United States v. Tyren Cervenak

99 F.4th 852

• Court: Court of Appeals for the Sixth Circuit
• Decided: April 26, 2024
• Docket: 23-3466
• Status: Published

Opinion

RECOMMENDED FOR PUBLICATION Pursuant to Sixth Circuit I.O.P. 32.1(b) File Name: 24a0094p.06

UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT

UNITED STATES OF AMERICA, ┐ Plaintiff-Appellee, │ │ > No. 23-3466 v. │ │ │ TYREN L. CERVENAK, │ Defendant-Appellant. │ ┘

Appeal from the United States District Court for the Northern District of Ohio at Cleveland. No. 1:22-cr-00204-1—John R. Adams, District Judge.

Decided and Filed: April 26, 2024

Before: GILMAN, McKEAGUE, and THAPAR, Circuit Judges.

_________________

COUNSEL

ON BRIEF: Catherine Adinaro Shusky, OFFICE OF THE FEDERAL PUBLIC DEFENDER, Cleveland, Ohio, for Appellant. Jason Manion, UNITED STATES ATTORNEY’S OFFICE, Cleveland, Ohio, for Appellee.

McKEAGUE, J., delivered the opinion of the court in which THAPAR, J., joined. THAPAR, J. (pp. 15–22), delivered a separate concurring opinion. GILMAN, J. (pp. 23–43), delivered a separate dissenting opinion. _________________

OPINION _________________

McKEAGUE, Circuit Judge. Tyren Cervenak challenges his 188-month sentence. Specifically, he challenges the district court’s application of a career-offender enhancement that increased the advisory sentencing guidelines range for his convictions. Cervenak argues that the No. 23-3466 United States v. Cervenak Page 2

district court erred by determining one of his prior state convictions—an Ohio robbery— qualified as a crime of violence under the sentencing guidelines.

Binding precedent forecloses Cervenak’s challenge to the career-offender enhancement. Under this precedent, his Ohio robbery conviction is a crime of violence. We AFFIRM.

I.

On August 11, 2022, Tyren Cervenak pleaded guilty in federal court to two counts of distributing crack cocaine and one count of possessing a firearm and ammunition as a convicted felon. Previously, Cervenak had been convicted of numerous crimes in Ohio state court. Relevant here are two felony convictions: one in 2016 for two counts of robbery while using a firearm and another in 2020 for trafficking heroin. Prior to sentencing in the federal case, pretrial services recommended that the district court consider Cervenak a career offender under section 4B1.1 of the sentencing guidelines. In the recommendation, pretrial services explained that two of Cervenak’s state convictions qualified as “either a crime of violence or a controlled substance offense.” Because Cervenak had at least two such convictions, he qualified as a career offender. Cervenak objected to his career-offender designation, arguing that a conviction under Ohio’s robbery statute does not qualify as a crime of violence under the sentencing guidelines. See Ohio Rev. Code § 2911.02(A)(2); U.S.S.G. §§ 4B1.1(a), 4B1.2(a). He did not dispute that his heroin conviction could serve as a predicate offense for the enhancement.

At the sentencing hearing, the court overruled Cervenak’s objection, stating it would set out the reasons for its decision in a forthcoming order. The court then accepted the guidelines calculations recommended by pretrial services: with an offense level of 29 and a criminal history category of VI, the guidelines yielded an advisory sentencing range of 151 to 188 months of imprisonment. Cervenak’s counsel reiterated Cervenak’s objection to the career-offender enhancement and argued for a downward variance from the guidelines range. The government responded by arguing that Cervenak was the “definition of a career offender” and that his conduct and criminal history justified imposing a sentence within the guidelines range. Sentencing Hr’g Tr., R.53 at PageID 263–65. The court agreed with the government and imposed a total sentence at the high end of the guidelines range: 188 months. No. 23-3466 United States v. Cervenak Page 3

After the hearing, the court issued an order explaining its reasoning for applying the career-offender enhancement to Cervenak’s guidelines calculations. In it, the district court agreed with Cervenak that, under this Court’s opinion in United States v. Butts, 40 F.4th 766 (6th Cir. 2022), his Ohio robbery conviction did not qualify as a crime of violence under the “force/elements” clause of the career-offender guidelines. See U.S.S.G. § 4B1.2(a)(1). The district court then assessed Cervenak’s robbery conviction under the enumerated-offenses clause of the guideline, which Butts declined to do. See U.S.S.G. § 4B1.2(a)(2); Butts, 40 F.4th at 768 n.1. The district court concluded that a conviction under Ohio Revised Code § 2911.02(A)(2) (“Ohio robbery”) falls within the definition of “robbery” included among the enumerated offenses in U.S.S.G. § 4B1.2(a)(2), meaning that Ohio robbery qualifies as a crime of violence under the guidelines. Cervenak timely appealed.

II.

The parties agree that we review de novo “a district court’s determination that a crime constitutes a crime of violence” under the guidelines. Butts, 40 F.4th at 770.

A.

Our recent decision in United States v. Carter governs the outcome of this case. 69 F.4th 361 (6th Cir. 2023), cert. denied, 144 S. Ct. 612 (2024). In that opinion, we held squarely that a conviction for Ohio robbery—under exactly the same statutory provision—qualifies as a crime of violence under the guidelines. Id. at 363. Carter—which we decided after the district court here sentenced Cervenak—declined to compare Ohio robbery to generic “robbery” under the enumerated-offenses clause. See id. Instead, we found that Ohio robbery categorically matched generic “extortion,” which is also an enumerated offense under U.S.S.G. § 4B1.2(a)(2).1 See id. “Under the Guidelines,” we explained, “extortion is ‘obtaining something of value from another’ by wrongfully using ‘force, fear of physical injury, or threat of physical injury.’ ” Id. (quoting U.S.S.G. § 4B1.2 cmt. n.1 (2021)). We determined that section (A)(2) of the Ohio robbery

1 We note that, at the time Cervenak was sentenced, “extortion” was defined only in the application notes to section 4B1.2. U.S.S.G. § 4B1.2 cmt. n.1 (2021). The guidelines have since been amended to include the same definition in the text of section 4B1.2. U.S.S.G. § 4B1.2(e)(2). Both Carter and United States v. Camp, 903 F.3d 594, 602–03 (6th Cir. 2018), confirm that we previously adopted the guideline commentary definition as our definition for generic extortion. No. 23-3466 United States v. Cervenak Page 4

statute2 included four elements: “(1) theft or attempted theft and (2) physically harming, attempting to harm, or threatening to harm (3) another person (4) during the theft or attempted theft.” Id. We concluded that each element of the Ohio robbery statute was either exactly the same or narrower than the corresponding definition under the generic form of extortion. Id. at 363–65.

Although we had determined in the past that Ohio’s robbery statute is twice divisible, Carter didn’t reach the second level of divisibility. See id. at 363–64. The categorical approach to determining whether a state crime is a crime of violence helps courts “identify the elements of the crime of conviction” where a state’s statute might otherwise be “opaque.” Mathis v. United States, 579 U.S. 500, 513 (2016). Under the approach, we break crimes into their constituent elements, which then enables us to compare the state offenses to various definitions—e.g., “use of force” or “extortion”—under federal law. See Taylor v. United States, 495 U.S. 575, 601 (1990). If a statute is overbroad—for example, if it criminalizes conduct that can, but need not always, involve the “use of force”—we apply a “modified” categorical approach. See United States v. Burris, 912 F.3d 386, 392–93 (6th Cir. 2019) (en banc). Under that approach, we ascertain whether a state statute is divisible, meaning there may be multiple distinct ways—by satisfying alternative elements—to commit an offense. Id. When a statute is divisible, we may consult state charging documents to determine which “alternative” formed the basis of the defendant’s prior conviction. Id. Under both the traditional and modified categorical approach, we ask whether the elements of the state offense are “the same as, or narrower than” the elements of the comparable federal definition. Descamps v. United States, 570 U.S. 254, 257 (2013).

In United States v. Butts, we applied the modified categorical approach to robbery under Ohio law to ask whether the offense has, as an element, the use or attempted use of force against another person. 40 F.4th 766 (6th Cir. 2022) (holding Ohio robbery didn’t qualify for a career- offender enhancement under the “force/elements” clause in section 4B1.2(a)(1)). We found that statute to be twice divisible. Id. at 770–71. First, the statute provides three different ways to commit robbery: one involving a weapon, one involving physical harm, and one involving the

2 Which provides that no person, “in attempting or committing a theft offense or in fleeing immediately after the attempt or offense, shall . . . [i]nflict, attempt to inflict, or threaten to inflict physical harm on another.” Ohio Rev. Code § 2911.02(A), (A)(2). No. 23-3466 United States v. Cervenak Page 5

use of force. Id. at 770 (quoting Ohio Rev. Code § 2911.02(A)). But covered conduct under the physical-harm form of robbery—the one at issue both in Butts and here—could potentially include conduct that did not involve the use of force. Id. at 771 (citing Ohio Rev. Code § 2911.02(A)(2)). That’s because the statute further subdivides into thirty-one different crimes based on the underlying “theft offense,” not all of which satisfied the use-of-force element. Consequently, we divided robbery a second time to compare convictions under the applicable theft offense to the federal “use of physical force” definition. Id. 771–72.

In applying this approach in Butts, we followed our earlier decision in United States v. Wilson, which undertook a double-divisibility inquiry to determine whether Ohio’s aggravated- robbery statute involved the use of force. See 978 F.3d 990, 996 (6th Cir. 2020). In Wilson, just as in Butts, Ohio’s aggravated-robbery statute was still overbroad after dividing the statute once. Id. at 995–96. So, Wilson required courts to divide the statute a second time to determine whether particular defendants’ convictions involved the use of force. Id. at 999.

In Carter, by contrast, we stopped at the first level of divisibility. This is because we concluded the elements of Ohio robbery categorically matched the elements of extortion under the guidelines, regardless of the predicate theft offense. Carter, 69 F.4th at 363–64. Thus, we did not need to reach the second level of divisibility. Specifically, we held that “theft offense” criminalized less conduct than the corresponding “obtaining” element of guidelines extortion. Id. at 364. The same was true of the remaining elements. Because the “elements of Ohio robbery are ‘the same as, or narrower than,’ the Guidelines’ elements of extortion,” we concluded Ohio robbery qualified as a crime of violence under U.S.S.G. § 4B1.2(a)(2). Id. at 365 (quoting Camp, 903 F.3d at 600).

B.

Cervenak’s only challenge to Carter’s application here is that Carter is premised on a mistake. Specifically, he argues that the Carter panel erred by failing to conduct the double- divisibility analysis mandated by Butts and Wilson and by assuming that “theft offense” under the Ohio statute categorically matches the “obtaining” element of generic extortion. Appellant’s No. 23-3466 United States v. Cervenak Page 6

Br. 16–18; Reply Br. 4–5. There are several reasons why we cannot disregard Carter to find that Cervenak’s conviction does not qualify as a crime of violence under the guidelines.

First, regardless of Cervenak’s argument that we are not bound because of an error in Carter’s analysis, a subsequent panel may not disregard the holding of another prior published panel opinion. See United States v. Ferguson, 868 F.3d 514, 515 (6th Cir. 2017). Cervenak’s only recourse is to take up the issue with the en banc Sixth Circuit or the Supreme Court.3 Carter determined that the Ohio robbery statute was a categorical match to guidelines extortion, meaning it qualifies as a crime of violence. See 69 F.4th at 363–65. We are bound by that determination. Because Cervenak is in exactly the same position as the defendant in Carter, Carter controls.

