Carlos Rendon v. Eric Holder, Jr.

FOR PUBLICATION

UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT

CARLOS ALBERTO RENDON, No. 10-72239 Petitioner, Agency No. v. A092-080-719

ERIC H. HOLDER, JR., Attorney General, ORDER Respondent.

Filed April 2, 2015

Before: Stephen Reinhardt, Raymond C. Fisher, and Mary H. Murguia, Circuit Judges.

Order; Dissent by Judge Graber; Dissent by Judge Kozinski

2 RENDON V. HOLDER

SUMMARY*

Immigration

The panel filed an order denying a petition for panel rehearing and, on behalf of the court, rejecting a sua sponte en banc call.

In its opinion filed August 22, 2014, the panel granted Carlos Alberto Rendon’s petition for review from the Board of Immigration Appeals’ decision finding him statutorily ineligible for cancellation of removal based on his conviction for attempted second-degree burglary, in violation of California Penal Code § 459. The panel held that under Descamps v. United States, 133 S. Ct. 2276 (2013), § 459 is indivisible and its use of disjunctive language did not render it divisible. The panel held that determining whether a disjunctively worded statute is divisible requires looking to whether the state treats the parts of the statute on opposite sides of the “or” as alternative elements or alternative means.

Dissenting from the denial of rehearing en banc, Judge Graber, joined by Judges O’Scannlain, Gould, Tallman, Bybee, Callahan, Bea, and Ikuta, wrote that when applying the modified categorical approach to a disjunctive statute, Descamps holds that the court must consult the Shepard documents to determine whether what the jury found or what the defendant pleaded to matches the federal definition of the relevant crime.

* This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader.

RENDON V. HOLDER 3

Dissenting from the denial of rehearing en banc, Judge Kozinski wrote to suggest an alternative reading of Descamps’s footnote 2. Judge Kozinski’s reading would, when confronted with a disjunctively-phrased statute, look to Shepard documents for the limited purpose of determining the statute’s elements. Judge Kozinski wrote that a court should first look to the documents to determine a crime’s elements, and on that basis decide whether the categorical or modified categorical approach is appropriate.

ORDER

The panel has voted to deny the petition for panel rehearing.

A sua sponte call for a vote on rehearing this case en banc was made by an active judge of this court. The call failed to receive a majority of the votes of the nonrecused active judges. Fed. R. App. P. 35. The sua sponte en banc call is rejected.

Judge Kozinski’s dissent from denial of rehearing en banc and Judge Graber’s dissent from denial of rehearing en banc are filed concurrently with this Order.

4 RENDON V. HOLDER

GRABER, Circuit Judge, with whom O’SCANNLAIN, GOULD, TALLMAN, BYBEE, CALLAHAN, BEA, and IKUTA, Circuit Judges, join, dissenting from the denial of rehearing en banc:

I respectfully but emphatically dissent from the denial of rehearing en banc. When applying the modified categorical approach, what must we do when we confront a textually disjunctive statute and are uncertain which statutory alternative underlies a previous conviction? The Supreme Court has answered that question in no uncertain terms: in the face of a disjunctive statute, we consult the documents of conviction, also known as the “Shepard documents,” to learn whether what the jury actually found, or what the defendant actually pleaded to, matches the federal definition of the relevant crime. The Court even told us what not to do: we need not parse state law to determine whether the statutory alternatives are “elements” or “means”; rather, we look to the Shepard documents whether the alternatives are elements or means. The panel’s opinion simply ignores, without explanation, the Court’s clear command. Remarkably, the opinion holds that we must do precisely what the Court instructed us not to do: parse state law to determine whether the statutory alternatives are elements or means. I regret our decision not to rehear this case en banc in order to respect our role as an intermediate appellate court that must follow the Supreme Court’s instructions.

