Nelida Cabeda v. Attorney General United States

971 F.3d 165

• Court: Court of Appeals for the Third Circuit
• Decided: August 18, 2020
• Docket: 19-1835
• Status: Published

Opinion

PRECEDENTIAL UNITED STATES COURT OF APPEALS FOR THE THIRD CIRCUIT _____________

No. 19-1835 _____________

NELIDA BEATRIZ CABEDA, Petitioner

v.

ATTORNEY GENERAL OF THE UNITED STATES OF AMERICA _______________

On Petition for Review of an Order of the United States Department of Justice Board of Immigration Appeals (BIA A042-791-483) Immigration Judge: John P. Ellington _______________

Argued January 14, 2020

Before: JORDAN, GREENAWAY, JR., and KRAUSE, Circuit Judges

(Filed: August 18, 2020) _______________ Thomas M. Griffin [ARGUED] Surin & Griffin 718 Arch Street, Suite 701N Philadelphia, PA 19106 Counsel for Petitioner

William P. Barr Andrew J. Oliveira [ARGUED] Gregory A. Pennington, Jr. United States Department of Justice Office of Immigration Litigation P.O. Box 878 Ben Franklin Station Washington, DC 20044 Counsel for Respondent _______________

OPINION OF THE COURT _______________

JORDAN, Circuit Judge.

We deal today with another appearance of what is known as the “categorical approach” to determining whether and how a conviction under state law will have consequences for the convicted criminal under federal law. We must apply it now in an immigration case, but, in whatever context it surfaces, it’s a fair bet that this formalistic framework may result in some counterintuitive and hard-to-justify outcome. And so it does here.

Argentine citizen Nelida Beatriz Cabeda, a woman in her thirties, was convicted in Pennsylvania state court of having involuntary deviate sexual intercourse with a teenage boy. That conviction ultimately led immigration authorities

2 to find her removable for having committed what they concluded is a state-law offense qualifying as an “aggravated felony,” 8 U.S.C. § 1227(a)(2)(A)(iii), specifically the “sexual abuse of a minor,” id. § 1101(a)(43)(A). Cabeda has petitioned for review of that decision, arguing that, notwithstanding her actual, admitted sexual abuse of a minor, she cannot be removed on that basis. That is so, she says, because the Pennsylvania statute under which she was convicted could conceivably be violated by conduct that falls short of satisfying all the elements of the federally defined crime of sexual abuse of a minor.

Regrettably, she is right. The categorical approach mandates our accedence to Cabeda’s demand that we ignore what she actually did and focus instead on what someone else, in a hypothetical world, could have done. That’s the analytical box the categorical approach puts us in. Thus, even though it is indisputable on this record – and, in fact, no one does dispute – that Cabeda repeatedly had sex with a minor, when we assess her conviction alongside the pertinent federal statutes, the categorical approach blinds us to the facts and compels us to hold that the crime of which she was convicted does not amount to the aggravated felony of “sexual abuse of a minor.” It is a surpassingly strange result but required by controlling law.

I. BACKGROUND

Cabeda is a citizen of Argentina who entered the United States in 1991 as a lawful permanent resident. Many years later, as alluded to above, she repeatedly engaged in vaginal and oral sex with a 15-year-old boy. She was 34 years old at the time and well aware of the boy’s age. The

3 encounters occurred over a period of several weeks and eventually came to light after the child’s mother found text messages of a sexual nature on his phone and took her concerns to the police.

Once confronted, Cabeda confessed and pled guilty in Pennsylvania state court to one count of Involuntary Deviate Sexual Intercourse, in violation of 18 Pa. Cons. Stat. § 3123(a)(7). She was sentenced to four to eight years’ imprisonment. Her crime drew the attention of federal authorities, and, after serving the minimum required term of her sentence, Cabeda was released into the custody of Immigration and Customs Enforcement agents in July of 2018. She was served with a Notice to Appear charging her with removability on two grounds: first, that she had committed the aggravated felony of sexual abuse of a minor; and second, that she had committed the crime of child abuse. Cabeda appeared before an Immigration Judge (“IJ”) and denied the charges.

After the hearing, she filed a motion to terminate the immigration proceedings, arguing that her Pennsylvania conviction did not qualify as either sexual abuse of a minor or child abuse. The IJ denied the motion. He concluded that Cabeda’s statute of conviction categorically qualified as the aggravated felony of “sexual abuse of a minor” for purposes of federal immigration law. The analysis proceeded in two steps. First, the IJ looked to the definition of “sexual abuse” set forth in 18 U.S.C. § 3509(a)(8), an approach previously endorsed by the Board of Immigration Appeals (“BIA”) in In re Rodriguez-Rodriguez, 22 I. & N. Dec. 991 (BIA 1999). The IJ next referred to the Supreme Court’s observation in Esquivel-Quintana v. Sessions, 137 S. Ct. 1562 (2017), that,

4 “in the context of statutory rape offenses that criminalize sexual intercourse based solely on the age of the participants, the generic federal definition of sexual abuse of a minor requires that the victim be younger than 16.” Id. at 1568. Thus, since Cabeda’s crime of conviction likewise requires that the victim be someone “less than 16 years of age,” 18 Pa. Cons. Stat. § 3123(a)(7), the IJ ruled that Cabeda’s crime qualified as sexual abuse of a minor. And, since sexual abuse of a minor is one variety of child abuse, the IJ also ruled that Cabeda had committed the crime of child abuse and was therefore removable on that ground too.