The dissent takes largely the same tack in challenging Carter’s application to this case— in essence, that the Carter panel was mistaken in its determination of the elements of Ohio robbery. But the dissent does not present any authority that indicates we may disregard the Carter panel’s holding. The dissent argues only that, because the parties failed to address squarely the predicate-theft question, we may not rely on the Carter panel’s legal conclusion that “theft offense” under Ohio law categorically matches the “obtaining” element of generic extortion. The nature of the federal judicial power underscores why we may not simply disregard this conclusion as a mere presumption. To the contrary, if litigants could “extract the opinion of a court on hypothetical” statutes or “dubious constitutional principles” by “agreeing on the legal issue presented,” our role in resolving cases or controversies “would be difficult to characterize as anything but advisory.” U.S. Nat’l Bank of Or. v. Indep. Ins. Agents of Am., 508 U.S. 439, 447 (1993). A court’s province is the law; we are not bound by litigants’ stipulations or concessions to legal principles. To be sure, when presented with a stipulation or concession, we may—in the first instance—assume a principle to be true and apply it to the case before us. Wright v. Spaulding, 939 F.3d 695, 704 (2019). But here, the Carter panel did no such thing. The panel never merely “acquiesce[d]” to either party’s concessions or assumptions. Id. Instead, it actively applied the “judicial mind” to make the legal conclusion that an

3 The defendant in Carter pursued both avenues for relief. Both petitions were denied. See United States v. Carter, No. 22-3699, 2023 WL 5746857 (6th Cir. Aug. 29, 2023); Carter v. United States, 144 S. Ct. 612 (2024) (mem.). No. 23-3466 United States v. Cervenak Page 7

unspecified theft offense under Ohio law was no broader than the “obtaining” element of extortion. See id. (quoting Carroll v. Carroll’s Lessee, 57 U.S. (16 How.) 275, 287 (1853)). And now, as we apply Carter’s holding to this case, we may no longer question that legal determination.

Second, Wilson and Butts are not applicable to the guidelines provision at issue here. Wilson assessed whether Ohio aggravated robbery had “as an element the use, attempted use, or threatened use of physical force” under the Armed Career Criminal Act. 978 F.3d at 993 (quoting 18 U.S.C. § 924(e)(2)(B)(i)). And Butts assessed whether Ohio robbery matched under the analogous “force/elements” clause in section 4B1.2(a)(1) of the guidelines. 40 F.4th at 769. Neither addressed whether Ohio robbery was a crime of violence under the enumerated-offenses clause of the guidelines. Carter later foreclosed that analysis entirely, holding that the elements of Ohio robbery categorically match the elements of extortion under the guidelines. 69 F.4th at 363.

C.

Even if we could reach the merits of whether Cervenak’s robbery conviction qualifies as a crime of violence, we would likely reach the same conclusion as Carter. There are at least two good reasons to think Carter’s holding was correct as it applies to robbery convictions that do not specify the underlying theft offense (“unspecified robbery convictions”).

1.

First, for Cervenak to prevail under his categorical-approach challenge, he must show a “realistic probability” that the state “would apply its statute to conduct that falls outside the generic definition of a crime.” Gonzales v. Duenas-Alvarez, 549 U.S. 183, 193 (2007). Indeed, Cervenak “must at least point to his own case or other cases in which the state courts in fact did apply the statute in the special (nongeneric) manner for which he argues.” Id. Here, Cervenak suggests the state could criminalize, under the predicate theft offenses, conduct that sweeps beyond the “obtaining property from another” element of extortion. See Appellant’s Br. 18; see also, e.g., Ohio Rev. Code § 2913.47 (prohibiting false statement in insurance application). But he provides no Ohio caselaw showing the state applies its robbery statute to conduct in No. 23-3466 United States v. Cervenak Page 8

which a defendant does not obtain property from another person. Indeed, some Ohio caselaw suggests the opposite—that obtaining property is a required element of a robbery conviction in Ohio. See State v. Brown, 198 N.E.3d 111, 120 (Ohio Ct. App. 2022) (“[W]hen no property is taken from a person, a threat of harm used against that person does not establish a robbery.”), rev’d on other grounds, 2024-Ohio-749 (Ohio 2024).

Instead, Carter argues only that the Supreme Court in Taylor rejected any requirement that the defendant show the state statute’s broad sweep. But Taylor did no such thing. It affirmed the Duenas-Alvarez requirement that, where the state and federal offenses “clearly overlapped,” federal courts should look to “state decisional law” to determine what conduct the state statute covers. United States v. Taylor, 596 U.S. 845, 858–59 (2022). Ohio robbery and generic extortion clearly overlap, meaning Cervenak must show that Ohio courts apply the robbery statute more broadly than extortion. He has not done so.

2.

Even if Cervenak could find a case in which Ohio applied its robbery statute to non- extortive conduct, it’s unlikely he would fare any better. Assuming some Ohio robberies do not qualify as extortion, we employ the modified categorical approach to determine if his conviction qualifies. Under that approach, we consult Cervenak’s Shepard documents to determine which version of robbery he committed. See Mathis, 579 U.S. at 500. Then, we identify the “elements” of that version and see if they match guidelines extortion. In identifying these elements, we look to how state courts apply the statute. See Johnson v. United States, 559 U.S. 133, 138 (2010) (“We are . . . bound by [a state’s] interpretation of state law, including its determination of the elements of [the applicable crime].”). Using that approach here, we believe that Cervenak’s conviction categorically matches extortion.

To start, the parties agree that Cervenak’s Shepard documents, like Carter’s, do not list the specific theft offense underlying his robbery conviction. At this point, it might be tempting to follow Ivy’s resolution: because Cervenak’s Shepard documents do not spell out his theft offense, perhaps we should presume his conviction rested upon the least-culpable offense permitted by the statute’s text. United States v. Ivy, 93 F.4th 937, 943 (6th Cir. 2024). And since No. 23-3466 United States v. Cervenak Page 9

certain theft offenses, such as burglary, don’t involve taking property, we would have to conclude that Cervenak’s unspecified robbery conviction does not categorically match extortion.

While tempting, this line of reasoning conflicts with a key tenet of the categorical approach: state courts have the final say on state law. See Cradler v. United States, 891 F.3d 659, 669 (6th Cir. 2018) (noting that federal courts cannot look “only to the words of the violated state statute” when applying the categorical approach); see also Hutchison v. Marshall, 744 F.2d 44, 46 (6th Cir. 1984) (“It is axiomatic that state courts are the final authority on state law.”). Accordingly, we must consult Ohio court decisions to discern the elements of an otherwise unspecified robbery conviction. See Johnson, 559 U.S. at 138.

Typically, we look to the “pronouncements of the state’s highest court to determine the full range of conduct that is encompassed by each statutory element.” Cradler, 891 F.3d at 669; see Johnson, 559 U.S. at 138. But, where we have no dispositive state supreme court caselaw available, we may “predict” how the Ohio Supreme Court “would resolve the issue from ‘all relevant data.’ ” Kingsley Assocs., Inc. v. Moll PlastiCrafters, Inc., 65 F.3d 498, 507 (6th Cir. 1995) (quoting Bailey v. V & O Press Co., 770 F.2d 601, 604 (6th Cir. 1985)). “Relevant data” include state appellate court decisions. Id. And we should follow those decisions unless given “persuasive data” that the Ohio Supreme Court would hold differently. Id.

Cervenak has not identified any cases in which Ohio courts have interpreted “theft offense” more broadly than generic extortion where the predicate theft offense is unspecified under an Ohio robbery indictment. And the Ohio court decisions that we have found belie the notion that an unspecified robbery conviction could rest upon any one of the thirty-one theft offenses listed in the statute. Instead, our reading of Ohio caselaw suggests that when a defendant’s indictment does not list a specific theft offense, the “regular practice” of Ohio courts is to require that the state prove the defendant committed “theft” under section 2913.02(A).4 See Burris, 912 F.3d at 402; cf. State v. Smith, 884 N.E.2d 595, 599 (Ohio 2008) (“[T]here are

4 That section defines “theft” as knowingly obtaining or exerting control over property or services, with the purpose of depriving the owner thereof, (1) without the consent of the owner or person authorized to give consent; (2) beyond the scope of the express or implied consent of the owner or person authorized to give consent; (3) by deception; (4) by threat; or (5) by intimidation. Ohio Rev. Code § 2913.02(A). No. 23-3466 United States v. Cervenak Page 10

two possible ways to commit the [use-of-force form of robbery]: robbery by theft or robbery by attempted theft.”).

Ohio jury instructions and caselaw support this conclusion. When Ohio prosecutes unspecified robbery, trial courts seem generally to charge the jury using the statutory definition of 2913.02(A) theft. See, e.g., Indictment at 2, State v. Tillison, No. 2016 CRC-I 000371 (Wayne Cnty. Ct. Com. Pl. Nov. 21, 2016) (charging defendant with unspecified robbery); Jury Instr. at 4–5, 7, id. (Feb. 14, 2017) (defining “theft offense” as 2913.02(A) theft); State v. Pigg, 2005- Ohio-2227, at ¶¶18–19, 24 (Ohio Ct. App. 2005) (describing how the trial court instructed the jury on 2913.02(A) theft where the elements of the underlying theft offense were otherwise unspecified and holding the trial court did not err); Jury Instr. at 14, State v. Stovall, No. G-4801- CR-0201703132-000 (Lucas Cnty. Ct. Com. Pls. Mar. 7, 2018) (instructing the jury that 2913.02(A) “ ‘[t]heft’ is an essential element of robbery” even in a case where the defendant was also charged with a different theft predicate—there, aggravated burglary); Indictment at 2–3, id. (Dec. 11, 2017).

Similarly, when defendants raise a sufficiency-of-the-evidence challenge against an unspecified robbery conviction, intermediate Ohio appellate courts typically assume that the state must show there was sufficient evidence of 2913.02(A) theft—not just any one of the thirty-one theft offenses. See, e.g., State v. Johnson, 2012-Ohio-2325, at ¶8 (Ohio Ct. App. 2012) (noting section 2913.02(A) “defines the underlying theft offense” for an unspecified robbery conviction); Bill of Particulars at 1–2, State v. Johnson, No. 09-CR-2215 (Franklin Cnty. Ct. Com. Pl. Mar. 2, 2011) (outlining the bill of particulars for the unspecified robbery conviction challenged in Johnson, 2012-Ohio-2325); State v. Pullen, 2015-Ohio-552, at ¶14 (Ohio Ct. App. 2015) (finding sufficient evidence to support that a jury could have found a 2913.02(A) theft, which in turn satisfies the unspecified “theft offense” element of Ohio robbery); Indictment at 1, State v. Pullen, No. 2013 CR 00933 (Montgomery Cnty. Ct. Com. Pl. Apr. 30, 2013) (charging same defendant with unspecified robbery); State v. McCloud, 2018-Ohio-3291, at ¶¶13–15 (Ohio Ct. App. 2018) (analyzing sufficiency and assuming 2913.02(A) theft); Indictment at 1, State v. McCloud, No. 2017 CR 01956/2 (Montgomery Cnty. Ct. Com. Pl., July 10, 2017); State v. Green, 2004-Ohio-3697, at ¶¶10–13 (Ohio Ct. App. 2004) (analyzing sufficiency and assuming No. 23-3466 United States v. Cervenak Page 11

2913.02(A) theft); Indictment at 1–2, State v. Green, No. 03 CR 003003 (Franklin Cnty. Ct. Com. Pl., May 2, 2003). Accordingly, in cases we’ve found, it seems that when Ohio prosecutes an unspecified robbery charge, the state must prove to a jury the defendant committed 2913.02(A) theft beyond a reasonable doubt.

