revenue.” Webb’s Fabulous Pharmacies, 449 U.S. at 163. O’Connor thus retained ownership of the interest generated by his funds while they were in Michigan’s custody.
C.
The district court reached the opposite conclusion on both points by concluding that the state owned both the principal and the interest. And reasoning from that basis, the district court proceeded to reject O’Connor’s due-process claims. But that false premise about who owned the property tainted the rest of the district court’s analysis.
To see why, consider the district court’s logic. First, the district court rejected O’Connor’s argument that Michigan deprived him of the principal without due process. As a threshold matter, the district court agreed that the state deprived O’Connor of the principal. In doing so, it relied on a Michigan Court of Appeals case decided after O’Connor I, which held the state takes ownership of presumptively abandoned property—i.e., the principal—when it takes custody over that property under the Act. See Kemerer v. State, --- N.W.3d ----, No. 362055, 2024 WL 4609911, at *9 (Mich. Ct. App. Oct. 29, 2024), appeal denied, 21 N.W.3d 201 (Mich. 2025). Despite that deprivation, the district court concluded that the existence of the Act itself and the notices that Michigan routinely publishes about unclaimed property gave O’Connor
No. 25-2104 O’Connor v. Eubanks, et al. Page 8
sufficient process. So it found that the state officials hadn’t violated O’Connor’s due-process rights.
Second, the district court determined that Michigan hadn’t deprived O’Connor of the interest on his accounts. The district court again followed Kemerer, which instructed that because the state owned the principal, it also owned any interest generated by that principal. So the district court concluded that O’Connor hadn’t been deprived of the interest because he never owned it in the first place. Without deprivation of a protected property interest, the court held that the state didn’t violate O’Connor’s due-process rights for the interest, either.
Those holdings both depend on the premise that the state took ownership of O’Connor’s property when it took custody over his accounts. But that’s wrong. As discussed above, Michigan law dictates that the original owner retains ownership of his property—along with the interest it generates—even after the state takes custody. And the district court shouldn’t have departed from the correct reading of the law by following Kemerer.
Although federal courts generally “may not disregard a decision of the state [intermediate] appellate court on point,” we may apply a different rule if “‘convinced by other persuasive data that the highest court of the state would decide otherwise.’” Kochins v. Linden- Alimak, Inc., 799 F.2d 1128, 1140 (6th Cir. 1986) (quoting Clutter v. Johns-Manville Sales Corp., 646 F.2d 1151, 1153 (6th Cir. 1981)); see also Hendershot v. Stanton, 162 F.4th 625, 629–30 (6th Cir. 2025). That’s true even in states like Michigan, Tennessee, and Kentucky that have only one intermediate appellate court. See Allstate Ins. Co. v. Thrifty Rent-A-Car Sys., Inc., 249 F.3d 450, 454 (6th Cir. 2001) (Michigan); Kochins, 799 F.2d at 1140 (Tennessee); Wallace Hardware Co. v. Abrams, 223 F.3d 382, 396 (6th Cir. 2000) (Kentucky). Although this puts a thumb on the scale in Kemerer’s favor, we need not follow that decision if there is “any clear indication” that the case was erroneously decided or contrary to past decisions of the state’s supreme court. Ruth v. Bituminous Cas. Corp., 427 F.2d 290, 293 (6th Cir. 1970). Here, the
No. 25-2104 O’Connor v. Eubanks, et al. Page 9
available data clearly indicate that the Michigan Supreme Court wouldn’t follow either of Kemerer’s holdings. Hendershot, 162 F.4th at 630.3
First, as discussed above, Kemerer’s holding that the state takes ownership under the Act is inconsistent with both the statute’s text and the Michigan Supreme Court’s interpretation of longstanding property-law principles.4 In fact, the Michigan Court of Appeals itself adopted a custodial view of the Act in dicta more than a decade before Kemerer. Flint Cold Storage v. Mich. Dep’t of Treasury, 776 N.W.2d 387, 392 (Mich. Ct. App. 2009) (noting that the Act “provides a mechanism by which the state may hold certain unclaimed property in trust for the benefit of the rightful owner”). And decisions interpreting other states’ versions of the Act have uniformly held that it creates a custodial scheme. Dani v. Miller, 374 P.3d 779, 794 (Okla. 2016); Clark v. Strayhorn, 184 S.W.3d 906, 911 (Tex. Ct. App. 2006); Canel v. Topinka, 818 N.E.2d 311, 325 (Ill. 2004); La. Health Serv. & Indem. Co. v. Tarver, 635 So. 2d 1090, 1099 (La. 1994); State v. Elsinore Shore Assocs., 592 A.2d 604, 606 (N.J. Super. Ct. App. Div. 1991); State ex rel. Marsh v. Neb. St. Bd. of Agric., 350 N.W.2d 535, 539 (Neb. 1984); Boswell v. Citronelle-Mobile Gathering, Inc., 294 So. 2d 428, 432 (Ala. 1974). Kemerer represents an outlier holding that the Michigan Supreme Court likely wouldn’t follow.