Cabeda appealed the IJ’s decision to the BIA. She argued that the IJ had erred in two ways – first, by failing to apply what she claims is a new generic definition of sexual abuse of a minor prescribed by Esquivel-Quintana, and second, by concluding that her crime of conviction was a categorical match for the federal generic crime of sexual abuse of a minor.1 The BIA agreed that the IJ had erred by

1 Cabeda additionally asserted before the BIA that her crime of conviction did not qualify as a “crime of child abuse,” 8 U.S.C. § 1227(a)(2)(E)(i). The BIA rejected her argument and affirmed the IJ on that count as well. She has not raised that argument in her petition to us, perhaps because a conviction for sexual abuse of a minor is an aggravated felony and prevents her from applying for cancellation of removal, whereas a conviction for child abuse, which is not an aggravated felony, leaves open the possibility of that relief. See Salmoran v. Att’y Gen., 909 F.3d 73, 83 (3d Cir. 2018) (stating that a removable alien may apply for cancellation of removal if crime of conviction is not an aggravated felony). The crime of child abuse is not listed in the Immigration and

5 failing to use Esquivel-Quintana’s “new” definition of sexual abuse of a minor. But it concluded that the Pennsylvania statutes at play in the comparison of state and federal law are nevertheless a categorical match to that new definition, even though the BIA never specified what it believed that new definition to be. It accordingly upheld the IJ’s ruling that Cabeda was removable as an aggravated felon.

Cabeda now petitions for review.

II. DISCUSSION2

The Immigration and Nationality Act designates certain crimes as “aggravated felon[ies.]” 8 U.S.C.

Nationality Act section laying out the exclusive list of aggravated felonies; it is instead found in a separate section of that statute listing domestic violence grounds for removability. See 8 U.S.C. § 1101(a)(43) (providing list of aggravated felonies); id. § 1227(a)(2)(A)(iii), (E)(i) (identifying crime of child abuse). In any event, Cabeda’s child abuse conviction remains as an independent ground for removal.

2 The BIA had jurisdiction pursuant to 8 C.F.R. § 1003.1(b). We have jurisdiction under 8 U.S.C. § 1252(a). We review the BIA’s legal conclusions de novo. Partyka v. Att’y Gen., 417 F.3d 408, 411 (3d Cir. 2005). “[F]indings of fact are conclusive unless any reasonable adjudicator would be compelled to conclude to the contrary.” 8 U.S.C. § 1252(b)(4)(B).

6 § 1101(a)(43). If an alien commits such a crime while present in the United States, he or she becomes subject to removal from this country. Id. § 1227(a)(2)(A)(iii). The IJ and BIA both determined that Cabeda had committed the aggravated felony of sexual abuse of a minor – an understandable conclusion, given her admitted and repeated sexual abuse of a minor. But Cabeda’s conduct is irrelevant; it’s her conviction that counts. Supreme Court precedent tells us we are not to “look … to the facts of the particular … case, but instead to whether the state statute defining the crime of conviction categorically fits within the generic federal definition of a corresponding aggravated felony.” Moncrieffe v. Holder, 569 U.S. 184, 190 (2013) (internal quotation marks and citation omitted).

In this case that means we are required to decide whether the pertinent Pennsylvania statutes defining and criminalizing involuntary deviate sexual intercourse are a categorical match to the federal generic crime of sexual abuse of a minor. The offenses proscribed by the state statutes “must be viewed in the abstract,” to see whether they “share[] the nature of the federal offense that serves as a point of comparison.” Id. Thus, “a state offense is a categorical match with a generic federal offense only if a conviction of the state offense necessarily involved facts equating to the generic federal offense.” Id. (internal quotation marks, alterations, omission, and citation omitted).

A. The Generic Federal Offense of Sexual Abuse of a Minor

The term “sexual abuse of a minor,” 8 U.S.C. § 1101(a)(43)(A), is not defined in the Immigration and

7 Nationality Act (“INA”). See Restrepo v. Att’y Gen., 617 F.3d 787, 792 (3d Cir. 2010) (noting that the INA “contains no definition of this phrase”). The INA does define other aggravated felonies by expressly cross-referencing various criminal statutes. See, e.g., 8 U.S.C. § 1101(a)(43)(B) (stating that 21 U.S.C.§ 802 provides the definition of “illicit trafficking in a controlled substance”); id. § 1101(a)(43)(C) (stating that 18 U.S.C. § 921 provides the definition of “illicit trafficking in firearms or destructive devices”); id. § 1101(a)(43)(D) (making “laundering of monetary instruments[,]” at least past a monetary threshold and as defined in 18 U.S.C. § 1956, an aggravated felony). But no such cross-reference is provided to give precise content to the term “sexual abuse of a minor.” That, of course, leads to some uncertainty in discerning the elements of that generic federal crime.

The BIA dealt with this uncertainty in its en banc decision in In re Rodriguez-Rodriguez, 22 I. & N. Dec. 991 (BIA 1999), by turning to 18 U.S.C. § 3509(a)(8), “a code section relating to the rights of child victims and witnesses in federal criminal cases.” Restrepo, 617 F.3d at 796. Section 3509(a)(8) defines sexual abuse to include “the employment, use, persuasion, inducement, enticement, or coercion of a child to engage in, or assist another person to engage in, sexually explicit conduct or the rape, molestation, prostitution, or other form of sexual exploitation of children, or incest with children.” 18 U.S.C. § 3509(a)(8). Later, applying the interpretive approach set forth in the Supreme Court’s landmark decision in Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837 (1984), we deferred to and adopted the BIA’s analysis in our opinion in Restrepo v. Attorney General. We reasoned that the absence

8 of a definition in the statute indicated that the meaning of “sexual abuse of a minor” is not clear and unambiguous. Restrepo, 617 F.3d at 795-97. We noted “that the BIA’s definition of sexual abuse of a minor [in Rodriguez- Rodriguez] is a reasonable one and that it [was therefore] appropriate to exercise Chevron deference.” Id. at 796. Accordingly, in our Circuit and for purposes of applying the categorical approach in the context of an immigration case, an analysis of the generic crime of “sexual abuse of a minor” depends upon 18 U.S.C. § 3509(a)(8) for guidance. Despite that, at Cabeda’s urging, the BIA ruled that the Supreme Court decision in Esquivel-Quintana effectively overruled Restrepo and altered that prior approach. That was error.