Under the categorical approach, then, Ohio courts appear to treat 2913.02(A) theft as an “element” of unspecified robbery convictions. See Mathis, 579 U.S. at 504 (“ ‘Elements’ are the constituent parts of a crime’s legal definition, which must be proved beyond a reasonable doubt to sustain a conviction.”). And as Carter correctly held, 2913.02(A) theft is narrower than the “obtaining” element of extortion.5 69 F.4th at 364. Thus, Ohio caselaw supports the conclusion that an unspecified robbery conviction categorically matches guidelines extortion.

The dissent provides only one example of an Ohio case in which an appellate court upheld a conviction of robbery with an unspecified predicate theft offense by assuming the predicate was something other than 2913.02(A) theft. See Dissenting Op. at 34; State v. Dixon, No. 89-CA-22, 1990 WL 94612 (Ohio Ct. App. June 20, 1990). This case does not compel a different conclusion here. For one, this case occurred before Ohio amended its robbery statute to criminalize conduct under the infliction-of-harm provision at issue here. See Dixon, 1990 WL 94612, at *2; Act of Aug. 10, 1995, § 2911.02, 1995 Ohio Laws 50. For another, the case does not undermine our conclusion that “theft offense” is no broader than the “obtaining” element of generic extortion, because, firstly, the defendant in Dixon attempted to cash a forged check. Dixon, 1990 WL 94612, at *2. And secondly, as explained below, the Ohio Supreme Court has since clarified that threats under the Ohio robbery statute, including by possessing a deadly weapon, “intimidate[] a victim and [are] designed to compel relinquishment of property without consent.” State v. Evans, 911 N.E.2d 889, 895 (Ohio 2009).

To be sure, we have only intermediate appellate caselaw as support for our conclusion. But cf. Smith, 884 N.E.2d at 599 (Ohio Supreme Court case holding that 2913.02 theft is a lesser included offense of Ohio robbery). But Cervenak has not provided any evidence to the contrary, and our research has not revealed any instances in which Ohio courts apply robbery to non-

5 The parties do not dispute this aspect of Carter’s holding, nor do they dispute whether Ohio robbery satisfies the remaining elements of guidelines extortion. No. 23-3466 United States v. Cervenak Page 12

extortive conduct. Since robbery-by-theft is a categorical match for extortion, both Carter’s and Cervenak’s robbery convictions thus qualify as crimes of violence.

D.

Next, Cervenak points to United States v. Ivy, 93 F.4th 937 (6th Cir. 2024), as further evidence that Carter should not govern here. But his argument runs into a roadblock: Ivy applies to convictions under a different Ohio statute—indeed, the same aggravated-robbery statute (though a different subsection) at issue in Wilson. Ivy, 93 F.4th at 943. Of course, Ivy did distinguish Carter after the government argued that Carter’s extortion holding should extend to aggravated robbery. See id. at 947. And Ivy did apply Wilson’s double-divisibility analysis to its finding, under the enumerated-offenses clause of the crime-of-violence guideline, that Ohio aggravated robbery criminalizes a broader range of conduct than both generic robbery and extortion. Id. at 943.

But Carter, which predates Ivy, still controls as to the enumerated-offense analysis for Ohio non-aggravated robbery. Moreover, Ivy identified only a “theoretical possibility” that Ohio would apply its statute to non-extortive conduct. Duenas-Alvarez, 549 U.S. at 194. As we noted earlier, that’s not enough: there must be evidence that Ohio actually applies the statute to such conduct. Ivy didn’t identify any.

And although Ivy implies Carter is limited to convictions that explicitly list 2913.02(A) theft in the charging documents, see id. at 947, nothing in Carter indicates that to be true. Indeed, the record in Carter makes clear Carter was convicted of Ohio robbery without any specific predicate offense listed in his charging documents. See Doc. 36-2 at 4–6, Carter, 69 F.4th 361 (No. 22-3699). And for the reasons given above, under Ohio law, 2913.02(A) theft was an element of Carter’s conviction. Therefore, Carter correctly concluded that an unspecified robbery conviction included “theft” as an element, which matches the “obtaining” element of generic extortion. See Carter, 69 F.4th at 363–64.

This holding applies to Cervenak, whose Ohio charging documents similarly contain no indication of the predicate theft offense. So, under Carter, Cervenak’s conviction for Ohio robbery qualifies as a crime of violence. No. 23-3466 United States v. Cervenak Page 13

E.

Finally, in a pair of Rule 28(j) supplemental letters, the parties discuss whether our court’s recent decision in United States v. Brown has any bearing on this case. See No. 22-3470, 2024 WL 778119 (6th Cir. Feb. 26, 2024). Both parties correctly conclude it does not. For one, we are not bound by our circuit’s unpublished precedents. Bell v. Johnson, 308 F.3d 594, 611 (6th Cir. 2002). Moreover, even if Brown were binding, it would not change our analysis. To be sure, Brown suggested that Ohio prosecutes aggravated robbery in cases that do not involve taking something of value “from the person or presence of another.” 2024 WL 778119, at *2 (quoting 18 U.S.C. § 2111). The Brown panel also noted that Ohio defendants can ask courts for bills of particulars in unspecified aggravated-robbery prosecutions. Id. at *3. But neither observation disturbs our reasoning.

First, Brown pointed to two Ohio decisions—Anderson and Smith—suggesting Ohio prosecutes aggravated robbery for theft offenses that do not involve “taking property” in the “presence of another.” Id. at *2; see State v. Anderson, No. 63474, 1993 WL 410377 (Ohio Ct. App. Oct. 14, 1993); State v. Smith, 2006-Ohio-4669 (Ohio Ct. App. 2006), abrogated by Evans, 911 N.E.2d at 894. But those two cases do not help Cervenak. Each involved defendants who had taken property in the presence of another, satisfying the “obtaining” element of generic extortion. The Anderson defendant either stole a car from its driver or received a stolen car from another thief, and the Smith defendant robbed cash from a gas station attendant. Anderson, 1993 WL 410377, at *1–4; Smith, 2006-Ohio-4669, at ¶¶3–6.

To be sure, Smith suggested that a defendant could commit aggravated robbery outside the presence of another person. Smith, 2006-Ohio-4669, at ¶34. But that suggestion does not undermine our analysis here. For one, the Ohio Supreme Court has abrogated Smith’s holding and implied that aggravated robbery does indeed require a victim to be present. See Evans, 911 N.E.2d at 895. Moreover, even if Smith were still good law, Carter or Cervenak would be no better off. By Smith’s own terms, non-aggravated robbery under 2911.02(A)(2) “require[s] another person to be present.” Smith, 2006-Ohio-4669, at ¶34. Thus, Anderson and Smith are inapposite here. No. 23-3466 United States v. Cervenak Page 14

Second, as discussed above, we read Ohio law to establish that 2913.02(A) theft is treated as an element of unspecified robbery. The cases in Brown—State v. Landgraf, State v. Smith, and Scott v. Tibbels—do not indicate otherwise. In Landgraf and Smith, Ohio courts simply held that the state’s failure to specify a theft offense in an aggravated-robbery indictment does not render the indictment insufficient under Ohio law. State v. Landgraf, 2006-Ohio-838, at ¶¶7–16 (Ohio Ct. App. 2006); State v. Smith, 2003-Ohio-903, at ¶¶14–16 (Ohio Ct. App. 2003). That’s because, according to those courts, the defendants could have asked for a bill of particulars. Landgraf, 2006-Ohio-838, at ¶11. But an indictment’s sufficiency under state law does not tell us what “elements” the state must prove to convict someone of robbery. Nor does the availability of a bill of particulars.

By contrast, the jury instructions and sufficiency-of-the-evidence precedents discussed earlier do tell us about a crime’s elements. Namely, they indicate that 2913.02(A) theft is considered an element of unspecified robbery. Indeed, Scott v. Tibbels reinforces that point: there, the state court required Ohio to prove that a defendant committed 2913.02(A) theft when the indictment did not specify a predicate theft offense. No. 3:12-cv-146, 2013 WL 3579925, at *6–10 (S.D. Ohio July 11, 2013) (quoting State v. Scott, 2010-Ohio-1919, at ¶¶45–59 (Ohio Ct. App. 2010)).

In sum, nothing in Brown counsels against applying Carter here.

* * *

The district court did not err by applying the career-offender enhancement to Cervenak’s guidelines sentencing range.

III.

Because binding precedent establishes that Ohio robbery qualifies as a crime of violence under the sentencing guidelines, we AFFIRM the judgment of the district court. No. 23-3466 United States v. Cervenak Page 15

____________________

CONCURRENCE ____________________

THAPAR, Circuit Judge, concurring. On October 20, 2015, Tyren Cervenak pointed a gun at two people and ordered them to hand over everything they had. Ohio prosecuted Cervenak for this conduct, and he pled guilty to two counts of robbery. Now, our court must answer an ostensibly simple question: did Cervenak’s robbery offense involve “obtaining something of value from another by . . . threat of physical injury”? See U.S.S.G. § 4B1.2(a)(2), (e)(2).

“[V]irtually all thoughtful, civilized persons not overly steeped in the mysteries of [the categorical approach]” would say “yes.” Florida v. Royer, 460 U.S. 491, 520 (1983) (Rehnquist, J., dissenting). Unfortunately, the categorical approach requires us to spend four dozen pages debating a question that should be answered with one word. That’s not a critique of the majority opinion, which I join in full. Rather, it’s a testament to the categorical approach’s well- documented absurdity. See, e.g., Mathis v. United States, 579 U.S. 500, 536–44 (2016) (Alito, J., dissenting); United States v. Taylor, 596 U.S. 845, 860–73 (2022) (Thomas, J., dissenting); United States v. Doctor, 842 F.3d 306, 312–19 (4th Cir. 2016) (Wilkinson, J., concurring); Cradler v. United States, 891 F.3d 659, 672 (6th Cir. 2018) (Kethledge, J., concurring); United States v. Burris, 912 F.3d 386, 407–10 (6th Cir. 2019) (en banc) (Thapar, J., concurring); United States v. Lopez, 75 F.4th 1337, 1346 (11th Cir. 2023) (Grant, J., concurring). As the author of United States v. Carter, I write separately to offer additional responses to the dissent.

Start with the dissent’s assertion that the majority opinion misapplies the modified categorical approach. The majority opinion concludes that courts must perform a “divisibility analysis” only when a statute criminalizes more conduct than the relevant federal definition—in other words, when a statute is “overbroad.”1 The dissent disagrees, arguing we must also divide

1 It’s worth highlighting that the majority opinion doesn’t actually rely on this rule in holding that robbery is a crime of violence. The “divide-only-if-overbroad” portion of the majority opinion simply explains why Carter No. 23-3466 United States v. Cervenak Page 16

non-overbroad statutes. But neither the Supreme Court nor our precedent mandates the dissent’s approach.

Start with the Supreme Court’s decision in Mathis. 579 U.S. 500. There, the Court explained that in certain situations, federal sentencing “requires” a divisibility analysis of the predicate conviction. Id. at 505. Which situations are those? The Court answered with an example: a court would need to “divi[de]” a burglary statute that created two different offenses, one that qualifies as “generic burglary,” and one that doesn’t. Id. In other words, the modified categorical approach is required for an overbroad statute.