A. The Court Expressly Rejected Rendon’s Approach

Petitioner Carlos Alberto Rendon was convicted of burglary as defined by California Penal Code section 459: “Every person who enters [certain locations] with intent to commit grand or petit larceny or any felony is guilty of

RENDON V. HOLDER 5

burglary.” The inquiry in this case hinges on intent: “intent to commit . . . larceny or any felony.” Id. (emphasis added). Everyone agrees that, if Petitioner was convicted of entering with intent to commit larceny, then the conviction meets the requisite intent element under the relevant immigration provision but that, if Petitioner was convicted of entering with intent to commit some other felony, such as arson, then the conviction does not meet the requisite intent element under the relevant immigration provision. Rendon v. Holder, 764 F.3d 1077, 1084 (9th Cir. 2014)

Analyzing the Supreme Court’s decision in Descamps v. United States, 133 S. Ct. 2276 (2013), Rendon holds that, for textually disjunctive statutes, we may look to the Shepard documents only if the alternatives are elements, such that the statute describes separate crimes; we may not look to the Shepard documents if the statute lists only alternative means of committing the same crime. Rendon, 764 F.3d at 1084–85. To make the elements/means determination, Rendon requires that we analyze state law to determine whether the disjunctive alternatives require juror unanimity. Id. at 1088–90

The Supreme Court expressly and unambiguously rejected that approach in Descamps. Descamps involved a conviction under the same California burglary statute. 133 S. Ct. at 2282. In the body of the opinion, the Court considered whether we may examine the Shepard documents to determine whether the entry was lawful. Id. at 2283–93. By contrast to the statutory list of alternative intents, the text of the statute says nothing about the lawfulness of the entry; it merely states: “Every person who enters . . . .” Cal. Penal Code § 459. The Court held that, for facts that lack a statutory hook (such as the lawfulness of the entry), courts may not look to the Shepard documents unless the fact is an

6 RENDON V. HOLDER

element of the crime, requiring juror unanimity. Descamps, 133 S. Ct. at 2285–86.

But the Court held that a different rule applies if the text of the statute contains a list of alternatives. Writing in dissent, Justice Alito used the methodology now mandated by Rendon: he examined the state statutes at issue in the Court’s previous modified-categorical-approach cases—statutes with an express list of alternatives in the statutory text. Descamps, 133 S. Ct. at 2297–98 & n.2 (Alito, J., dissenting). He reviewed state court decisions concerning juror unanimity—just as Rendon now instructs—and discovered that the state courts might not require juror unanimity with respect to at least two of the statutes in the earlier cases; in the abstract, those statutes likely described alternative means, not alternative elements. Id. Justice Alito then argued that the majority’s analysis—at least as he understood it—contradicted the Court’s earlier cases. Id.

The Descamps majority’s response is key, so I repeat it here in full:

The dissent delves into the nuances of various States’ laws in an effort to cast doubt on this understanding of our prior holdings, arguing that we used the modified categorical approach in cases like Taylor, Shepard, and Johnson “in relation to statutes that may not have been divisible” in the way that we have just described. Post, at 2297 (ALITO, J.). But if, as the dissent claims, the state laws at issue in those cases set out “merely alternative means, not alternative elements” of an offense, post, at 2298, that is news to us. And

RENDON V. HOLDER 7

more important, it would have been news to the Taylor, Shepard, and Johnson Courts: All those decisions rested on the explicit premise that the laws “contained statutory phrases that cover several different crimes,” not several different methods of committing one offense. Johnson, 559 U.S., at 144 (citing Nijhawan [v. Holder, 557 U.S. 29, 41 (2009)]). And if the dissent’s real point is that distinguishing between “alternative elements” and “alternative means” is difficult, we can see no real-world reason to worry. Whatever a statute lists (whether elements or means), the documents we approved in Taylor and Shepard—i.e., indictment, jury instructions, plea colloquy, and plea agreement—would reflect the crime’s elements. So a court need not parse state law in the way the dissent suggests: When a state law is drafted in the alternative, the court merely resorts to the approved documents and compares the elements revealed there to those of the generic offense.

Descamps, 133 S. Ct. at 2285 n.2 (majority op.) (brackets and ellipsis omitted) (boldface type emphases added).