In Esquivel-Quintana, the Supreme Court considered the immigration ramifications of a conviction under a California law defining statutory rape to include “sexual intercourse with a minor who is more than three years younger than the perpetrator,” 137 S. Ct. at 1567 (quoting Cal. Penal Code § 261.5(c)). More specifically, the question was “whether a conviction under a state statute [thus] criminalizing consensual sexual intercourse between a 21- year-old and a 17-year-old qualifies as sexual abuse of a minor under the INA.” Id. The Court began by reiterating the applicability of the categorical approach, saying that the pertinent section of the INA “makes aliens removable based on the nature of their convictions, not based on their actual conduct.” Id. (citing 8 U.S.C. § 1227(a)(2)(A)(iii) (“Any alien who is convicted of an aggravated felony at any time after admission is deportable.”)). The Court then focused on the precise question before it.

9 Noting again the parameters of the California statutory rape statute and California’s definition of a “minor” as anyone under 18, the Court observed that “the conduct criminalized under this provision would be, at a minimum, consensual sexual intercourse between a victim who is almost 18 and a perpetrator who just turned 21.” Id. at 1568. The categorical analysis required the presumption that the behavior at issue was the least culpable that could be prosecuted under the statute, so consensual sex between one partner a day shy of eighteen and another on his or her 21st birthday was the frame of reference the Court assumed for judging whether the offense of conviction constituted sexual abuse of a minor under the INA.

Because the INA does not provide a definition of “sexual abuse of a minor,” the Court turned to “the normal tools of statutory interpretation[,]” beginning with the words of the statute itself. Id. at 1569. Since Congress added sexual abuse of a minor to the INA in 1996 as an aggravated felony triggering removal, the Supreme Court looked to a then- current dictionary for a definition of what constitutes “sexual abuse,” saying the ordinary meaning of those words “included ‘the engaging in sexual contact with a person who is below a specified age or who is incapable of giving consent because of age or mental or physical incapacity.’” Id. (quoting Merriam-Webster’s Dictionary of Law 454 (1996)). Because the INA requires that the abuse be “of a minor,” the Court decided the statutory focus is “on age, rather than mental or physical incapacity.” Id.

Turning to the parties’ arguments, the Court said that the government wanted a federal law definition of “sexual abuse of a minor” requiring only that behavior “(1) is illegal,

10 (2) involves sexual activity, and (3) is directed at a person younger than 18 years old.” Id. (citation omitted). But, the Court observed, that “turns the categorical approach on its head by defining the generic federal offense of sexual abuse of a minor as whatever is illegal under the particular law of the State where the defendant was convicted.” Id. at 1570. That effectively means “there is no ‘generic’ definition at all.” Id.

Instead of accepting the government’s unrestricted whatever-a-state-outlaws approach to defining the term, the Supreme Court agreed with the petitioner’s much narrower focus on the age of consent that is customary in many statutory rape laws across the country. It was noteworthy, the Court decided, that the word “aggravated” precedes the word “felony” in the INA’s listing of “sexual abuse of a minor” as a deportable crime, and that the crime is listed in the same subparagraph of the statute as murder and rape, two of the most heinous crimes. “The structure of the INA therefore suggests that sexual abuse of a minor encompasses only especially egregious felonies.” Id. Ultimately, after looking at other contextual clues in the INA and surveying several states’ policy choices on the age of consent for sexual activity, the Court said, “the general consensus from state criminal codes points to the same generic definition as dictionaries and federal law: Where sexual intercourse is abusive solely because of the ages of the participants, the victim must be younger than 16.” Id. at 1572.

That ended the matter. The Court did not have to give a full definition of what constitutes sexual abuse of a minor, and it did not do so. In fact, it expressly resisted attempts to push it past the limits of what was required to answer the

11 specific question before it. There was no need to consider Chevron deference or any other issue that might bear more broadly on the interpretation of the term “sexual abuse of a minor,” “because the statute, read in context, unambiguously foreclose[d] the [BIA]’s interpretation[,]” which had treated the petitioner’s crime as an aggravated felony. Id. All that mattered was that the “petitioner was not convicted of an aggravated felony and [was] not, on that basis, removable.” Id. at 1568.

While there is a lot to learn from the Supreme Court’s Esquivel-Quintana decision, the primary take-away for the present matter is that the Court very deliberately ruled narrowly. It did not purport to establish a full definition of “sexual abuse of a minor,” and it did not, in either purpose or effect, undermine our existing precedent in Restrepo in a way that would permit us to ignore that precedent. One may sense some tension between the statutory interpretation undertaken in Restrepo and the analytical approach employed in Esquivel-Quintana. Most notably, our opinion in Restrepo was premised heavily on the broad applicability of Chevron deference to the BIA’s expertise, see Restrepo, 617 F.3d at 793 (asserting that “Congress may have intended for the BIA to utilize its expertise to define the phrase” sexual abuse of a minor), while Esquivel-Quintana relied on ordinary tools of statutory interpretation and declined to resort to Chevron deference in answering the specific question confronted, see Esquivel-Quintana, 137 S. Ct. at 1572 (stating that “the statute, read in context, unambiguously forecloses the [BIA]’s interpretation” and thus Chevron did not apply). But the implication that Chevron deference is unnecessary in one specific instance does not “sufficiently undercut the decisional basis” of Restrepo to allow us to say that its “rule

12 no longer has any vitality[.]” West v. Keve, 721 F.2d 91, 93 (3d Cir. 1983) (citation omitted).3 In short, we are still bound by Restrepo. Esquivel-Quintana has not changed that.4 3 That is true even though the Supreme Court’s decision in another recent case, Kisor v. Wilkie, 139 S. Ct. 2400 (2019), casts doubt on the kind of broadly deferential approach taken in Restrepo. See id. at 2415 (considering the question of what judicial deference is owed to an agency’s interpretation of its own regulations – so-called Auer deference – and deciding that “Auer … gives agencies their due, while also allowing – indeed, obligating – courts to perform their reviewing and restraining functions”). 4 In the accompanying opinion concurring in part, our colleague Judge Krause says that the tension between Esquivel-Quintana and Restrepo is “irreconcilable[.]” (Concur. Op. at 4.) That is because, in her view, “[i]n multiple respects, the statutory analysis of Esquivel-Quintana is entirely inconsistent with that of Restrepo.” (Concur. Op. at 4.) She thus concludes that “Esquivel-Quintana has so undermined [Restrepo’s] analysis that … Restrepo is no longer good law.” (Concur. Op. at 4-5.) That is where we part company. As already acknowledged, it is true that Esquivel- Quintana and Restrepo undertook the task of statutory interpretation on different methodological tracks. Our decision in Restrepo was written at a time when deference to agency decision making often proceeded as a matter of course, while the Supreme Court’s decision in Esquivel- Quintana reflects what may be seen as a more searching and nuanced approach. But shifting interpretive methodologies are not usually viewed as carrying the force of stare decisis, at least not when the decisions employing them do not purport