A few years later, our circuit reached the same conclusion. In United States v. Burris, our en banc court laid out the categorical and modified categorical approaches. 912 F.3d at 392–93. We prefaced our discussion of the modified categorical approach with a big “if.” Specifically, courts apply the modified categorical approach “if at least one of the statute’s alternative sets of elements . . . categorically matches the Guidelines . . . and at least one of the sets of elements does not.” Id. at 393 (cleaned up) (emphasis added). Again, we divide a criminal statute when it’s broader than a Guidelines definition.

Granted, neither case explicitly stated the converse: “don’t divide a statute when it’s not overbroad.” But that logically follows. After all, the whole point of the modified categorical approach is to discern which version of an overbroad statute a defendant violated. See Mathis, 579 U.S. at 505. When there’s no overbreadth problem, there’s no need to dive deeper. For instance, imagine a murder statute that “divides” into exactly two crimes: murder on a weekday, and murder on a weekend. And assume both versions meet the Guidelines’ crime-of-violence definition. If a court had to sentence a murderer for a subsequent crime, would it really need to walk through a divisibility analysis to figure out which version of murder he committed? Of course not. It wouldn’t make any sense to apply the modified categorical approach to a statute that’s already a categorical match. Tellingly, the dissent doesn’t identify a single case that does so.

didn’t engage in the Wilson-Butts analysis. By contrast, the majority opinion grounds its conclusion primarily on stare decisis, and secondarily on Duenas-Alvarez and Ohio law. See Majority op. 3, 7–12. No. 23-3466 United States v. Cervenak Page 17

Of course, that’s only half the battle. We still must address the question of the day: does Ohio robbery categorically match Guidelines extortion? Carter concluded it does. United States v. Carter, 69 F.4th 361, 361–62 (6th Cir. 2023). Since then, litigants like Cervenak have raised counterpoints—not raised in Carter—to that conclusion. The majority opinion effectively responds to those points. Below, I add a few more responses.

First, the majority opinion correctly looks to Ohio law. Because “state courts are the ultimate expositors of state law,” we don’t usually get to decide what Ohio’s statutes mean, no matter how “plain” we think their text is. Mullaney v. Wilbur, 421 U.S. 684, 691 (1975); see United States v. Mitchell, 743 F.3d 1054, 1059 (6th Cir. 2014) (“[I]n determining the meaning or scope of robbery, we are bound by the [state supreme court’s] interpretation, including its guidance on the elements of the crime.”). This principle applies with full force to the categorical approach. Cradler, 891 F.3d at 669–70; Johnson v. United States, 559 U.S. 133, 138 (2010).

That’s why Duenas-Alvarez requires defendants to identify a state-court decision applying the relevant statute to conduct outside the federal definition. See Gonzales v. Duenas- Alvarez, 549 U.S. 183, 193 (2007); see also Moncrieffe v. Holder, 569 U.S. 184, 206 (2013) (noting defendant “would have to demonstrate that the State actually prosecutes the relevant offense in [overbroad] cases”). This makes sense: if a state has never applied a statute to overbroad conduct, what authority do federal judges have to extend state criminal law to a new context? Cf. Sours v. Gen. Motors Corp., 717 F.2d 1511, 1520–21 (6th Cir. 1983) (“We hesitate to announce a decision, unless absolutely required to do so, in an important area of state law where there is no sure guidance from the state courts.”); New York v. Ferber, 458 U.S. 747, 769 n.24 (1982) (noting, in the First Amendment context, that a state court’s construction of a statute’s scope binds a federal court’s overbreadth inquiry).

The dissent’s efforts to distinguish away Duenas-Alvarez are unavailing. The dissent cites Taylor and Camp for the premise that defendants need not cite examples of overbroad applications when a statute is clear enough. But there’s an important caveat: both Taylor and Camp dealt with convictions under Hobbs Act—a federal statute. Taylor, 596 U.S. at 848; No. 23-3466 United States v. Cervenak Page 18

United States v. Camp, 903 F.3d 594, 596 (6th Cir. 2018). And federal courts can decide for themselves whether a federal statute applies to overbroad conduct. Taylor, 596 U.S. at 859. Thus, there’s no reason to make the defendant cite examples. Id.

But that same logic doesn’t apply to state statutes. Federal judges don’t get to decide what state law means—much less what conduct a state has decided to criminally punish. Taylor made this point abundantly clear and, if anything, endorses the application of Duenas-Alvarez here. Two paragraphs after the dissent’s excerpt from Taylor, the Supreme Court explained what distinguished that case from Duenas-Alvarez:

Duenas-Alvarez required a federal court to make a judgment about the meaning of a state statute. Appreciating the respect due state courts as the final arbiters of state law in our federal system, this Court reasoned that it made sense to consult how a state court would interpret its own State’s laws. Meanwhile, no such federalism concern is in play here. The statute before us asks only whether the elements of one federal law align with those prescribed in another.

Id. (citation omitted). Here, Cervenak’s case requires us to determine whether Ohio would apply its robbery statute to nonextortive conduct. So we’re on the Duenas-Alvarez side of the line. See United States v. Paulk, 46 F.4th 399, 403 n.1 (6th Cir. 2022) (holding that Duenas-Alvarez, not Taylor, applies when state law defines the predicate offense). That means we can’t conclude Ohio robbery is overbroad unless there’s an Ohio case applying that statute to nonextortive conduct. Cf. United States v. Butts, 40 F.4th 766, 772 (6th Cir. 2022) (applying Duneas-Alvarez to Ohio robbery post-Taylor). Neither Cervenak, nor the majority opinion, nor the dissent, nor I can find such a case. Under Duenas-Alvarez, that ends the debate. See Paulk, 46 F.4th at 404 (holding an offense was a crime of violence when the defendant and court couldn’t find a case applying the statute in an overbroad context); United States v. Garth, 965 F.3d 493, 498 (6th Cir. 2020) (same).

Moncrieffe is not to the contrary. The dissent quotes an excerpt from that case stating that federal courts applying the categorical approach “cannot discount [the Controlled Substances Act’s] text.” Moncrieffe, 569 U.S. at 197. But the Controlled Substances Act is a federal statute, so of course federal courts must consult the statutory text to determine its scope. By contrast, when Moncrieffe determined whether a Georgia statute was broader than the Act, No. 23-3466 United States v. Cervenak Page 19

the Court—expressly invoking Duenas-Alvarez—consulted state caselaw. Id. at 191, 194 (citing Taylor v. State, 581 S.E.2d 386, 388 (Ga. Ct. App. 2003); Hadden v. State, 353 S.E.2d 532, 533– 34 (Ga. Ct. App. 1987)). Thus, Moncrieffe supports the majority opinion’s decision to consult state courts, rather than embark on its own exegesis of another sovereign’s statute.

In sum, as “plain” as it might seem that Ohio’s robbery statute applies to nonextortive conduct, that’s not our call to make. We must instead follow the majority opinion’s approach and turn to Ohio law.

The Ohio Supreme Court has made clear that robbery always involves taking something of value from another. In 2008, that court held that “theft [under section 2913.02(A)] is a lesser included offense of robbery.”2 State v. Smith, 884 N.E.2d 595, 596 (Ohio 2008). And under Ohio law, a crime is a lesser-included offense only if “the greater offense cannot, as statutorily defined, ever be committed without the lesser offense, as statutorily defined, also being committed.” Id. at 597–99 (quotation omitted). That means Ohio robbery can’t be committed without also committing theft. In other words, theft is part of the “minimum conduct criminalized” under Ohio’s robbery statute. Dissenting op. 31. And since theft involves taking something of value from another, Ohio robbery categorically matches Guidelines extortion.3 Carter, 69 F.4th at 363–64.

Against Smith and the cases the majority opinion cites, the most the dissent can muster is a single robbery prosecution from thirty-five years ago. See State v. Dixon, No. 89-CA-22, 1990 WL 94612 (Ohio Ct. App. June 20, 1990). Like Cervenak’s indictment, the indictment in that case didn’t specify the predicate theft offense. Id. at *2. But when the defendant raised a sufficiency-of-the-evidence challenge, the Ohio court upheld the conviction after finding

2 Notably, the robbery conviction in Smith stemmed from an indictment that specified theft as the underlying theft offense. See Indictment at 2, State v. Smith, No. B 0503447 (Hamilton Cnty. Ct. Com. Pl. Apr. 15, 2005). But nothing in Smith limits its holding to such convictions. 3 The dissent highlights the majority opinion’s use of hedging language when discussing Ohio law. To be sure, elements of a crime are what the state must prove, not what the state typically has to prove. But that distinction gives me no pause here because, for the reasons I give herein, theft is always an element of Ohio robbery. No. 23-3466 United States v. Cervenak Page 20

sufficient evidence of forgery, not theft. Id. According to the dissent, this example sinks the contention that section 2913.02(A) theft is an element of all unspecified Ohio robbery convictions.

I disagree. First, the dissent’s lone example predates the Ohio Supreme Court’s decision that theft is a lesser-included offense of every robbery charge. See Smith, 884 N.E.2d at 596. By contrast, all but two of the cases the majority opinion cites post-date that decision. And all treat theft as an element of unspecified robbery.

Second, Dixon is unpublished. Ordinarily, that wouldn’t make a difference: Ohio treats published and unpublished decisions as equally controlling. Ohio S. Ct. Rules for the Reporting of Ops. 3.4. But before 2002, that wasn’t so: unpublished opinions weren’t authoritative pronouncements of law. Ohio S. Ct. Rules for the Reporting of Ops. 2(G)(1)–(2) (2001) (repealed 2002). Since Dixon predates that cutoff by over a decade, I hesitate to ascribe it weight here.

Finally, even if Dixon proves that theft isn’t an element of all unspecified robberies, this case would still come out the same way. As the majority opinion points out, the defendant in Dixon attempted to obtain something of value. See 1990 WL 94612, at *2. So that conviction is still a match for Guidelines extortion and doesn’t satisfy Cervenak’s burden under Duenas- Alvarez.

The dissent’s references to Stone and Dunn fare no better. To be sure, the defendants in those cases were convicted of “theft offenses,” despite not having obtained anything of value. See State v. Stone, 2024-Ohio-177 (Ohio Ct. App. 2024); State v. Dunn, 2023-Ohio-4413 (Ohio Ct. App. 2023). But neither defendant was convicted of robbery. Cervenak’s case requires us to determine whether someone can be convicted of Ohio robbery without obtaining anything of value. See Duenas-Alvarez, 549 U.S. at 193. So the non-robbery convictions in Stone and Dunn are irrelevant. No. 23-3466 United States v. Cervenak Page 21

IV.

The dissent’s remaining critiques of the majority opinion are similarly unpersuasive. At one point, the dissent claims the majority opinion is “interpret[ing] prosecutorial trends” rather than assessing the elements of Ohio robbery. Dissenting op. 33. Not so. In concluding theft is an element of unspecified robbery, the majority opinion consults state-court jury instructions and sufficiency-of-the-evidence analyses. These are examples of how Ohio courts apply state law, not patterns in how Ohio prosecutors charge defendants. Specifically, these examples show that, to convict a defendant of unspecified robbery, a jury must find that he committed theft. Thus, these examples prove theft is an element of Ohio robbery. See Mathis, 579 U.S. at 504.

The dissent then attempts to downplay the probative value of the sufficiency-of-the- evidence examples. By the dissent’s reading, the Ohio courts in these cases weren’t saying that theft is an element of unspecified robbery. Instead, says the dissent, those courts simply sifted through trial records to determine which theft offense was most “appropriate” for a sufficiency challenge. Dissenting op. 35. And it just so happened that theft was the best fit for all the majority opinion’s cited cases.