The Descamps majority could not have been plainer. When there is no statutory list, it is critical whether a fact is an element of the crime. But the elements/means distinction is not relevant if the statute lists alternatives: “Whatever a statute lists (whether elements or means),” courts should consider “the documents we approved in Taylor and Shepard.” Id. (emphasis added). “So a court need not parse

8 RENDON V. HOLDER

state law in the way the dissent[, and now the Rendon panel opinion,] suggests.” Id. “When a state law is drafted in the alternative, the court merely resorts to the approved documents and compares the elements revealed there to those of the generic offense.” Id. (emphasis added). In other words, it does not matter whether, in the abstract, the statutory phrases describe elements or means; what matters is only what the Shepard documents in that particular case say. We “need not” look to state court cases; we “merely”—that is, only—look to the Shepard documents. Id.

In scores of cases, we and our sister circuits have applied that simple instruction and have looked to the Shepard documents whenever confronted with a textually disjunctive statute.1 That approach is clearly correct: it is not our job, as

1 See, e.g., United States v. Gomez, 757 F.3d 885, 900 n.13 (9th Cir. 2014); United States v. Quintero-Junco, 754 F.3d 746, 752 (9th Cir. 2014); Ragasa v. Holder, 752 F.3d 1173, 1176 (9th Cir. 2014); Murillo- Prado v. Holder, 735 F.3d 1152, 1156–57 (9th Cir. 2013) (per curiam); Duenas-Alvarez v. Holder, 733 F.3d 812, 815 (9th Cir. 2013); United States v. Ceron, 775 F.3d 222, 228 (5th Cir. 2014) (per curiam); Franco- Casasola v. Holder, 773 F.3d 33, 38 (5th Cir. 2014); Avendano v. Holder, 770 F.3d 731, 734 (8th Cir. 2014); United States v. Trent, 767 F.3d 1046, 1057 (10th Cir. 2014), cert. denied, 2015 WL 732162 (U.S. Feb. 23, 2015) (No. 14-7762); United States v. Prater, 766 F.3d 501, 510 (6th Cir. 2014); United States v. Brown, 765 F.3d 185, 192 (3d Cir. 2014); United States v. Estrella, 758 F.3d 1239, 1248 (11th Cir. 2014); United States v. Martinez, 756 F.3d 1092, 1095–96 (8th Cir. 2014); Garcia v. Holder, 756 F.3d 839, 844 (5th Cir. 2014) (per curiam); United States v. Conde- Castaneda, 753 F.3d 172, 176 (5th Cir.), cert. denied, 135 S. Ct. 311 (2014); United States v. Herrera-Alvarez, 753 F.3d 132, 139–40 (5th Cir. 2014); United States v. Carter, 752 F.3d 8, 17 (1st Cir. 2014); United States v. Mitchell, 743 F.3d 1054, 1065–66 (6th Cir.), cert. denied, 135 S. Ct. 158 (2014); United States v. Marrero, 743 F.3d 389, 396 (3d Cir. 2014), cert. denied, 135 S. Ct. 950 (2015); Sarmientos v. Holder, 742 F.3d 624, 629 (5th Cir. 2014); United States v. Covington, 738 F.3d 759, 764

RENDON V. HOLDER 9

an intermediate appellate court, to overrule the Supreme Court’s plain and applicable pronouncements, whether brilliantly or poorly reasoned. Indeed, the Supreme Court repeatedly has held that we must follow its holdings unless and until the Supreme Court itself overrules them. See United States v. Hatter, 532 U.S. 557, 567 (2001) (“The Court of Appeals was correct in applying [existing precedent] to the instant case, given that ‘it is this Court’s prerogative alone to overrule one of its precedents.’” (quoting State Oil Co. v. Khan, 522 U.S. 3, 20 (1997))); Agostini v. Felton, 521 U.S. 203, 237 (1997) (“We reaffirm that ‘if a precedent of this Court has direct application in a case, yet appears to rest on reasons rejected in some other line of decisions, the Court of Appeals should follow the case which directly controls, leaving to this Court the prerogative of overruling its own decisions.’” (brackets omitted) (quoting Rodriguez de Quijas v. Shearson/Am. Express, Inc., 490 U.S. 477, 484 (1989))). There is no authority for the proposition that we may ignore part of an opinion if we have doubts about its correctness or wisdom. No further inquiry or analysis is required or permitted: when the Supreme Court has instructed us clearly and unambiguously, we must follow its instructions.