13 to overrule past precedent. We have noted that the Supreme Court “typically avoids methodological stare decisis[,]” Am. Farm Bureau Fed’n v. U.S. E.P.A., 792 F.3d 281, 307 n.8 (3d Cir. 2015), while observing that “federal courts do not treat interpretive methodology as a traditional form of ‘law[.]’” Id. (quoting Evan J. Criddle & Glen Staszewski, Against Methodological Stare Decisis, 102 Geo. L.J. 1573, 1576 (2014)). See also Philip P. Frickey, Interpretive-Regime Change, 38 Loy. L.A. L. Rev. 1971, 1976 (2005) (noting that the Supreme Court’s methodological statements “are not binding on the Supreme Court or even on lower courts”). We certainly agree that cases like Esquivel-Quintana and Kisor provide an analytical approach we ought to follow now, but that does not mean the substantive conclusions reached in earlier cases have all been overruled. If Esquivel-Quintana did what our colleague claims for it – that is, if it meant that our prior precedential decisions were all being overruled to the extent they gave broad Chevron deference to the BIA’s interpretation of immigration statutes – we think there would have been more to signal so dramatic a step than the mere observation that, in the particular case then before it, the Supreme Court saw Chevron as having no application. See Esquivel-Quintana v. Sessions, 137 S. Ct. at 1572 (“We have no need to resolve whether the rule of lenity or Chevron receives priority in this case because the statute, read in context, unambiguously forecloses the Board's interpretation. Therefore, neither the rule of lenity nor Chevron applies.”). So we are not persuaded that Restrepo has been overruled by Esquivel-Quintana. That is not “turn[ing] vertical stare decisis on its head,” as our colleague says.

14 Indeed, we have already taken the position, at least implicitly, that Esquivel-Quintana speaks to the question of statutory rape, not more broadly to the definition of the generic offense of sexual abuse of a minor.5 See Mondragon- Gonzalez v. Attorney General, 884 F.3d 155, 160 (3d Cir. 2018) (noting that Esquivel-Quintana clarified the meaning of “the generic offense of statutory rape”).6 And other circuits

(Concur. Op. at 19.) It is giving necessary respect to our existing precedent, even when we ourselves might be inclined to decide things differently now. 5 In fact, on the topic of statutory rape, the opinion has the even narrower focus of the age of consent. See Esquivel- Quintana, 137 S. Ct. at 1572-73 (“We hold that in the context of statutory rape offenses focused solely on the age of the participants, the generic federal definition of ‘sexual abuse of a minor’ under § 1101(a)(43)(A) requires the age of the victim to be less than 16.”). 6 It is notable that nowhere in Mondragon-Gonzalez did we suggest, much less embrace, the view that Esquivel- Quintana enacted a sweeping change that will affect all matters in which we and the BIA have been called upon to interpret a statute. In fact, even though Mondragon-Gonzalez presented the question of whether to defer to the BIA’s definition of the crime of child abuse, we simply stated that “Esquivel-Quintana ha[d] no application … at all.” Mondragon-Gonzalez, 884 F.3d at 160. That pronouncement appears incompatible with the kind of sea change in our jurisprudence that our concurring colleague says has occurred.

15 apparently agree. See Acevedo v. Barr, 943 F.3d 619, 623 (2d Cir. 2019) (holding that Esquivel-Quintana “did not foreclose the BIA’s use of [§ 3509(a)(8)] in other instances” and thus prior circuit decisions to grant deference to that approach were still binding); Correa-Diaz v. Sessions, 881 F.3d 523, 527 (7th Cir. 2018) (noting that the Supreme Court “declined to rule more broadly on the generic federal definition” and decided only “one precise question: the definition of ‘minor’ under § 1101(a)(43)(A) in the context of statutory rape offenses focused solely on the age of the participants”). Pursuant to Restrepo, then, we will continue to defer to the BIA’s use of 18 U.S.C. § 3509(a)(8) as the primary guide to defining the generic federal crime of sexual abuse of a minor.

That, however, does not end our task. Section 3509(a)(8) does not specify a mens rea requirement, and we cannot defer to a nullity, so we must look elsewhere to discern the mens rea required to establish the generic federal crime. Following the analytical pattern laid out in Esquivel- Quintana, we thus “consider the structure of the INA, the inherent egregious nature of an aggravated felony, and closely-related statutes.” Acevedo, 943 F.3d at 624. Those sign posts all lead us to the conclusion that sexual abuse of a minor is a crime that requires, at a minimum, a knowing state of mind.

First, “[t]he structure of the INA … suggests that sexual abuse of a minor encompasses only especially egregious felonies.” Esquivel-Quintana, 137 S. Ct. at 1570. Again, “the INA lists sexual abuse of a minor in the same subparagraph as ‘murder’ and ‘rape,’” and the “[s]urrounding

16 provisions” of the law, listing extremely serious offenses, “guide our interpretation[.]” Id. “When considering the mens rea required for a crime to serve as ‘sexual abuse of a minor,’ the court must keep in mind this categorization.” Acevedo, 943 F.3d at 624. The Supreme Court’s suggestion that sexual abuse of a minor is an “especially egregious felon[y]” indicates that a mens rea of lower culpability such as recklessness will not suffice as an element of the generic crime.

Second, the term “aggravated felony” itself implies a certain “inherent seriousness[.]” Id. After all, a conviction for such an offense “carries significant immigration consequences, including providing a basis for the removal from the United States of a lawfully present immigrant, or, as in this case, disqualifying a removable immigrant alien from discretionary relief from removal.” Rangel-Perez v. Lynch, 816 F.3d 591, 601-02 (10th Cir. 2016) (citation omitted). An aggravated felony conviction, then, “must be for conduct that Congress has determined warrants such significant and serious treatment.” Id. at 602. And that too signals that a knowing mens rea is a requirement.