Second, Kemerer erred by concluding that the original property owner doesn’t own the interest generated while the property is in state custody. 2024 WL 4609911, at *8. If Michigan owned the accounts, then it would also own the interest they generated. Id. at *9. But because Kemerer is wrong that Michigan owns the property, it’s also wrong that Michigan owns the
3Michigan suggests that its highest court’s refusal to review Kemerer makes it more likely it would adhere to that decision if asked. True, “state appellate court precedent may be particularly persuasive where the state supreme court denies leave to appeal,” but such a denial has never been held to tie our hands completely. Hendershot, 162 F.4th at 630 (citing Ruth, 427 F.2d at 293); cf. Maryland v. Balt. Radio Show, Inc., 338 U.S. 912, 919 (1950) (statement of Frankfurter, J.) (noting that the United States Supreme Court’s “denial of a petition for a writ of certiorari . . . carries with it no implication whatever regarding the Court’s view on the merits of a case which is has declined to review”).
4Kemerer does try to grapple with parts of the Act. For example, it contends that Michigan’s ability to sell assets in its custody under the Act confirms that the state owns them. See Kemerer, 2024 WL 4609911, at *9 (citing Mich. Comp. Laws § 567.243(1), (4)). But the Kemerer court failed to mention that Michigan must still repay the full value of that property to the original owner if he asserts title to the assets after the sale. Mich. Comp. Laws § 567.245. The state’s duty to repay any proceeds from the sale sounds more like custody than title.
No. 25-2104 O’Connor v. Eubanks, et al. Page 10
interest. The Michigan Supreme Court is thus unlikely to agree with Kemerer’s flawed conclusion.
D.
In short, the district court erred by concluding that Michigan took ownership of the principal—and the interest it generated—in O’Connor’s accounts while they were in state custody. And because of that threshold error, the rest of its due-process analysis was incorrect.
If Michigan had taken ownership of the principal, that would present different issues. A prospective plaintiff would have two potential constitutional claims. First, he could bring a takings claim, alleging that Michigan had taken the principal without just compensation.5 Second, he could bring a due-process claim, alleging that the state had taken his principal without providing adequate process before doing so.
But if the state took ownership of the principal, that plaintiff probably couldn’t bring a separate due-process or takings claim with respect to any interest generated after the state took ownership of the property. That’s because, as discussed above, interest follows principal. So if the state owned the principal, it would also own any interest generated. And that would mean the state probably hadn’t “deprived” the plaintiff of his property, so there would be no due- process violation. Likewise, the plaintiff wouldn’t have a takings claim because he didn’t own the interest in the first place.
We need not resolve these thorny issues here because O’Connor retained ownership of his property—and the interest it generated—while it was in state custody. Accordingly, we remand for the district court to perform the due-process analysis starting from that correct premise.
That analysis will require the district court to decide what due-process framework governs Michigan’s refusal to return O’Connor’s interest. When the district court assumed Michigan took ownership of the principal, it found that Texaco, Inc. v. Short controlled
5Of course, O’Connor’s takings claim has already been dismissed from this case. See O’Connor I, 83 F.4th at 1022.
No. 25-2104 O’Connor v. Eubanks, et al. Page 11
O’Connor’s claim that he was deprived of that principal without due process. R. 59, Pg. ID 934– 39 (citing Texaco, Inc. v. Short, 454 U.S. 516 (1982)). But on remand, the district must decide in the first instance whether Texaco applies to the interest.
One final wrinkle: If the district court finds that O’Connor’s due-process claims should proceed, it must also reconsider whether to grant O’Connor’s request for class discovery.
* * *
We vacate the district court’s dismissal of O’Connor’s due-process claims. We remand the case for further proceedings consistent with this opinion.