That reading can’t possibly be correct. Due process requires that defendants know the elements of their alleged crime from the get-go. See Valentine v. Konteh, 395 F.3d 626, 631 (6th Cir. 2005). So it can’t be the case that an appellate court may retroactively define a conviction’s elements based on the evidence presented at trial. That’s simply not how our criminal justice system works.

Finally, our respect for Ohio’s sovereignty gives us one final reason to resist the dissent’s conclusion. If we abandon Ohio’s interpretation of its robbery law for the dissent’s, we cast a shroud of constitutional doubt over the state’s robbery convictions. Circuit precedent requires charging documents to list every element of a crime as a matter of due process. See id. So, if an unspecified theft offense in an indictment could refer to any of thirty-one possible offenses, then every unspecified robbery conviction might be subject to challenge. Contra Scott v. Tibbels, No. 12-CV-146, 2013 WL 3579925 (WHR) (MRM), at *6–10 (S.D. Ohio July 11, 2013), R&R adopted, 2014 WL 5384284 (S.D. Ohio Oct. 21, 2014) (denying habeas to a defendant who No. 23-3466 United States v. Cervenak Page 22

challenged his Ohio aggravated-robbery indictment, which failed to specify the predicate theft offense).

The dissent’s approach would “disturb [Ohio’s] significant interest in repose for concluded litigation . . . and intrude on [Ohio’s] sovereignty to a degree matched by few exercises of federal judicial authority.” Harrington v. Richter, 562 U.S. 86, 103 (2011) (cleaned up). The majority opinion appropriately declines to go along.

Today’s decision correctly applies the categorical approach by following Ohio’s interpretation of its robbery law. I’ll admit that the text of Ohio’s robbery statute makes it tempting for federal judges to use their “legal imagination.” Duenas-Alvarez, 549 U.S. at 193. “Fortunately, we need not resort to tea leaves or to judicial tarot cards—or to a bold substitution of our own views for those of the Ohio Supreme Court . . . .” Sours, 717 F.2d at 1521. Ohio courts have resolved the issue: robbery convictions involve a taking or attempted taking of property. Smith, 884 N.E.2d at 599. We must defer to that interpretation. No. 23-3466 United States v. Cervenak Page 23

DISSENT _________________

RONALD LEE GILMAN, Circuit Judge, dissenting. The majority holds that Tyren Cervenak’s challenge to the career-offender enhancement is without merit because his prior state-court conviction for robbery qualifies as a “crime of violence” under the enumerated- offenses clause of the U.S. Sentencing Guidelines. Specifically, the majority concludes that an offender such as Cervenak, who was convicted of robbery under Ohio Rev. Code Ann. § 2911.02(A)(2), was necessarily found to have obtained “something of value” from the victim by the use or threatened use of force. This leads the majority to conclude that a conviction for Ohio robbery is always a categorical match for the enumerated offense of extortion under the Guidelines, thus making Ohio robbery a crime of violence that potentially subjects an offender to an enhanced federal sentence.

I respectfully dissent because I disagree with the majority’s conclusion that Cervenak’s Ohio robbery conviction constitutes a crime of violence under the Guidelines. In reaching its conclusion, the majority erroneously relies on United States v. Carter, 69 F.4th 361 (6th Cir. 2023), a case where the relevant analysis for purposes of this appeal was premised on the incorrect assumption that all persons convicted of Ohio robbery had engaged in “theft” as statutorily defined. See Kitts v. United States, 812 F. App’x 336, 340 (6th Cir. 2020) (“Assumptions are not holdings that bind future panels.”). The majority then compounds its error by misapplying the categorical approach, in conflict with the Supreme Court’s decisions in Moncrieffe v. Holder, 569 U.S. 184 (2013), and Mathis v. United States, 579 U.S. 500 (2016), as well as with this court’s precedents in United States v. Wilson, 978 F.3d 990 (6th Cir. 2020), United States v. Butts, 40 F.4th 766 (6th Cir. 2022), and United States v. Ivy, 93 F.4th 937 (6th Cir. 2024).

I. THE OHIO ROBBERY STATUTE

To understand the problem here, one needs to grapple with the mind-bending complexity of Ohio’s robbery statutes. I emphasize “statutes” in the plural because there are actually two No. 23-3466 United States v. Cervenak Page 24

very lengthy statutes that define the offense of Ohio robbery. The first is § 2911.02 of the Ohio Revised Code Annotated, which breaks down “robbery” into three elements, each focusing on the severity of the threat (i.e., the use of a deadly weapon, the threat of bodily harm without the use of a deadly weapon, or simply the threat of force).

If any one of the above elements is met, then § 2911.02 references a second statute, Ohio Rev. Code Ann. § 2913.01(K), which lists 31 different “theft offenses.” An offender commits second-degree Ohio robbery by perpetrating any one of those 31 theft offenses while inflicting, attempting to inflict, or threatening to inflict physical harm on another. Ohio Rev. Code. Ann. § 2911.02(A)(2).

The “rub” here is that neither Cervenak’s state-court indictment nor his judgment of conviction specify which of the 31 “theft offenses” listed in § 2913.01(K) support his robbery conviction. This is a major problem because not all of the 31 listed theft offenses are a categorical match for the crime of “extortion” under the Guidelines.

The majority solves this problem by simply assuming that Cervenak’s unspecified theft offense was “theft” under Ohio Rev. Code Ann. § 2913.02, which is only one of the 31 listed theft offenses under §2913.01(K). Based on precedent from both the Supreme Court and this court, however, this is an unjustified assumption.

II. THE REQUIRED ANALYSIS FOR DIVISIBLE STATUTES

In Mathis v. United States, 579 U.S. 500 (2016), the Supreme Court discussed how the categorical approach changes for “more complicated (sometimes called ‘divisible’)” statutes. Id. at 505. Divisible statutes “list elements in the alternative, and thereby define multiple crimes,” which means that a defendant could be convicted under a single statute containing “two different offenses, one more serious than the other.” Id. Because this structure “mak[es] the comparison of elements harder,” courts “require[] a way of figuring out which of the alternative elements listed . . . was integral to the defendant’s conviction.” Id. To “address that need,” the Supreme Court “approved the modified categorical approach for use with statutes having multiple alternative elements.” Id. (quotation marks omitted). Under this approach, courts must look to the documents referred to in Shepard v. United States, 544 U.S. 13 (2005), (i.e., indictment, plea No. 23-3466 United States v. Cervenak Page 25

agreement, jury instructions, etc.) to determine what crime a defendant was convicted of. The court “can then compare that crime, as the categorical approach commands, with the relevant generic offense.” Id. at 506.

A. The established divisibility analysis as applied in Wilson and Butts

In United States v. Wilson, 978 F.3d 990 (6th Cir. 2020), this court applied the lessons of the Supreme Court in Mathis and held that Ohio’s aggravated-robbery statute was doubly divisible. Id. at 999. The court explained that, at the first divisibility step, the sentencing court must look to the limited class of Shepard documents to determine what crime, and with which elements, the defendant was convicted of. Id. Next, the court must determine which “theft offense” formed the basis of the conviction. And once the court has completed that inquiry, it has to do “what the categorical approach demands: compare the elements of the crime of conviction . . . with the . . . elements clause” of the Guidelines. Id. (quoting United States v. Burris, 912 F.3d 386, 393 (6th Cir. 2019)). Ohio’s robbery statute, moreover, has the exact same structure as the aggravated-robbery statute in Wilson, with both statutes incorporating the 31 theft offenses listed in Ohio Rev. Code Ann. § 2913.01(K) as alternative elements. These theft- offense elements also supply the mens rea for both statutes. See State v. Horner, 935 N.E.2d 26, 32–33 (Ohio 2010).

In United States v. Butts, 40 F.4th 766 (6th Cir. 2022), this court expounded upon Wilson by analyzing Ohio’s robbery statute (the same one at issue in the present case) for a conviction predicated on “theft” as codified in Ohio Rev. Code Ann. § 2913.02. This court held that such a conviction does not constitute a crime of violence under the “elements clause” of § 4B1.2(a) of the Guidelines. Id. at 768. The court further observed that an offense requiring only a reckless state of mind does not meet the Guidelines’ crime-of-violence standard. Id. at 770.

Relevant here, the Butts court found that the elements of Ohio Rev. Code Ann. § 2911.02(A) (1)–(3) “define[] multiple crimes by alternatively listing elements,” and concluded that “Ohio’s robbery statute is divisible.” Id. It also observed that, “when a statute is divisible, we look to the defendant’s particular underlying conviction to determine whether that offense is a predicate crime of violence.” Id. at 771. No. 23-3466 United States v. Cervenak Page 26

The Butts court further noted that “[w]e have held that the phrase ‘theft offense’ in Ohio’s aggravated robbery statute (one very similarly structured to §2911.02 (A)(2)) divides the statute into multiple theft offenses, and we see no reason to treat Ohio’s non-aggravated robbery statute differently.” Id. at 770–71 (internal citation omitted). It then stated that “[b]ecause § 2911.02’s theft element is divisible, we look to so-called Shepard documents, or state-court records reflecting the charge, to determine which Ohio theft offense predicated Butts’s robbery charge.” Id. at 771. From this analysis, the court held that, under the elements clause, Ohio robbery is not a crime of violence when § 2913.02 theft is the predicate offense because the force element of Ohio robbery under § 2911.02(A)(2) could be committed recklessly or negligently. Id. at 772. The court concluded that recklessness and negligence are not the states of mind that would warrant Guidelines’ sentencing enhancements. Id. at 774.

B. Ohio’s robbery statute is doubly divisible

This court has previously explained that, “like Ohio aggravated robbery, Ohio robbery is doubly divisible.” United States v. Ivy, 93 F.4th 937, 947 (6th Cir. 2024). Ohio robbery is first divisible because an offender can commit three different versions of robbery under subsections (A)(1), (2), or (3) of § 2911.02. Since each subsection is a different version of robbery, the statute “define[s] multiple crimes.” See Wilson, 978 F.3d at 996 (alteration in original) (quoting Mathis, 579 U.S. at 505). And the statute is divisible again because an offender must also commit any one of the 31 “theft offenses” cross-referenced in Ohio Rev. Code Ann. § 2913.01(K) in order to be convicted of Ohio robbery. See id. One of the 31 “theft offenses” listed in § 2913.01(K), for example, is § 2911.32, which deals with tampering with coin machines. Another “theft offense” referred to in § 2913.01(K) is safecracking under § 2911.31. Ohio robbery is therefore doubly divisible because a conviction requires an offender to (1) commit any one of three versions of robbery while (2) committing any one of the 31 theft offenses listed in §2913.01(K). See Butts, 40 F.4th at 770–72.

Cervenak’s robbery conviction arises from a violation of Ohio Rev. Code Ann. § 2911.02(A)(2). The statute creates a second-degree felony for an offender who “[i]nflict[s], attempt[s] to inflict, or threaten[s] to inflict physical harm on another” while “attempting or No. 23-3466 United States v. Cervenak Page 27

committing a theft offense.” But Cervenak’s state-court indictment and judgment of conviction do not specify the predicate theft offense for his robbery conviction.

III. CARTER DOES NOT CONTROL THIS CASE

The majority’s foundational error is concluding that United States v. Carter, 69 F.4th 361 (6th Cir. 2023), “governs the outcome of [the present] case.” Majority Op. at 3. The Supreme Court’s decision in Mathis outlined what is required when a court is faced with a divisible statute. This court’s decisions in Wilson and Butts applied Mathis’s analysis to Ohio’s aggravated-robbery and robbery statutes respectively. Because of the argument presented on appeal in Carter, however, that panel had no occasion to consider and follow the established divisibility analysis. I thus turn to an explanation of why I believe that Carter does not control the outcome of the present case.