Third and finally, our consideration of a closely related statute, 18 U.S.C. § 2243, confirms that, to prove sexual abuse of a minor, the prosecution must show that the perpetrator acted knowingly. Section 2243, titled “[s]exual abuse of a minor or ward[,]” is a helpful analog to section 3509(a)(8). The Supreme Court reached for it when clarifying a different part of the definition of sexual abuse of a minor. See Esquivel-Quintana, 137 S. Ct. at 1570-71 (turning to 18 U.S.C. § 2243 to determine the age of consent for the generic federal definition of sexual abuse of a minor).

17 Other courts, too, have looked to it when defining the mens rea for sexual abuse of a minor. See, e.g., Acevedo, 943 F.3d at 624 (importing mens rea from 18 U.S.C § 2243); Rangel- Perez, 816 F.3d at 604 (same). Section 2243 actually incorporates two distinct mens rea requirements. First, it requires knowing conduct as to the sexual act in question. 18 U.S.C. § 2243(a). 7 Second, it establishes that no knowledge at all is required with respect to the victim’s age; in that respect, it is a strict liability statute. Id. § 2243(d)(1). We hold that the generic federal crime of sexual abuse of a minor includes at least the first of those mens rea requirements; we have no cause today to address whether it also includes the second.8

7 Our concurring colleague claims that turning to § 2243 for guidance while declining to overrule Restrepo lacks “logical coherence.” (Concur. Op. at 31.) But Restrepo itself acknowledged that Rodriguez-Rodriguez only treated section 3509(a)(8) as a “guide” and not “as a restrictive or limiting definition[.]” Restrepo, 617 F.3d at 796 n.10. It is thus consistent to both continue to adhere to binding precedent calling for us to defer to the BIA’s reliance on § 3509(a)(8), and, at the same time, to turn to other statutory aids when § 3509(a)(8) fails to provide the necessary guidance. 8 With respect to the mens rea regarding the age of the victim, the Fifth Circuit has held that a state statute requiring only recklessness is a categorical match for the generic federal offense of sexual abuse of a minor. United States v. Rivas, 836 F.3d 514, 515 & n.2 (5th Cir. 2016) (citing Contreras v. Holder, 754 F.3d 286, 295 (5th Cir. 2014)). As just noted, however, there is no need for us to reach that issue.

18 In summary, section 3509(a)(8) continues to provide the contours of the substance of the offense, but section 2243 sets forth the necessary mental state. With the generic crime thus sufficiently outlined, we turn next to the relevant Pennsylvania statutes to see whether, by comparison, the least culpable conduct for which a conviction could be had under them would likewise fit the definition of the generic crime.

B. Pennsylvania Involuntary Deviate Sexual Intercourse

The necessary comparison quickly shows that Cabeda is correct in asserting that there is no categorical match between the Pennsylvania statutes and the generic federal offense of sexual abuse of a minor. The critical difference is found in the mens rea requirements – the state offense can be committed recklessly, whereas the federal generic crime requires a knowing mental state with regard to the sexual conduct.

Cabeda’s offense of conviction is 18 Pa. Cons. Stat. § 3123(a)(7), under which, a “person commits a felony of the first degree when the person engages in deviate sexual intercourse with a complainant… who is less than 16 years of age and the person is four or more year older than the complainant and the complainant and person are not married to each other.” The term “deviate sexual intercourse” is in turn defined in 18 Pa. Cons. Stat. § 3101 as “[s]exual intercourse per os or per anus between human beings and any form of sexual intercourse with an animal. The term also includes penetration, however slight, of the genitals or anus of another person with a foreign object for any purpose other

19 than good faith medical, hygienic or law enforcement procedures.”

Notably absent from either of those statutory provisions is any mens rea requirement. The Pennsylvania criminal code has a gap-filling provision,18 Pa. Cons. Stat. § 302(c), for just such a circumstance: “When the culpability sufficient to establish a material element of an offense is not prescribed by law, such element is established if a person acts intentionally, knowingly or recklessly with respect thereto.” 18 Pa. Cons. Stat. § 302(c). Thus, the Pennsylvania code allows for prosecution under section 3123(a)(7) on the basis of behavior that is only reckless.9

9 Section 302 is not divisible in a way that would make the modified categorical approach appropriate. Although § 302(c) lists three types of mens rea in the disjunctive, Pennsylvania authority suggests that they are alternate means rather than elements. See Mathis v. United States, 136 S. Ct. 2243, 2253 n.3 (2016) (noting that alternate “mental states are interchangeable means of satisfying a single mens rea element”). Section 302(c) itself refers to “the culpability sufficient to establish a material element of an offense[,]” and it provides alternative mental states for determining whether “such element is established[.]” Thus, the statute itself distinguishes between the elements of an offense and the alternative means, listed therein, of satisfying those elements. Pennsylvania caselaw too appears to treat the § 302(c) mental states as alternative means of satisfying a single statutory element. See, e.g., Commonwealth v. Gallagher, 924 A.2d 636, 637 (Pa. 2007) (noting in prosecution for child luring that § 302(c) imposes a “duty to prove that [defendant] acted intentionally, knowingly, or at least recklessly”);

20 That conclusion is confirmed by Pennsylvania caselaw, which shows that other, closely related, statutes can be violated recklessly. For example, section 3125 of the same criminal code title prohibits “penetration, however slight, of the genitals or anus of a complainant with a part of the person’s body for any purpose other than good faith medical, hygienic or law enforcement procedures” where, among other possible circumstances, “the person has substantially impaired the complainant’s power to appraise or control his or her conduct by administering or employing, without the knowledge of the complainant, drugs, intoxicants or other means for the purpose of preventing resistance[.]” 18 Pa. Cons. Stat. § 3125(a). The prohibition set forth in that date- rape statute contains nearly identical wording to the statute Cabeda violated, and Pennsylvania courts have said that “the minimum mens rea for these offenses is recklessness.” Commonwealth v. Cosby, 224 A.3d 372, 419 (Pa. Super. Ct. 2019); accord Commonwealth v. Torsilieri, No. 2300 EDA 2018, 2019 WL 3854450, at *5 (Pa. Super. Ct. Aug. 16,