B. Footnote 2 of Descamps Is Consistent With the Main Holding

I recognize that some judges have expressed concern about how to reconcile footnote 2 with the more general holding of Descamps and that some cases have applied

(6th Cir. 2014); United States v. Hockenberry, 730 F.3d 645, 669 (6th Cir. 2013), cert. denied, 134 S. Ct. 1044 (2014); United States v. Barker, 723 F.3d 315, 320 (2d Cir. 2013) (per curiam).

10 RENDON V. HOLDER

Rendon’s methodology, mostly by ignoring Descamps’ footnote 2.2 I agree that, at first blush, it is not obvious how to reconcile the reasoning in the main body of Descamps, which concerned facts with no statutory hook, with the Court’s instructions in footnote 2, which concerned statutory alternatives. If the elements/means distinction is critical to the former category, as the Court held, it is not self-evident why the distinction is irrelevant to the latter category, as the Court also held.

Closer examination reveals that the two holdings may be reconciled. As an initial matter, much of the Court’s description in the text foreshadows its holding, in footnote 2, that alternative statutory phrases are different from general facts with no statutory hook. “‘The “modified categorical approach” that we have approved permits a court to determine which statutory phrase was the basis for the conviction.’” Descamps, 133 S. Ct. at 2285 (brackets omitted) (emphasis added) (quoting Johnson v. United States, 559 U.S. 133, 144 (2010)). Accordingly, we may “determine which statutory phrase [here, intent to commit larceny or any felony] was the basis for the conviction.” Id. We look to the Shepard documents when a statute “sets out one or more elements of the offense in the alternative—for example, stating that burglary involves entry into a building or an automobile.” Id. at 2281. Here, the burglary statute “sets out

2 See Omargharib v. Holder, 775 F.3d 192, 198–99 (4th Cir. 2014) (agreeing with Rendon); United States v. Abbott, 748 F.3d 154, 159 (3d Cir. 2014) (looking beyond the statutory text, without explanation); United States v. Pate, 754 F.3d 550, 554–55 (8th Cir.) (same), cert. denied, 135 S. Ct. 386 (2014); Omargharib, 775 F.3d at 201 (Niemeyer, J., concurring) (questioning the meaning of footnote 2); Trent, 767 F.3d at 1058–61 (same); United States v. Howard, 742 F.3d 1334, 1343 n.3 (11th Cir. 2014) (same).

RENDON V. HOLDER 11

one or more elements [the element of intent] in the alternative—for example, stating that burglary involves entry [with intent to commit larceny] or [any felony].” Id.

The textual approach is in no way problematic— constitutionally, statutorily, or otherwise. Consider a state drug-possession criminal statute that contains a list of controlled substances, some of which meet the federal definition and some of which do not. If the jury instructions required the jury to find that the defendant possessed, say, cocaine (a drug on the federal schedules), then we know that the jury unanimously found that the defendant possessed cocaine. But if the instructions required only that the defendant possessed an unspecified “controlled substance,” then we do not know whether the jury found that he or she possessed cocaine or some drug not listed in the federal schedules, such as “khat.” Similarly, if the instructions required the jury to find that the defendant possessed either cocaine or khat, then we also do not know whether the jury found that he or she possessed a federally scheduled drug unless a special verdict form or judgment makes clear that the only substance possessed was cocaine. In all cases, the modified categorical approach will return a match only if we know that what the jury found is within the federal definition. It is irrelevant whether, in the abstract, a jury would have to be unanimous in the unusual case in which evidence suggested that a defendant possibly possessed both cocaine and khat.