In Carter, the only issue on appeal was whether the less-demanding mens rea requirement of Ohio robbery criminalized more conduct than general robbery under the Guidelines, such that Ohio robbery should not be considered a crime of violence. One of the government’s alternative arguments in response was that extortion under the Guidelines is a categorical match for Ohio robbery.

The government’s extortion argument in Carter was predicated on the assumption, made without any elaboration, that theft under Ohio Rev. Code Ann. § 2913.02 was the applicable predicate theft offense. In his reply brief, Carter did not address this argument. He argued instead that the panel should not compare Ohio robbery to extortion under the Guidelines because, as a matter of law, Ohio robbery should be compared only to Guidelines robbery. Carter never argued that his Ohio robbery conviction did not categorically match extortion.

The Carter court then analyzed Ohio’s robbery statute by assuming, like the government, that the predicate theft offense in that case was theft under Ohio Rev. Code Ann. § 2913.02. Under this “assumption-analysis,” the panel concluded that Ohio robbery qualifies as a crime of violence because Ohio robbery categorically matches the elements of extortion under the Guidelines. Carter, 69 F.4th at 363–64. But Carter’s categorical-match conclusion was possible only because of its assumption-analysis. No. 23-3466 United States v. Cervenak Page 28

Crucially, the Carter panel did not consider which predicate theft offense was present in that case because that question was not relevant in addressing the defendant’s argument. Although the defendant mentioned that “it [is] unclear what the underlying theft offense was” in his supplemental brief submitted to the district court, Carter’s appellate counsel did not discuss the relevance of this fact on appeal. Counsel mentioned that fact only when recounting the case’s procedural history. The panel instead assumed that “theft” under § 2913.02 was the defendant’s predicate theft offense without any analysis of the question. See Carter, 69 F.4th at 363–64 (“We count four elements: (1) theft or attempted theft and (2) physically harming, attempting to harm, or threatening to harm (3) another person (4) during the theft or attempted theft. These elements are the same as, or narrower than, the elements of Guidelines extortion.”) (emphasis added).

That assumption by the Carter panel is consequential because “[a]ssumptions are not holdings that bind future panels.” Kitts v. United States, 812 F. App’x 336, 340 (6th Cir. 2020) (holding that a panel’s prior determination that the defendant’s second-degree burglary qualified as a predicate offense under the ACCA was not binding because the panel “assumed that Tennessee defined its first element, breaking and entering, in conformance with general burglary”) (emphasis in original); see also Wright v. Spaulding, 939 F.3d 695, 704 (6th Cir. 2019) (rejecting the idea that “binding precedent” should be made “out of a court’s simple acquiescence in the parties’ . . . assumptions”).

The defendant in Carter—as relevant here—argued only that the panel should not, as a matter of law, compare Ohio robbery as defined in § 2911.02(A)(2) to extortion as defined in the Guidelines. Unlike Cervenak, Carter never argued that the reference in § 2911.02 to the 31 theft offenses listed in §2913.01(K) creates a categorical mismatch to Guidelines extortion, meaning “that the question before us in today’s case was simply not before us in [Carter].” See Nemir v. Mitsubishi Motors Corp., 381 F.3d 540, 559 (6th Cir. 2004). And so Carter “does not touch upon [Cervenak’s] argument.” See Kitts, 812 F. App’x at 340.

This court’s decision in Ivy confirms my point. The Ivy court explained that the Carter defendant “failed to argue the double divisible nature of the Ohio robbery statute as established in Wilson . . . . And because the Carter panel ‘was not presented with the argument that [Ohio No. 23-3466 United States v. Cervenak Page 29

robbery] is twice divisible, it had no occasion to consider’ the issue that we must now resolve.” United States v. Ivy, 93 F.4th 937, 947 (6th Cir. 2024) (cleaned up); see also Mathis v. United States, 579 U.S. 500, 505–06 (2016) (explaining that when a statute is divisible, courts apply the modified categorical approach).

This court in Ivy explicitly declined to extend Carter’s reasoning to a question that was nearly identical to the one before us because Ivy recognized that Carter’s entire analysis was based on a “presum[ption]” that the defendant did not refute. Ivy, 93 F.4th at 947. But here, Cervenak correctly argues that we may not engage in that presumption. That is because, “if the Shepard documents in a particular case do not make clear under which subsection of the relevant statute a defendant was convicted, sentencing courts must ‘presume that the conviction rested upon nothing more than the least of the acts criminalized.’” United States v. Burris, 912 F.3d 386, 406 (6th Cir. 2019) (en banc) (quoting Moncrieffe v. Holder, 569 U.S. 183, 190–91 (2013)); see also Ivy, 93 F.4th at 947 (“[W]e cannot rely on theft as the predicate offense here because theft is not listed as the predicate offense in his Shepard documents.”).

In sum, the parties’ arguments in Carter caused the panel to not consider how (1) the double divisibility of Ohio’s robbery statute affected the court’s analysis, or (2) if the reference in § 2911.02(A) to the 31 theft offenses listed in §2913.01(K) creates a categorical mismatch to Guidelines extortion. Because Carter’s analysis was based only on an assumption and did not address the issues before us in the present appeal, Carter does not establish a binding precedent on these points. See Nemir, 381 F.3d at 559 (“Questions which merely lurk in the record, neither brought to the attention of the court nor ruled upon, are not to be considered as having been so decided as to constitute precedents.”) (quoting Webster v. Fall, 266 U.S. 507, 511 (1925)).

IV. THE MAJORITY’S ANALYSIS CONFLICTS WITH DECISIONS FROM THE SUPREME COURT AND THIS COURT

Unlike the majority, I would not extend Carter’s “assumption-analysis” to the present case because precedents from both the Supreme Court and this court foreclose the assumption that the majority makes. I will now discuss the several ways in which I believe that the majority has erred. No. 23-3466 United States v. Cervenak Page 30

A. The divisibility analysis from Mathis, Wilson, Butts, and Ivy controls this case

First, the majority errs by disregarding the fact that Ohio’s robbery statute is doubly divisible. See Ivy, 93 F.4th at 947. (“[L]ike Ohio aggravated robbery, Ohio robbery is doubly divisible.”); see also Butts, 40 F.4th at 770–72 (explaining that the Ohio robbery statute is doubly divisible). The Supreme Court instructs that courts apply the modified categorical approach to divisible statutes when determining whether an offense is a crime of violence. Mathis v. United States, 579 U.S. 500, 517 (2016) (“The first task for a sentencing court faced with an alternatively phrased statute is thus to determine whether its listed items are elements or means. If they are elements, the court should do what we have previously approved: review the record materials to discover which of the enumerated alternatives played a part in the defendant’s prior conviction, and then compare the element (along with all others) to those of the generic crime.”).

This court has likewise required application of the modified categorical approach when analyzing a divisible statute to determine whether an offense is a crime of violence. United States v. Burris, 912 F.3d 386, 406 (6th Cir. 2019) (en banc) (“This divisibility analysis is nevertheless now clearly required under the Supreme Court’s categorical-approach and modified-categorical-approach jurisprudence.”); Wilson, 978 F.3d at 999 (“Because [the statute] is . . . divisible, we must apply the modified categorical approach.”) (quotation marks omitted); Butts, 40 F.4th at 771 (applying the modified categorical approach to the same statute at issue here “[b]ecause [the statute’s] theft element is divisible”); Ivy, 93 F.4th at 943 (applying the modified categorical approach to a statute with the same structure as the one here because the statute “is twice divisible,” addressing a nearly identical question) (citation and internal quotation marks omitted).

The majority fails to consider this precedent because it incorrectly states that the modified categorical approach applies only if a statute is “overbroad,” i.e., when the statute at issue criminalizes more conduct than the generic offense in question. Majority Op. at 4 (“If a statute is overbroad . . . , we apply a ‘modified’ categorical approach.”). That proposition, however, is contradicted by Wilson, Burris, Butts, and Ivy, and directly conflicts with what the Supreme Court has stated in Mathis. See Mathis, 579 U.S. at 517; see also Hernandez- Maldonado v. Barr, 773 F. App’x 280, 283 n.1 (6th Cir. 2019) (noting that “if a statute is No. 23-3466 United States v. Cervenak Page 31

divisible, the overbreadth analysis is less straightforward, because we must apply the modified categorical approach”) (cleaned up) (emphasis in original). Had the majority considered Cervenak’s Shepard documents and correctly applied the modified categorical approach, it surely would have conducted the divisibility analysis required by both the Supreme Court and this court.

B. The majority’s analysis fails to consider the minimum conduct criminalized by Ohio robbery

The majority’s assumption that Cervenak’s theft offense is “theft” under Ohio Rev. Code Ann. § 2913.02 also conflicts with the Supreme Court’s instruction that courts “must presume that the conviction rested upon nothing more than the least of the acts criminalized.” Moncrieffe v. Holder, 569 U.S. 184, 191–93 (2013) (cleaned up). And this court’s application of Moncrieffe requires the same analysis. See United States v. Burris, 912 F.3d 386, 406 (6th Cir. 2019) (en banc) (explaining that “if the Shepard documents in a particular case do not make clear under which subsection of the relevant statute a defendant was convicted, sentencing courts must ‘presume that the conviction rested upon nothing more than the least of the acts criminalized’” (quoting Moncrieffe, 569 U.S. at 190–91)); see also Ivy, 93 F.4th at 946 (“[W]e must consider the minimum conduct criminalized by any of [the theft offenses].”) (cleaned up) (emphasis in original).

The majority bypasses this required analysis. It instead critiques this court’s recent decision in Ivy, Majority Op. at 9, 12, ventures off into interpreting the “regular practice” of what Ohio courts supposedly require to “prosecute[] unspecified [Ohio] robbery,” id. at 9, and correspondingly examines when “defendants raise a sufficiency-of-the-evidence challenge against an unspecified robbery conviction.” Id. at 10.

C. The majority’s five additional errors

Based on the majority’s citation of a few Ohio prosecutions in which § 2913.02(A) theft was the predicate theft offense at issue, the majority concludes that “it seems that when Ohio prosecutes an unspecified robbery charge, the state must prove to a jury the defendant committed No. 23-3466 United States v. Cervenak Page 32

2913.02(A) theft beyond a reasonable doubt.” Id. at 11. I believe that this conclusion is erroneous for the following five reasons:

1. The majority misapplies the categorical approach

The first and most fundamental problem with the majority’s analysis is that it misapplies the categorical approach and therefore uses the wrong legal standard. In discussing why § 2913.02(A) theft is an element of unspecified robbery convictions, the majority states that “Ohio courts appear to treat 2913.02(A) theft as an ‘element’ of unspecified robbery convictions.” Id. at 11.

But when analyzing the elements of a state crime, the categorial approach demands certainty, not simply an interpretation of prosecutorial trends. The Supreme Court has repeatedly emphasized that a “state offense is a categorical match only if a conviction of that offense ‘necessarily involved . . . facts equating to [the] generic [federal offense].’” Moncrieffe, 569 U.S. at 184–85 (quoting Shepard v. United States, 544 U.S. 13, 24 (2005)) (alterations in original) (emphasis added).