No. 25-2104 O’Connor v. Eubanks, et al. Page 12
__________________
CONCURRENCE
__________________
NALBANDIAN, Circuit Judge, concurring. I agree with the majority that Dennis O’Connor has a viable due-process claim to his interest. I also agree that the district court will need to determine the correct due-process framework for that claim on remand. But I write separately to make two further points: first, why I’d treat Kemerer as binding in a world without Hendershot; and second, why I don’t think Texaco establishes the amount of process due in this case because of the degree of state involvement in the unclaimed-property process.
I.
Because I wouldn’t stray from Hendershot, I agree with the majority’s reasoning that the state never exercised ownership over O’Connor’s principal. My agreement is based on our decision to disregard the Michigan Court of Appeals’ opposite holding in Kemerer v. State, __ N.W.3d __, 2024 WL 4609911, at *9 (Mich. Ct. App. 2024), which we deemed out-of-step with Michigan Supreme Court jurisprudence in Hendershot v. Stanton, 162 F.4th 625, 629–31 (6th Cir. 2025).
But I write separately to explain why I think Kemerer’s opposite holding should control our analysis—Hendershot notwithstanding—because it states the law of Michigan as applied by Michigan’s courts. Kemerer supplies one legal rule applicable to this case because the Rules of Decision Act, 28 U.S.C. § 1652, requires federal courts exercising diversity jurisdiction to “sit[] as a state court.” See BMW Stores, Inc. v. Peugeot Motors of Am., Inc., 860 F.2d 212, 214 (6th Cir. 1988) (quoting Comm’r of Internal Revenue v. Estate of Bosch, 387 U.S. 456, 465 (1967)); Grant v. Bill Walker Pontiac-GMC, Inc., 523 F.2d 1301, 1303 (6th Cir. 1975) (citing Erie R.R. Co. v. Tompkins, 304 U.S. 64 (1938)). State courts across Michigan are, in turn, bound by the Michigan Court of Appeals’ published, unreviewed decisions—like Kemerer. Mich. Ct. R. 7.215(C)(2); Esordi v. Township of Macomb, __ N.W.3d __, 2025 WL 2495908, at *5 (Mich. Ct. App. 2025). So if a state court must follow Kemerer, we must too.
No. 25-2104 O’Connor v. Eubanks, et al. Page 13
To see why Kemerer should apply, we start with the Rules of Decision Act’s text. That statute, first enacted as part of the Judiciary Act of 1789, provides that “[t]he laws of the several states, except where the Constitution or treaties of the United States or Acts of Congress otherwise require or provide, shall be regarded as rules of decision in civil actions in the courts of the United States, in cases where they apply.” 28 U.S.C. § 1652. The category of “cases where they apply” includes those, like this one, arising under the federal courts’ diversity jurisdiction. Advey v. Celotex Corp., 962 F.2d 1177, 1181 (6th Cir. 1992) (“Under the Erie doctrine, federal courts are to apply state substantive law in diversity cases.” (citing Erie R.R. Co. v. Tompkins, 304 U.S. 64 (1938)).
But figuring out what Congress meant when it referred to “[t]he laws of the several states” has bedeviled the courts since the founding era. See, e.g., Robinson v. Campbell, 16 U.S. 212, 221–23 (1818) (addressing whether “the laws of the several states” binding on federal courts include the states’ legal or equitable remedies). In its most influential early exposition on the Act, the Supreme Court held that the “laws of the several states” didn’t include state common law, at least insofar as state common law conflicted with “general commercial law”—i.e., “not the law of a single country only,” but rather, at least in “questions of a more general nature,” the law “of the commercial world.” Swift v. Tyson, 41 U.S. 1, 18–19 (1842). Over the following decades, federal courts expanded general law’s domain until little state common law was left to control in federal diversity cases. See Jack Landman Goldsmith & Curtis Bradley, General Law Revivalism and the Problem of 1938, at 11–12 (U. Chi. L. Sch., Pub. L. & Legal Theory Research Paper No. 26-5); 11 Mark V. Tushnet, The Oliver Wendell Holmes Devise: History of the Supreme Court of the United States, The Hughes Court 857 (2021) (describing the “incremental[]” expansion of general law into areas customarily governed by state law); see also Erie, 304 U.S. at 79 (describing the “rule declared in Swift” as “rest[ing] upon the assumption” that federal courts can assess general-law questions independently of state court decisions). But nearly a century later, the Supreme Court repudiated the Swift doctrine in Erie, 304 U.S. at 71– 78, one of its “most . . . consequential decisions.” Ramos v. Louisiana, 590 U.S. 83, 117–18 (2020) (Kavanaugh, J., concurring in part). There, the Supreme Court declared that “federal courts exercising jurisdiction in diversity of jurisdiction cases would apply as their rules of decision the law of the state, unwritten as well as written.” Erie, 304 U.S. at 72–73.