The Supreme Court rejected that reasoning with respect to facts with no statutory hook because, at trial, “extraneous facts and arguments may confuse the jury” and, at a plea colloquy, defendants might decline to object to extraneous facts so as not to “irk the prosecutor or court.” Descamps,

12 RENDON V. HOLDER

133 S. Ct. at 2289. Where the text is silent, we might worry that the defendant would not “squabbl[e] about superfluous factual allegations.” Id. But, critically, when it comes to a statutory list of alternatives from which a jury must find—or a defendant must admit—one alternative, those concerns have no realistic practical application. It is exceedingly unlikely that a jury would find, or that a defendant would admit, an inapplicable item on the statutory list. Rather, the Shepard documents allow the court to know whether a defendant admitted, or a jury found, a specific statutory alternative. Although one might conjure up a scenario in which a defendant somehow falsely admitted that he committed the crime involving a particular statutory alternative, the categorical approach ultimately is grounded in reality, not in an exercise in improbable theoretical musings. See Gonzales v. Duenas-Alvarez, 549 U.S. 183, 193 (2007) (requiring, in the application of the categorical approach, “a realistic probability, not a theoretical possibility,” that the state would apply its statute in a particular way).

C. The Issue Is Exceptionally Important

The modified categorical approach arises frequently in both immigration and criminal cases—categories that comprise a substantial majority of our docket. Rendon’s rule will apply to future panels of this court, district court judges in the Ninth Circuit, the Board of Immigration Appeals (“BIA”), and immigration judges. Going forward, instead of simply reading the statutory text, we now will be required to delve into the nuances of a seemingly endless variety of state laws in order to determine whether, in the abstract, a jury must unanimously agree as to which statutory alternative the

RENDON V. HOLDER 13

defendant committed—a notoriously uncertain inquiry.3 In other words, Rendon not only contradicts the Supreme Court’s express holding, but it sets us—and district judges, the BIA, and immigration judges—down a complicated path of examining state law that leads to uncertain results.

Rendon also sows confusion in our existing caselaw. For example, the Supreme Court has looked to the Shepard documents when analyzing textually disjunctive statutes such as burglary in Massachusetts and battery in Florida. Shepard v. United States, 544 U.S. 13, 17 (2005); Johnson, 559 U.S.

3 For example, when Justice Alito examined the juror unanimity question concerning the state statutes of the Court’s three earlier cases, he was able to reach only tentative conclusions with respect to two of them. See, e.g., Descamps, 133 S. Ct. at 2298 (Alito, J., dissenting) (“I have not found a Massachusetts decision squarely on point, but there is surely an argument that the Massachusetts Legislature did not want to demand juror agreement on this question.”); id. (“It is a distinct possibility (one not foreclosed by any Florida decision of which I am aware) that a conviction under this provision does not require juror agreement . . . .”). With respect to the last case, even a tentative conclusion was hard to reach. See id. at 2298 n.2 (“The remaining case, Taylor v. United States, 495 U.S. 575 (1990), may also have involved a statute that was not divisible, but the situation is less clear. . . . It is not entirely clear whether a Missouri court would have required jurors to agree on a particular choice from this list. . . . [I]t is possible that the Missouri court did not mean to say that the type of building was an element . . . .” (citations omitted)); see also Omargharib, 775 F.3d at 200 (Niemeyer, J., concurring) (“While Judge Floyd concludes that the applicable Virginia law defines alternative means, thereby precluding use of the modified categorical approach under current law, I find it especially difficult to comprehend the distinction. Virginia’s law could just as easily be viewed as prescribing two crimes . . . .”); United States v. Ramirez-Macias, 584 F. App’x 818, 820 (9th Cir. 2014) (unpublished) (Hawkins, J., concurring) (stating that, whether California law requires juror unanimity with respect to the type of drug possessed “is fairly open to debate” and “much rides on the answer to this question [of juror unanimity]”).

14 RENDON V. HOLDER

at 137. If we determine, as Justice Alito asserted, Descamps, 133 S. Ct. at 2298, that those crimes list alternative means, will we decline to look to the approved documents, thereby effectively overruling Shepard and Johnson? Or will we hold that those state statutes are exceptions to Rendon’s rule for the arbitrary reason that the Supreme Court happened to decide cases involving those statutes?