Beginning in Taylor v. United States, 495 U.S. 575, 599 (1990), the Supreme Court explained that sentencing courts may look to charging documents and jury instructions under the categorical approach only where “the jury necessarily had to find” facts equating the generic offense with what the defendant was convicted of. Id. at 602. In Shepard, the Supreme Court re-emphasized the certainty required when analyzing the elements of a defendant’s conviction under the categorial approach. The Court upheld the “demanding requirement that any sentence . . . rest on a showing that a prior conviction ‘necessarily’ involved . . . facts equating to [a] generic” offense. Shepard, 544 U.S. at 24 (quoting Taylor v. United States, 495 U.S. 575, 599 (1990)).

In Moncrieffe, the Supreme Court summarized the applicable standard from Taylor and Shepard as follows: “[T]o satisfy the categorical approach, [the] state . . . offense . . . must ‘necessarily’ prescribe conduct that is an offense under the [federal generic definition].” Moncrieffe, 569 U.S. at 192. In other words, a “state offense is a categorical match . . . only if a No. 23-3466 United States v. Cervenak Page 33

conviction of the state offense ‘necessarily involved . . . facts equating to [the] generic [federal offense].’” Id. at 190 (alterations in original).

In the present case, the majority’s analysis errs because it falls well short of concluding that an unspecified Ohio robbery “necessarily involved” § 2913.02(A) theft. See id. The majority instead interprets prosecutorial trends by using language such as “it seems” and “[it] appear[s]” to qualify its conclusion that § 2913.02(A) theft is always the appropriate theft offense for unspecified Ohio robbery. Majority Op. at 11. But this is irreconcilable with the categorical approach. Moncrieffe, 569 U.S. at 194–95 (“Ambiguity [about which statutory provision applies to a conviction] means that the conviction did not ‘necessarily’ involve facts that correspond to an offense punishable as [a crime of violence].”).

The majority’s related error stems from its interpretation of how Ohio allegedly prosecutes robbery, which leads the majority to conclude that “theft” is always the default “theft offense” under Ohio robbery. Majority Op. at 11. But this type of atextual analysis was rejected by the Supreme Court in Moncrieffe, where the Supreme Court examined whether the state drug statute at issue established conduct that would be punishable as a felony under the Controlled Substances Act (CSA). The CSA contains a felony provision and a misdemeanor provision with respect to certain drug offenses. Based on how the relevant drug crimes were prosecuted, the government argued that the CSA’s felony provision was the appropriate comparator. Specifically, the government asserted that “the felony provision [is] the default because, in practice, that is how . . . criminal prosecutions for marijuana distributions operate.” Moncrieffe, 569 U.S. at 196–97.

But the Supreme Court rejected this argument because the text of the statute foreclosed the government’s interpretation. The Court reasoned that “[t]he CSA’s text makes neither provision the default,” and that “we cannot discount [the statute’s] text, . . . which creates no default punishment . . . .” Id. at 196–97. It therefore concluded that the state statute was not a categorical match for the generic federal offense at issue.

In the present case, the majority has made the same error as the government did in Moncrieffe. The majority specifically examines Ohio robbery cases by interpreting how No. 23-3466 United States v. Cervenak Page 34

“criminal prosecutions . . . operate,” id., and then crafts a perceived default theft offense in conflict with the text of Ohio’s robbery statute. Majority Op. at 11 (“[I]t seems that when Ohio prosecutes an unspecified robbery charge, the state must prove to a jury the defendant committed 2913.02(A) theft beyond a reasonable doubt.”). But as the Supreme Court determined in Moncrieffe, “we cannot discount [a statute’s] text, . . . which creates no default punishment.” Moncrieffe, 569 U.S. at 197. Tellingly, the majority has failed to identify anything in the text of Ohio’s robbery statute that creates a default theft offense. The majority therefore employs a form of analysis explicitly rejected by the Supreme Court.

2. Ohio courts have upheld “unspecified robbery convictions” under other predicate theft offenses

As stated by the majority, “it seems that” Ohio courts require prosecutors to prove § 2913.02(A) theft beyond a reasonable doubt for all “unspecified robbery charge[s].” Majority Op. at 11. The majority relatedly asserts that “when a defendant’s indictment does not list a specific theft offense, the ‘regular practice’ of Ohio courts is to require that the state prove the defendant committed ‘theft’ under section 2913.02(A).” Id.

But that is not the “regular practice” of the Ohio courts. In State v. Dixon, No. 89-CA-22, 1990 WL 94612 (Ohio Ct. App. June 20, 1990), for example, the defendant was convicted of Ohio robbery where his indictment identified only “a theft offense.” Id. at *2. The defendant had sought to use a forged identification card to cash a forged check, and when his attempt to cash the forged check failed, he violently tried to exit the store. An Ohio court ultimately upheld the robbery conviction by noting that “a ‘theft offense’ [under Ohio robbery] includes a violation of R.C. 2913.31, ‘forgery.’” Id. And because the evidence supported the defendant’s use of the forged instruments, the court upheld the robbery conviction because the appropriate predicate theft offense was “forgery” under Ohio Rev. Code Ann. § 2913.31. Id. The Dixon case therefore belies the majority’s conclusion that § 2913.02(A) theft is always an element of unspecified robbery convictions. No. 23-3466 United States v. Cervenak Page 35

3. Ohio courts rely on evidence that federal courts are not permitted to review, so comparing our analysis to what Ohio courts do on direct review of a conviction is inappropriate

The next problem with the majority’s analysis is that what Ohio appellate courts do on direct review is an inappropriate analog for our review under the categorical approach. Under the categorical approach, we may not consider the defendant’s actual conduct in determining whether an offense is a crime of violence under the Guidelines. See Mathis, 579 U.S. at 510 (“A sentencing judge may look only to ‘the elements of the [offense], not to the facts of [the] defendant’s conduct.’”) (alterations in original) (quoting Taylor v. United States, 495 U.S. 595, 601 (1990)). Indeed, “we examine what the state conviction necessarily involved, not the facts underlying the case.” Moncrieffe, 569 U.S. at 190.

But Ohio appellate courts must ascertain what a defendant actually did to determine whether sufficient evidence supported the conviction at issue. Dixon, 1990 WL 94612, at *2 (explaining that the sufficiency-of-the-evidence standard is based on “an examination of the record at the trial”). Ohio appellate courts therefore have significantly more information to ascertain the appropriate underlying theft offense because that evidence (e.g., trial testimony, recordings, photos, affidavits, police reports, etc.) is revealed at trial.

The cases cited by the majority prove as much. For instance, in the first case cited by the majority, State v. Johnson, No. 11AP-717, 2012 WL 1895790 (Ohio App. Ct. 2012), the court considered various forms of evidence revealed at trial. The court reviewed “six[] videos [that] show[ed] parts of the struggle between appellant and the Kroger employees” during the robbery, as well as trial testimony from three individuals, one of whom testified “that he saw appellant steal two cartons of cigarettes and a pack of lighters.” Id. at *3, *8. In another case cited by the majority, State v. McCloud, 2018-Ohio-3291, 2018 WL 3954284 (Ohio App. Ct. 2018), the court considered “testimony and significant circumstantial evidence, including [a] security video and the contemporaneous observations of a police officer,” which led the jury to conclude that “[the defendant] took [the victim’s] money.” Id. at *3.

Ohio appellate courts are therefore not, as the majority suggests, choosing § 2913.02(A) theft because it is the only available “theft offense” to support an “unspecified robbery No. 23-3466 United States v. Cervenak Page 36

conviction.” Rather, Ohio appellate courts are determining the appropriate theft offense, as revealed through trial evidence, which happens to be § 2913.02(A) theft in certain cases based on all the facts in the record.

But under the categorical approach, we may not consider this extra evidence or conduct a factual analysis. The categorical approach in fact bars us from looking at the slew of evidence that Ohio appellate courts must consider in the particular case at issue. See Mathis v. United States, 579 U.S. 500, 505 (2016) (holding that a court employing the categorical approach “looks to a limited class of documents”). As a result, to compare our analysis under the categorical approach with what an Ohio appellate court does on direct review is inappropriate given (1) the stark difference in what we may review, and (2) the fundamental difference in the analysis that each court is carrying out.

4. The majority’s critique of this court’s decision in Ivy is misplaced

Still another reason why the majority’s analysis is erroneous is because it fails to heed this court’s recent decision in United States v. Ivy, 93 F.4th 937 (6th Cir. 2024). According to the majority:

Cervenak’s Shepard documents . . . do not list the specific theft offense . . . . [I]t might be tempting to follow Ivy’s resolution . . . [to] presume his conviction rested upon the least-culpable offense permitted by the statute’s text, . . . [but] . . . [w]hile tempting, this line of reasoning conflicts with a key tenet of the categorical approach: state courts have the final say on state law.

Majority Op. at 8–9.

The problem with the majority’s “key tenet” is that it flatly conflicts with what this court requires when the Shepard documents do not specify a statutory subsection under which a defendant was convicted in state court. Specifically, this court’s decision in United States v. Burris, 912 F.3d 386, 406 (6th Cir. 2019) (en banc) (quoting Moncrieffe v. Holder, 569 U.S. 183, 190–91 (2013)), states that “if the Shepard documents in a particular case do not make clear under which subsection of the relevant statute a defendant was convicted, sentencing courts must ‘presume that the conviction rested upon nothing more than the least of the acts criminalized.’” And more problematically, the majority’s new tenet conflicts with the Supreme Court’s No. 23-3466 United States v. Cervenak Page 37

instruction that “we must presume that the conviction rested upon nothing more than the least of the acts criminalized.” Moncrieffe, 569 U.S. at 190–91 (cleaned up). The Ivy panel, by contrast, applied these precedents because that is what the Supreme Court in Moncrieffe requires, and also what this court’s en banc decision in Burris mandates.

5. The “realistic probability” analysis from Gonzales v. Duenas- Alvarez, 549 U.S. 183, 193 (2007), does not apply in the present case

My fifth and final reason for disagreeing with the majority’s analysis is based on its conclusion that Cervenak’s argument fails because he must point to an Ohio case applying Ohio robbery to conduct that falls outside of the Guidelines’ definition of extortion. To the contrary, “where the meaning of the statute is ‘plain,’ the defendant need not provide a case to demonstrate a realistic probability that the statute is broader than the generic offense.” United States v. Camp, 903 F.3d 594, 602 (6th Cir. 2018) (quoting United States v. Lara, 590 F. App’x 574, 584 (6th Cir. 2014)).

In Ivy, this court analyzed the text of two predicate theft offenses: (1) trespassing in a habitation when a person is present or likely to be present under § 2911.12(B), and (2) breaking and entering under § 2911.13(B). As the panel in Ivy concluded, “the plain text reveals, in committing these offenses, an offender is not required to take anything of value from another person.” Ivy, 93 F.4th at 946–47. This conclusion is consistent with decisions by Ohio courts applying these two statutes. See State v. Stone, 2024-Ohio-177, 2024 WL 208125, at *1 (Ohio Ct. App. Jan. 19, 2024) (upholding a conviction under Ohio Rev. Code. Ann. § 2911.12(B) where the defendant did not obtain anything of value); State v. Dunn, 2023-Ohio-4413, 2023 WL 8474188, at *6 (Ohio Ct. App. Dec. 7, 2023) (upholding a conviction under § 2911.13(B) where the defendant did not obtain anything of value). Because Ivy has already concluded that the text of these two predicate offenses is plain, we should not be engaging in a “realistic probability” analysis here.