Commonwealth v. Mayfield, 832 A.2d 418, 427 (Pa. 2003) (noting in prosecution for sexual assault that § 302(c) “require[s] the Commonwealth to prove at least recklessness”). We have also suggested as much. See Aguilar v. Att’y Gen., 663 F.3d 692, 695 & n.7 (3d Cir. 2011) (holding that sexual assault in violation of 18 Pa. Cons. Stat. § 3124.1 constitutes an INA “crime of violence” aggravated felony even though it can be committed recklessly under § 302(c), and noting that “the trial judge instructed the jury that they must find ‘that the defendant acted knowingly or at least recklessly regarding [the complainant’s] nonconsent’”).

21 2019). The mens rea catchall provision in section 302(c) is the underpinning for those decisions.

Similarly, section 3126(a) prohibits “indecent contact[,]” with indecent contact defined in section 3101 as “[a]ny touching of the sexual or other intimate parts of the person for the purpose of arousing or gratifying sexual desire, in any person.” Again, Pennsylvania courts have concluded that the crime thus defined can be committed recklessly, based on section 302(c). See Torsilieri, 2019 WL 3854450, at *5 (explaining that the default mens rea of recklessness applies); Commonwealth v. Carter, 418 A.2d 537, 540-41 (Pa. Super. Ct. 1980) (same).

That means that the Pennsylvania statutes are categorically broader than the federal generic crime of sexual abuse of a minor, since the federal offense must be committed knowingly, but the Pennsylvania crimes can be committed recklessly. Now, one might be forgiven for thinking that, as a matter of common sense, it is scarcely conceivable that one could, as a factual matter, recklessly commit the crime that Pennsylvania calls involuntary deviate sexual intercourse.10 That improbability, one might further think, should mean that the Pennsylvania statute actually is a categorical match for the generic crime of sexual abuse of a minor, because there is no realistic probability that Pennsylvania could or would enforce its statute in a way that would sweep in reckless conduct. Following that reasoning would allow for a more sensible

10 And yet, given the breadth of the statutory language prohibiting penetration “however slight,” reckless violation of the law is not as absurd as it might seem at first glance. See infra at n.12.

22 result here, the semantic strictures of the categorical approach notwithstanding. Unfortunately, that analytical route is also barred by binding precedent.

It is true the Supreme Court has stated that, at least under certain circumstances, the categorical approach “requires a realistic probability, not a theoretical possibility, 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). And it is further true that, in other circuits, Cabeda’s arguments may well have failed because of the improbability of applying a statute like Pennsylvania’s to prosecute reckless conduct. See United States v. Castillo-Rivera, 853 F.3d 218, 223 (5th Cir. 2017) (en banc) (“There is no exception to the actual case requirement articulated in Duenas-Alvarez where a court concludes a state statute is broader on its face.”).

“Our Court’s precedent, however, takes [an] alternative approach.” Salmoran v. Att’y Gen., 909 F.3d 73, 81 (3d Cir. 2018). We have held that “where the elements of the crime of conviction are not the same as the elements of the generic federal offense … the realistic probability inquiry … is simply not meant to apply.” Id. (internal quotation marks and citations omitted). So, once we conclude that the textual breadth of a statute is more expansive than the federal generic crime because the mens rea elements are different, a petitioner need not show that there is a realistic chance that the statute will actually be applied in an overly broad manner. See Zhi Fei Liao v. Att'y Gen., 910 F.3d 714, 723 (3d Cir. 2018) (noting that “it is unnecessary to conduct a realistic probability inquiry” when “the elements of [the state] conviction … [do] not match the elements of the generic

23 federal offense”); Singh v. Att’y Gen., 839 F.3d 273, 286 n.10 (3d Cir. 2016) (noting that the realistic probability inquiry does not apply when “the elements of the crime of conviction are not the same as the elements of the generic federal offense”).

Thus, the mismatch between the mens rea of the federal generic crime and the Pennsylvania involuntary deviate sexual intercourse statute leads inevitably to the conclusion that they are not a categorical match. We are left with no option, then, but to conclude that Cabeda’s multiple statutory rapes of a 15-year-old boy do not qualify as sexual abuse of a minor within the meaning of the INA. What a world.

C. The Dissent’s Analytical Path Is Unavailable

The dissent, understandably, seeks to avoid this result, but we cannot endorse the legal reasoning it uses along the way. Our dissenting colleague seeks to retroactively reframe the Salmoran line of cases as applying only when there is a “clear difference between the statute of conviction and the federal generic offense[.]” (Dissenting Op. at 3.) Thus, in his view, the realistic probability inquiry is foreclosed only when the petitioner “definitively demonstrate[s] a difference” between them. (Dissenting Op. at 5.) In close cases, then, and only in close cases, will the inquiry apply. But nowhere in Salmoran is there any indication that we were laying down a mere tie-breaking rule. Instead, as already noted, what we said was that “where the elements of the crime of conviction are not the same as the elements of the generic federal offense … the realistic probability inquiry … is simply not meant to apply.” Salmoran, 909 F.3d at 81; see also Zhi Fei Liao, 910

24 F.3d at 723 (holding that “the BIA erred in conducting a ‘realistic probability’ inquiry where the elements of petitioner’s controlled substance conviction under Pennsylvania state law did not match the elements of the generic federal offense”). In line with that holding, we engage in an ordinary statutory interpretation analysis of the statute of conviction. And once we complete that analysis, we compare our result to the federal generic offense. We do not then take into account how difficult the statutory interpretation question was in determining whether the realistic-probability inquiry applies.11 Tempting as it is, then, we cannot accept our dissenting colleague’s reading of Salmoran.