A similar analysis applies to our own cases. In cases we decided after Descamps and before Rendon, we simply have looked to the Shepard documents when confronted with a textually disjunctive statute. See, e.g., Murillo-Prado v. Holder, 735 F.3d 1152, 1156–57 (9th Cir. 2013) (using Shepard documents after noting that the Arizona racketeering statute provides a disjunctive list); Duenas-Alvarez v. Holder, 733 F.3d 812, 815 (9th Cir. 2013) (using Shepard documents after noting that the California vehicle-theft statute provides a disjunctive list). If a later panel determines that, applying Rendon’s methodology, the statutes list alternative means, may the panel overrule the earlier cases?4 If not, must the panel simply look to the Shepard documents, even though the panel believes that Rendon forbids that approach?

This is not an idle concern. In Coronado v. Holder, 759 F.3d 977 (9th Cir. 2014), cert. denied, 2015 WL 852424 (U.S. Mar. 2, 2015) (No. 14-461), decided before Rendon, we held that we may look to the Shepard documents to determine the controlled substance involved in a previous conviction

4 Of course, the decision of one three-judge panel cannot overrule another three-judge panel unless intervening higher authority “is clearly irreconcilable with our prior circuit authority,” Miller v. Gammie, 335 F.3d 889, 900 (9th Cir. 2003) (en banc), which is questionable in this context.

RENDON V. HOLDER 15

under California drug-crime laws. Although a post-Rendon case applied that approach, at least one judge questioned the correctness of Coronado under Rendon’s approach. United States v. Ramirez-Macias, 584 F. App’x 818, 820 (9th Cir. 2014) (unpublished) (Hawkins, J., concurring). Petitions for rehearing en banc remain outstanding in Ramirez-Macias and several other cases involving California drug-crime laws.

Those drug laws illustrate the absurdity of Rendon’s approach. In nearly every drug conviction in California, there is only one substance at issue. In order for the jury to convict, it must find that the defendant possessed5 that substance. Similarly, in order for a court to accept a defendant’s plea, the defendant must admit to possessing that substance. It is unclear whether a prosecutor ever charges a defendant—in a single count—with possessing “any one of several” controlled substances. Indeed, despite thousands of drug convictions over decades of enforcement, there is no California court case on point answering whether a jury must be unanimous in that exceedingly unusual circumstance. Yet Rendon seems to require that we ask that exact question. Moreover, if we determine that, in the abstract, juror unanimity is not required, then it seems we must ignore all California drug convictions in every context, even if everyone agrees that there was only one substance at issue in the case at hand. The absurdity of that conclusion is only amplified by the fact that the Supreme Court told us precisely not to undertake this entire inquiry, now mandated by Rendon.

5 I use possession in text, but the analysis applies equally to all drug convictions, including delivery, sale, conspiracy, manufacture, etc.

16 RENDON V. HOLDER

D. Conclusion

Rendon holds that we may look to the Shepard documents only if a statute lists alternative elements, but not if the statute lists alternative means. Rendon, 764 F.3d at 1084–85. Descamps held that we look to the Shepard documents “[w]hatever a statute lists (whether elements or means).” 133 S. Ct. at 2285 n.2. Rendon requires us to look to state law to determine juror unanimity, the same inquiry undertaken by the dissent in Descamps. Rendon, 764 F.3d at 1088–90. Descamps held that “a court need not parse state law in the way the dissent suggests: When a state law is drafted in the alternative, the court merely resorts to the approved documents.” 133 S. Ct. at 2285 n.2. In the face of this clear conflict, Rendon is silent.

As described above, Rendon’s undue and unsupported complicating of the modified categorical approach will have substantial effects on our immigration and criminal cases. But my concern extends beyond even those considerable shores. I am troubled on a deeper level by the panel’s decision simply to ignore an on-point holding of the Supreme Court and by the full court’s decision not to rehear this case en banc to correct that egregious error. The proper functioning of our tiered judicial system requires that we follow the Supreme Court’s holdings, even if we disagree with them. With regret, I dissent from the denial of rehearing en banc.