The majority’s rationale also fails because its justification for applying the “realistic probability” analysis from Gonzales v. Duenas-Alvarez, 549 U.S. 183, 193 (2007), has no applicability to the present case. In Duenas-Alvarez, the defendant—a noncitizen contesting his removal—sought to compare the generic meaning of “theft” as defined in federal law to a No. 23-3466 United States v. Cervenak Page 38

California theft statute. He argued that the California statute criminalized more conduct than generic theft under federal law. The defendant specifically maintained that California’s “natural and probable consequence” doctrine extended the statute to aiding-and-abetting crimes (not covered by the plain language of the statute) in a way that was not applicable to generic theft. Under the defendant’s interpretation, he would not be subject to deportation because he did not commit a “theft” under federal law.

The defendant’s argument nevertheless failed because he could not provide examples showing that California enforced its theft statute in a distinct manner from how generic theft was enforced under federal law. And the Supreme Court was concerned that the defendant’s unique interpretation of the California law would be nothing “more than the application of legal imagination.” Id. In light of the defendant’s “legal imagination,” the Supreme Court demanded a showing that there would be a realistic probability that the statute is enforced in the unique way advanced by the defendant. Id.

But here, our analysis focuses on 31 discrete theft offenses plainly listed in Ohio’s robbery statute that contain statutorily defined elements. See § 2913.01(K). And Ohio courts have upheld convictions under at least two of the theft offenses where the defendants did not obtain anything of value. See State v. Stone, 2024-177, 2024 WL 208125, at *1 (Ohio Ct. App. Jan. 19, 2024); State v. Dunn, 2023-Ohio-4413, 2023 WL 8474188, at *6 (Ohio Ct. App. Dec. 7, 2023). This key fact clearly shows “more than the application of legal imagination to [the statute’s] language.” Duenas-Alvarez, 549 U.S. at 193; see also Camp, 903 F.3d at 602 (“[W]here the meaning of the statute is ‘plain,’ the defendant need not provide a case to demonstrate a realistic probability that the statute is broader than the generic offense.”) (citation omitted). The majority’s analysis to the contrary is akin to “legal imagination,” given that its analysis effectively removes 30 of the 31 theft offenses from Ohio’s robbery statute. See Duenas-Alvarez, 549 U.S. at 193.

Moreover, in United States v. Taylor, 596 U.S. 845, 857–58 (2022), the Supreme Court provided another situation where courts should find the analysis from Duenas-Alvarez inapplicable. That is where there is no overlap between the two offenses at issue when comparing them to determine whether one of the offenses is a crime of violence. In that specific No. 23-3466 United States v. Cervenak Page 39

case, the Court analyzed whether attempted Hobbs Act robbery qualified as a crime of violence under 18 U.S.C. § 924 (c)(3)(A), and the Court held that it did not.

The Supreme Court also highlighted why Duenas-Alvarez was inapplicable by noting: “In Duenas-Alvarez, the elements of the relevant state and federal offenses clearly overlapped and the only question the Court faced was whether the state courts also applied the statute in a special (nongeneric) manner.” Taylor, 596 U.S. at 859 (cleaned up). It also stated that “[h]ere, we do not reach that question because there is no overlap to begin with. Attempted Hobbs Act robbery does not require proof of any of the elements §924(c)(3)(A) demands. That ends the inquiry, and nothing in Duenas-Alavrez suggests otherwise.” Id. (emphasis in original).

So too here. The theft offenses described in § 2911.12(B) and § 2911.13(B), which are 2 of the 31 “theft offenses” listed in §2913.01(K), do not require proof of any of the elements that Guidelines extortion demands. In particular, nothing in those two theft offenses require an offender to take “something of value” from another person. See U.S.S.G. § 4B1.2. This lack of overlap is thus another reason why Duenas-Alvarez does not apply. The government’s argument to the contrary (that the majority adopts) was rejected because the Supreme Court deemed it “atextual” and “doubtful on its face” where the text of the statutes was clear. Id. at 858.

In sum, the meanings of at least two predicate theft offenses in Ohio’s robbery statute are plain, obviating the need to engage in the “realistic probability” analysis. See Camp, 903 F.3d at 602. Endorsing the majority’s proposition to the contrary contravenes the decisions in both Camp and Ivy, as well as the Supreme Court’s admonishment of similar “atextual” arguments made by the government in Taylor.

And as Taylor noted more generally, a defendant’s failure to show that the government prosecutes the crime in question is irrelevant when the text of the relevant statutes clearly shows a categorical mismatch:

The government faults [the defendant] for failing to identify a single case in which it has prosecuted someone for attempted . . . robbery without proving a communicated threat. But what does that prove? . . . Put aside the oddity of placing a burden on the defendant to present empirical evidence about the government’s own prosecutorial habits . . . . Put aside, too, the practical challenges such a burden would present in a world where most cases end in plea No. 23-3466 United States v. Cervenak Page 40

agreements, and not all of those cases can make their way into easily accessible commercial databases. An even more fundamental problem lurks here . . . . The government’s theory cannot be squared with the statute’s terms.

Taylor, 596 U.S. at 857

V. OHIO ROBBERY WITHOUT A LISTED PREDICATE THEFT OFFENSE IS NOT A CRIME OF VIOLENCE UNDER THE ENUMERATED-OFFENSES CLAUSE OF THE SENTENCING GUIDELINES

Having explained why I disagree with the majority’s analysis, I now turn to how I would resolve this case. I begin by analyzing whether Ohio robbery without a listed predicate theft offense is a categorical match for extortion under the Guidelines. Next, I will proceed by analyzing whether Ohio robbery without a listed predicate theft offense is a categorical match for robbery under the Guidelines.

A. Ohio robbery without a listed predicate offense criminalizes more conduct than extortion as defined in the Guidelines

Ohio robbery that inflicts, attempts to inflict, or threatens to inflict physical harm on another, without a listed predicate offense, is a crime of violence under the Guidelines’ enumerated-offenses clause only if its “elements are the same as, or narrower than,” the elements of extortion. Camp, 903 F.3d at 600 (quoting Descamps v. United States, 570 U.S. 254, 257 (2013)). The Guidelines’ commentary defines extortion, so “[w]e look to that definition rather than a generic definition.” Id. at 602. At the time of Cervenak’s sentencing, the Guidelines defined extortion as “obtaining something of value from another by the wrongful use of (A) force, (B) fear of physical injury, or (C) threat of physical injury.” U.S.S.G. § 4B1.2 (2018).

This court’s task is to decide if a conviction for Ohio robbery under § 2911.02(A)(2) always requires an offender to obtain “something of value” from another person. See id. If Ohio robbery so requires, then it does not cover more conduct than the Guidelines’ definition of extortion. But if it does not, then Ohio robbery criminalizes more conduct than the Guidelines definition of extortion and, as such, it cannot be considered a crime of violence.

“Because § 2911.02[(A)(2)]’s theft element is divisible, we look to the so-called Shepard documents, or state-court records reflecting the charge, to determine which Ohio theft offense No. 23-3466 United States v. Cervenak Page 41

predicated [Cervenak’s] robbery charge.” See United States v. Butts, 40 F.4th 766, 771 (6th Cir. 2022). The state-court records here do not list the underlying predicate offense. And where the “Shepard documents in a particular case do not make clear under which subsection of the relevant statute a defendant was convicted, sentencing courts must ‘presume that the conviction rested upon nothing more than the least of the acts criminalized.’” United States v. Burris, 912 F.3d 386, 406 (6th Cir. 2019). (quoting Moncrieffe v. Holder, 569 U.S. 183, 190–91 (2013)); see also Ivy, 93 F.4th at 946 (“[W]e must consider the minimum conduct criminalized by any of [the theft offenses].”) (cleaned up) (emphasis in original).

Not all of the predicate theft offenses listed in § 2913.01(K) require a defendant to “obtain[] something of value from another,” as required by the Guidelines definition of extortion. See U.S.S.G. § 4B1.2. Among the 31 “theft offenses” listed in § 2913.01(K) is § 2911.12(B), which deals with trespassing in a habitation when a person is present or likely to be present. That statute provides that “[n]o person, by force, stealth, or deception, shall trespass in a permanent or temporary habitation of any person when any person other than an accomplice of the offender is present or likely to be present.” Ohio Rev. Code. Ann. § 2911.12(B). Another “theft offense” referred to in § 2913.01(K) is breaking and entering. That statute proscribes “trespass[ing] on the land or premises of another, with purpose to commit a[ny] felony.” Id. § 2911.13(B).

With respect to the above two predicate offenses, this court in Ivy explained that, “[a]s the plain text reveals, in committing these offenses, an offender is not required to take anything of value from another person.” Ivy, 93 F.4th at 946–47; see also Brown v. United States, No. 22-3470, 2024 WL 778119, at *3 (6th Cir. 2024) (“A person can commit Ohio aggravated robbery, however, without obtaining a thing of value from another.”). These theft offenses plainly show that a § 2911.02(A)(2) robbery conviction is not limited to the Guidelines’ definition of extortion. And because Ohio robbery without a specified predicate theft offense criminalizes more conduct than Guidelines extortion, Cervenak’s robbery conviction is not a categorical match. No. 23-3466 United States v. Cervenak Page 42

B. Ohio robbery without a listed predicate offense criminalizes more conduct than robbery as defined in the Guidelines

Carter dealt only with extortion and did not address how the Guidelines define robbery. This was presumably because, at the time of Cervenak’s sentencing, the Guidelines did not define the term. When the Guidelines do not define an enumerated offense, we use the offense’s generic meaning. See Camp, 903 F.3d at 602. To ascertain that meaning, we look at “how the crime is described across jurisdictions, as well as consulting sources such as the Model Penal Code.” United States v. Rede-Mendez, 680 F.3d 552, 556 (6th Cir. 2012) (citing Taylor v. United States, 495 U.S. 575, 598 & n.8 (1990)).

This court in Ivy held that generic robbery under the Guidelines “requires a defendant to (1) misappropriate property in a way that involves ‘immediate danger’ to another person (2) with a mens rea of at least recklessness.” Ivy, 93 F.4th at 944. The court further concluded that, when considering the least serious conduct with respect to the same list of theft offenses at issue here, “an offender may commit many other theft offenses without immediate danger to another person.” Id. at 945. After analyzing the mens rea of aggravated robbery, the court noted that the statute has no mens rea requirement, that a hypothetical defendant could “negligently” commit aggravated robbery, and that the Ohio courts would uphold the conviction. Id. As a result, this court in Ivy found that aggravated robbery does not fit within the meaning of generic robbery because generic robbery has a means rea of at least recklessness, whereas Ohio aggravated robbery can be committed with a mens rea of only negligence.

The Ohio robbery statute in the present case, like the related statute in Ivy, does not contain a mens rea element. Instead “the underlying theft offense specifies the mens rea that the state must prove to convict a person of robbery.” Butts, 40 F.4th at 770 (citing State v. Tolliver, 19 N.E.3d 870, 875 (Ohio 2014)). So, like the statute in Ivy, Ohio robbery without a listed predicate offense in the charging documents “has no means rea . . . [and] covers more conduct than the generic offense of robbery, and [thus] does not constitute a crime of violence under the enumerated-offenses clause.” Ivy, 93 F.4th at 945 (cleaned up). No. 23-3466 United States v. Cervenak Page 43

VI. CONCLUSION

In sum, because Carter did not address the arguments made before us, Carter does not control our analysis. I would instead resolve this case by looking to Moncrieffe, Mathis, Butts, Wilson, and Ivy rather than to Carter because doing so correctly applies the binding precedent of both this court and the Supreme Court. Under my analysis, I would vacate Cervenak’s sentence and remand the case to the district court for resentencing. Accordingly, I respectfully dissent.