But even if the dissent were correct that Salmoran requires a “clear difference” between the elements of the state statute and the federal generic offense, such a difference is present here. Our dissenting colleague believes that Pennsylvania’s Involuntary Deviate Sexual Intercourse statute requires “a mens rea of knowledge, if not intent.” (Dissenting Op. at 6.) He bases that conclusion on two premises: first, that a mens rea of recklessness would produce absurd results which the legislature could not possibly have intended; and second, that it is a mistake to rely on the default culpability requirements of § 302(c).

11 Nor are we persuaded by the dissent’s approach of first looking for prosecutions and then concluding, after finding none, that the realistic-probability inquiry does not apply. That is precisely backwards. We only look through judicial records for prosecutions after we determine whether the realistic-probability analysis applies.

25 As to the first premise, it is putting the cart before the horse to start with the absurdity doctrine and then work backwards from there to interpret the text so as to avoid a preconceived absurd result. Instead, we first interpret the statute according to its text and only then analyze whether that text leads to an absurd result. To do otherwise leads to a distortion in statutory interpretation as we strain to avoid the pre-identified absurd result. That risk is amply demonstrated by the free-form and purposive approach the dissent takes to arrive at a preferred statutory interpretation. Regardless of any intuitions we may have about whether Pennsylvania’s involuntary deviate sexual intercourse statute ought to be viewed as a sufficient match with the provisions of federal law defining the generic offense, our discomfort with the outcome in this case doesn’t allow us to rewrite our own precedent or Pennsylvania law.12

12 The dissent’s purposive analysis fails even on its own terms. The dissent claims that the Pennsylvania Supreme Court has articulated one possible purpose of § 3123(a)(7) as “to protect minors younger than 16 years of age from older teenage and adult sexual aggressors.” Commonwealth v. Albert, 758 A.2d 1149, 1154 (Pa. 2000). But the language relied up on by our dissenting colleague is from the court’s analysis of hypothetical reasons offered by the government in response to a rational-basis challenge to the statute. The court’s recognition of a legitimate state interest sufficient to withstand rational-basis scrutiny says nothing about what mens rea would suffice to accomplish the statute’s purpose. In fact, the court specifically noted that there was no statutory purpose defined anywhere. See id. at 1152 (“At the outset, we note that neither the legislative provision[] at issue here nor the accompanying legislative

26 As to the second premise – that § 302(c) is inapplicable in this context – our dissenting colleague cites no authority for that proposition, save a couple of opinions declining to apply it in an unrelated context.13 And, as

history … disclose any official statement by the legislature regarding the rationale or policy motivating [its] enactment.”). Moreover, it is not for us to make Pennsylvania’s policy choice in setting the level of culpable mens rea for sexual offenses. As already noted, it seems strange to imagine how the deviate sexual intercourse statute could be violated recklessly (see supra n.10 and accompanying text), but suppose an adult were to aggressively and inappropriately touch a child, and that conduct resulted in the penetration “however slight” required for a violation of §§ 302(c), 3101, and 3123(a)(7). In such a situation, penalizing reckless conduct is not on its face absurd. The dissent’s example of a parent feeding her child is plainly inapposite. Such an innocent action carries no culpable mens rea whatsoever and thus is not subject to criminal penalty.

13 Our dissenting colleague cites Commonwealth v. Hart, 28 A.3d 898 (Pa. 2011), as an example of a Pennsylvania court engaging in broader statutory interpretation “rather than merely applying § 302(c).” But at issue in Hart was an interpretive dispute over actus reus, not mens rea – specifically, over “whether the mere offer of an automobile ride to a child constitutes an attempt to ‘lure’ the child.” Id. at 900. In concluding that an offer for a ride must be accompanied by “other enticement or inducement,” id. at

27 900, 909, the court unsurprisingly saw little need to engage with section 302(c). No one argues that section 302(c) will answer every relevant question a criminal statute might raise. What matters for our purposes is whether it answers the relevant question we must answer – namely, the minimum mens rea for which a defendant could be convicted. It does. And the answer it gives is different from – and broader than – the mens rea for the federal generic offense. Nor is the dissent’s reliance on Commonwealth v. Ludwig, 874 A.2d 623 (Pa. 2005), persuasive. There, the state supreme court recognized that where a criminal statute “does not explicitly provide for an applicable mens rea, the General Assembly has provided a default culpability provision in Section 302(c) . . . that is to be applied.” Id. at 630 (emphasis added). True, in applying section 302, the court stated that the mens rea default does not apply where a contrary mens rea “is . . . prescribed by law.” Id. (quoting 18 Pa. C.S.A. § 302(c)). But it did so only because the relevant offense, third-degree murder, had a “consistent[]” and “well- settled” mens rea (malice) at common law. Id. at 630–31. There is no such common-law tradition for involuntary deviate sexual intercourse with a minor. Finally, the dissent cites to a lone concurrence by one justice that was joined by none of the other six justices on the bench. See Commonwealth v. Moran, 104 A.3d 1136, 1151- 52 (Pa. 2014) (Todd, J., concurring). The justices in the majority reiterated their “repeated[] h[o]ld[ing] [that] § 302 provides the default level of culpability where a criminal statute does not include an express mens rea.” Id. at 1149 (majority op.) (citing Gallagher, 924 A.2d at 639; Ludwig, 874 A.2d at 630; and Mayfield, 832 A.2d at 427). Even the

28 already discussed, Pennsylvania courts have applied § 302(c) to closely related statutory language.14 So while it is true that the text of Pennsylvania’s Involuntary Deviate Sexual Intercourse statute does not itself contain a mens rea, the clear guidance from the courts is that we should apply the gap- filling provisions of § 302(c).15 There is nothing unusual

two justices in dissent were in full agreement on that point. See id. at 1154 (Baer, J., concurring and dissenting). 14 See supra pp. 17-18. 15 Pennsylvania courts have repeatedly emphasized the broad applicability of § 302(c) to the Commonwealth’s criminal laws. The state supreme court has relied on it, for instance, in explaining that even for a criminal statute that lacks any mention of mens rea, the legislature’s intent as to mens rea is not unclear. See Commonwealth v. Mayfield, 832 A.2d 418, 427 (Pa. 2003) (assessing 18 Pa. C.S.A. § 3124.2 – a provision neighboring § 3123(a)(7) – that criminalizes institutional sexual assault and importing a mens rea of recklessness). Indeed, once an interpreting court determines that the relevant offense was not intended “to be a strict liability crime” the court “need . . . do[] nothing more than advert to § 302(c) and require the Commonwealth to prove at least recklessness.” Id.; see also Commonwealth v. Gallagher, 924 A.2d 636, 638 (Pa. 2007) (“As a rule, in . . . instances [where the statute does not express a mens rea requirement], Section 302(c) of the Crimes Code prescribes the default culpability requirement . . . .” (emphasis added)); Commonwealth v. Parmar, 710 A.2d 1083, 1088–89 (Pa. 1998) (“The bribery statute does not have an explicit

29 about supplementing the text of a specific provision with a cross-reference to a generally applicable statutory provision despite the unfortunate result it leads to when made a component of the categorical approach to analyzing Cabeda’s conviction.