No. 25-2104 O’Connor v. Eubanks, et al. Page 14
Despite Erie’s command to follow state common law in diversity cases, neither it nor its progeny set out a universal methodology for reading state common law. That’s perhaps unsurprising. After all, state common law flows from state courts. And states have “great latitude to establish the structure and jurisdiction of their own courts.” See Howlett ex rel. Howlett v. Rose, 496 U.S. 356, 372 (1990). So two state judiciaries could plausibly have very different ways of expounding their common law, defying a one-size-fits-all approach for applying state law under Erie.
Indeed, instead of attempting to apply a universal approach to state common law, federal courts most aptly describe their approach to diversity cases as simply requiring them to “sit[] as a state court.” See BMW Stores, 860 F.2d at 214 (citation modified). So we’ve got a relatively straightforward question to ask ourselves: How would a Michigan court treat Kemerer? Michigan’s rules of procedure answer that question clearly. Kemerer, as a published opinion of the Michigan Court of Appeals, “creates binding precedent” for courts across the state, including subsequent panels of the Court of Appeals, “until the Michigan Supreme Court enters a decision altering . . . [the] decision[] or its rationale.” Esordi, 2025 WL 2495908, at *5. In other words, the published and unreviewed decisions of the Michigan Court of Appeals are the “[t]he law . . . of the state[]” binding on federal courts in diversity cases. 28 U.S.C. § 1652. And Kemerer falls into that category.
Hendershot disregarded Kemerer by applying the oft-repeated formulation of the Erie rule limiting federal courts’ adherence to intermediate state appellate authority to the extent that it “predict[s] what the state supreme court will do.” Hendershot, 162 F.4th at 630. And we’ve repeated that formulation in cases applying Michigan law. See, e.g., Allstate Ins. Co. v. Thrifty Rent-A-Car Sys., Inc., 249 F.3d 450, 454 (6th Cir. 2001); Kingsley Assocs., Inc. v. Moll PlastiCrafters, Inc., 65 F.3d 498, 509 (6th Cir. 1995). But that formulation only makes sense insofar as it helps us discover “[t]he law . . . of the state[].” 28 U.S.C. § 1652. And in many states, it probably does—but not Michigan. In Ohio, for example, intermediate state appeals courts’ published and unreviewed decisions on an unsettled question don’t bind every other state court—only the courts within their geographic jurisdiction. See Stapleton v. Holstein, 723 N.E.2d 164, 166 (Ohio Ct. App. 1998) (“Only Ohio Supreme Court decisions and reported
No. 25-2104 O’Connor v. Eubanks, et al. Page 15
opinions of this court are binding upon trial courts of this district.”). So we’d only treat them as persuasive, not conclusive. But given Michigan’s judicial structure, there’s no point measuring a published and unreviewed decision of the Michigan Court of Appeals against our own prediction of what the Michigan Supreme Court might do, because the Michigan Court Rules already designate the former as binding statewide law. See Mich. Ct. R. 7.215(C)(2); Esordi, 2025 WL 2495908, at *5.1 In other words, we can apply instead of predict.
So because Erie requires us to treat state law the same way a state court would, I’d argue Kemerer has bound us since the day it was published. But because we’re more directly bound by our opposite conclusion in Hendershot, a published opinion, I concur with the majority.2
II.
Turning to the district court’s task on remand: the majority is right that the district court will have to decide for the first time the proper due-process framework for O’Connor’s claim to his interest. But I write separately to offer some thoughts as to why I think Texaco, Inc. v. Short, 454 U.S. 516 (1982), is off the table.
The Fourteenth Amendment prohibits states from “depriv[ing] any person of . . . property . . . without due process of law.” U.S. Const. amend. XIV, § 1. That command embodies a “flexible concept that varies with [the] particular situation.” Zinermon v. Burch, 494 U.S. 113, 127 (1990). So “[d]ifferent circumstances call for different processes.” Cunningham v. Blackwell, 41 F.4th 530, 536 (6th Cir. 2022). But at its “core,” due process “requires notice and
1Indeed, the Supreme Court clarified, just two years after Erie, that a federal diversity court’s central aim is to “ascertain from all the available data what the state law is,” not to “reject the state rule merely because it has not received the sanction of the highest state court.” West v. Am. Tel. & Tel. Co., 311 U.S. 223, 236–37 (1940). In other words, a “state is not without law save as its highest court has declared it.” Id. at 236. And when the available data comprised lower-court decisions declaring rules of law applicable statewide, the Supreme Court treated them as conclusive on federal courts sitting in diversity. Fidelity Union Tr. Co. v. Field, 311 U.S. 169, 179 (1940) (New Jersey’s judicial system in 1940 included a Court of Chancery whose decisions were “treated as binding in later cases in chancery” and directly reviewable only by the state’s highest court).