RENDON V. HOLDER 17

KOZINSKI, Circuit Judge, dissenting from the denial of rehearing en banc:

Judge Graber’s dissental provides many excellent reasons why this case should have been reheard en banc. I write only to suggest an alternative reading of Descamps’s footnote 2, one that I believe better reconciles the footnote’s text with the rest of the Court’s opinion.

In my view, footnote 2 indicates that, when confronted with a statute phrased in the disjunctive, we should look to the Shepard documents for the limited purpose of determining what that statute’s elements are. Footnote 2 states, “Whatever a statute lists (whether elements or means), the [Shepard] documents . . . reflect the crime’s elements.” Descamps v. United States, 133 S. Ct. 2276, 2285 (2013) (emphasis added). It also says, “When a state law is drafted in the alternative, [a] court merely resorts to the approved documents and compares the elements revealed there to those of the generic offense.” Id. (emphasis added). These sentences clearly indicate that the Court believed a crime’s elements can be discerned from the Shepard documents, and that we are permitted to look to them for that purpose.

This does not mean, however, that we should use the modified categorical approach irrespective of whether a statute lists “elements” or “means.” Rather, Descamps permits us to peek at the Shepard documents in order to determine which approach to use. If those documents show that a statutory term was an element of the offense, we may employ the modified categorical approach and use the documents to determine whether a defendant committed a state crime falling within the ambit of the relevant federal statute. But if the Shepard documents instead show that the

18 RENDON V. HOLDER

statutory alternative was simply a means of committing the offense, then we are not permitted to further use those documents to determine whether the defendant in fact committed an offense falling within the federal definition.

In this case, therefore, the panel should have initially looked to the Shepard documents for the sole and limited purpose of determining whether “intent to commit larceny” was an element of the offense. If the documents revealed it was not an element, the panel could conclude—as it did—that California Penal Code section 459 is categorically broader than an attempted theft offense under 8 U.S.C. § 1101(a)(43)(U), and that petitioner’s conviction under the state statute did not render him ineligible for cancellation. That is so even if the Shepard documents also revealed that petitioner, as a factual matter, had an intent to commit larceny and had therefore attempted to commit a “theft offense” within the meaning of the federal statute. In short, a court should first look to the Shepard documents to determine a crime’s elements, and then, on that basis, decide whether the categorical or modified categorical approach is appropriate.

Such a methodology is the best, and perhaps only, way to reconcile the text of Descamps’s footnote 2 with the rest of the opinion. Given Descamps’s singular focus on conducting “an elements-based inquiry” rather than an “evidence-based one,” Descamps, 133 S. Ct. at 2287, it would make little sense for us to treat elements and means interchangeably. On the other hand, I cannot endorse the panel’s implicit conclusion that footnote 2 is just a cluster of errant sentences, irreconcilable with the rest of the opinion and free for us to disregard. Descamps makes clear that “a court need not parse state law” to determine a crime’s elements, and the panel’s

RENDON V. HOLDER 19

holding is in open conflict with that pellucid instruction. Id. at 2285 n.2.

I admit that examining the Shepard documents to determine an offense’s elements comes with its own set of problems. The documents may not clearly reveal which statutory terms are, in fact, elements. And, even if they do, reliance on such documents will inevitably mean that convictions under the same state statute will sometimes be treated differently for federal law purposes. But there’s no perfect way of handling disjunctive statutes, and the approach footnote 2 adopts seems to me as good as any. It limits judicial fact-finding to the narrow and inescapable task of determining what an offense’s elements are. And it doesn’t require us to engage in the laborious and often inscrutable exercise of parsing state law.

More importantly, whether or not one agrees with Descamps’s methodology, we are obligated to implement it. Given the degree of conflict and confusion amongst the circuits, the Supreme Court will surely revisit this issue sooner rather than later. And, when it does so, I believe it will tell us in no uncertain terms that the panel’s opinion here has led us badly astray.