2Following Kemerer would not mean that O’Connor would lose on appeal. Yes, Kemerer affects the titletransfer question. But there’s still the question of the process due before Michigan can deprive O’Connor of his property. So I would still remand because I don’t think the district court picked the right due-process framework— set out in Texaco, Inc. v. Short, 454 U.S. 516 (1982)—because the rationale underlying that framework relies on an effectively passive state. Michigan has significant involvement in its abandoned-property regime, which implicates due process in a way that the scheme in Texaco did not.
No. 25-2104 O’Connor v. Eubanks, et al. Page 16
an opportunity to be heard ‘at a meaningful time and in a meaningful manner.’” Garcia v. Fed. Nat’l Mortg. Ass’n, 782 F.3d 736, 741 (6th Cir. 2015) (quoting Armstrong v. Manzo, 380 U.S. 545, 552 (1965)). And determining whether the government met that core obligation in a given context “turns on three factors: the government’s interest, the individual’s stake in the matter, and the suitability of the procedures.” Cunningham, 41 F.4th at 536.
Instead of balancing the parties’ interests to determine whether Michigan afforded sufficient process, the district court assumed (in the decision we’ve now vacated) that Texaco’s narrower framework governed. That case centered on Indiana’s statute of limitations for mineral rights. Texaco, 454 U.S. at 536–37. The Indiana statute “put[] an end to interests in coal, oil, gas, or other minerals which have not been used for twenty years” by declaring that unused subsurface interests reverted to the owner of the surface property. Id. at 518 (citation modified). Faced with a due-process challenge to that statute, the Supreme Court held that “the due process standards . . . appl[icable] to an ‘adjudication’” do not apply to a law that is, in effect, a “self- executing statute of limitations.” Id. at 535–36. All that due process demands in that context is publication of the statute, paired with a reasonable amount of time for the public to familiarize itself with the statute before it takes effect. Id. But if the state instead adopts a system by which a “property interest [is] taken only after a specific determination that the deprivation was proper,” due process requires individualized notice and an opportunity to be heard. Id. at 537. Indeed, the Supreme Court emphasized in a later case that “it is the self-executing feature of a statute of limitations that makes [the individualized-notice requirement] inapposite.” Tulsa Pro. Collection Servs., Inc. v. Pope, 485 U.S. 478, 486 (1988) (citation modified).
Michigan’s unclaimed property regime is distinct on two dimensions from the Indiana statute at issue in Texaco. First, the Michigan statute isn’t self-executing. It requires “holders” of unclaimed property—entities who have “possession of a property belonging to another”—to continuously monitor their inventories for property that they believe are abandoned and turn it over to the state. Mich. Comp. Laws §§ 567.222(i), 567.251. Second, Michigan’s unclaimed property regime involves “significant state action,” whereas Texaco blessed a statute that required such “limited [state] involvement” that it didn’t “implicate the protections of the Due Process Clause.” Pope, 485 U.S. at 486–87. Indeed, it’s difficult to imagine more state
No. 25-2104 O’Connor v. Eubanks, et al. Page 17
involvement than Michigan’s. The state continuously monitors holders of property to make sure they’re complying with their reporting and turn-over obligations. Mich. Comp. Laws § 567.243. And once it receives property, it can then liquidate and deposit the proceeds in the state’s general fund. Id. § 567.251.
So because Texaco’s rationale hinges on Texaco’s facts—facts that are markedly different from the ones with which we’re confronted by Michigan’s unclaimed-property regime—I don’t think it should weigh substantially in the district court’s calculus determining the proper due-process framework for O’Connor’s claim to his interest on remand.
* * *
Bound as we are by our own precedent, I agree with the majority’s decision to disregard Kemerer and hold that O’Connor retained ownership over his principal even while it was in state custody. And guided by the prudence of judicial process, I also agree with the majority’s decision to leave the resolution of the correct due-process standard up to the district court in the first instance. So I concur fully in the majority opinion, and I write separately only to offer some thoughts about why our precedent shouldn’t have cast Kemerer aside, and why these circumstances probably demand more than the statutory notice approved by the Supreme Court in Texaco.