So, while we are in complete accord with our colleague’s desire to mitigate the workings of the categorical approach, we cannot take the analytical path he suggests. In the end, we are left to shake our heads at the path we are on. But, having followed that path as required, we conclude that there is not a categorical match between Cabeda’s statute of conviction and the corresponding generic federal crime.

mens rea requirement on its face, but it is subject to the culpability requirements of Section 302 . . . . Th[ose] culpability requirements . . . apply to all crimes in the Crimes Code, like bribery, [as well as] those outside the Crimes Code . . . .”). The Pennsylvania code itself bears out § 302(c)’s broad application. As the state supreme court explained in Parmar, see 710 A.2d at 1089, although the legislature crafted exceptions to section 302(c)'s general rule, those exceptions apply only to "summary offenses" and those for which "a legislative purpose to impose absolute ability . . . plainly appears[.]" 18 Pa. C.S.A. § 305(a)(1)–(2). Neither of those exceptions, however, applies here. We are thus bound to apply § 302(c).

30 III. CONCLUSION

For the foregoing reasons, we will grant the petition for review, vacate the BIA’s decision, and remand for further proceedings consistent with this opinion.16

16 Nothing in our decision today means that Cabeda is entitled to cancellation of removal, if and when she applies for it. Our holding simply means that she is not subject to the automatic deportation that results from a conviction for an aggravated felony.

31 GREENAWAY, JR., Circuit Judge, concurring in part, dissenting in part, and dissenting in the judgment.

At the outset, I join the majority’s well-reasoned analysis with respect to the three conclusions in Section II.A: (1) Esquivel-Quintana did not provide a new federal generic definition of a crime of sexual abuse of a minor; (2) we continue to defer to 18 U.S.C. § 3509(a)(8) as the primary guide to defining the generic offense; and (3) the mens rea of the federal crime of sexual abuse of a minor is “knowing conduct as to the sexual act in question.”1 Maj. Op. at 18.

I also share my colleagues’ frustration with the depths of the mental gymnastics that the categorical approach manufactures and the counterintuitive results it often produces; however, here the proper application of the categorical approach does not result in such a head-scratching outcome.

Initially, where I part with my colleagues is in their reliance on Salmoran v. Attorney General to excuse Cabeda from demonstrating that there is “a realistic probability, not a theoretical possibility, that the State would apply its statute to conduct that falls outside the generic definition of a crime.” 909 F.3d 73, 77 (3d Cir. 2018) (Greenaway, J.) (internal quotations and citations omitted). Their analysis in Section II.B leads to the challenging interpretation that a defendant could be culpable for recklessly “engag[ing] in deviate sexual

1 Because I find that Esquivel-Quintana did not provide a new federal generic definition of a crime of sexual abuse of a minor, I respectfully decline to join JUDGE KRAUSE’s concurrence.

1 intercourse” with a minor, whatever that actually entails. 18 Pa. Cons. Stat. § 3123(a).

In contrast, my view is that the categorical approach does not require us to go down that rabbit hole. I therefore reach the opposite result, that Cabeda was convicted of an aggravated felony consisting of the sexual abuse of a minor. 8 U.S.C. § 1227(a)(2)(A)(iii); 8 U.S.C. § 1101(a)(43)(A). Accordingly, I would deny the petition for review for lack of jurisdiction.2

In explaining how I reach this conclusion, I will first situate Salmoran within the context of this Court’s treatment of the realistic probability requirement. Then I will offer two reasons why Cabeda is not excused from demonstrating a realistic probability that Pennsylvania courts would apply a mens rea of recklessness to the sexual act under Cabeda’s statute of conviction, 18 Pa. Cons. Stat. § 3123(a)(7). Namely, this case is distinguishable from Salmoran and other cases where we found it unnecessary to engage in a realistic probability inquiry, and that the majority’s interpretation of § 3123(a)(7) is flawed, or, at best, inconclusive and does not render the realistic probability requirement moot. Because I find that Cabeda cannot meet that additional burden (of proving a realistic probability), her crime is a match for the

2 This Court has jurisdiction to review final orders of removal pursuant to 8 U.S.C. § 1252(a), but we lack jurisdiction to review an order to remove an alien who commits an aggravated felony, 8 U.S.C. § 1252(a)(2)(C). However, this Court may retain jurisdiction to address the prerequisite of “whether an alien was convicted of a non-reviewable aggravated felony.” Stubbs v. Att’y Gen., 452 F.3d 251, 253 n.4 (3d Cir. 2006).

2 federal generic offense and aggravated felony of sexual abuse of a minor.

I. The Categorical Approach and the Realistic Probability Requirement

As the majority explains, the categorical approach requires us to determine if the material elements of the federal generic offense match those of the state statute of conviction. Maj. Op. at 6–7 (citing Moncrieffe v. Holder, 569 U.S. 184, 190 (2013)). Where the petitioner claims that an element of a statute encompasses more conduct than the federal crime, the petitioner must have some practical basis for his or her argument. Moncrieffe, 569 U.S. at 191; Gonzales v. Duenas- Alvarez, 549 U.S. 183, 193 (2007); Lewin v. Att’y Gen., 885 F.3d 165, 168 (3d Cir. 2018). In other words, the petitioner’s interpretation of the statute cannot be based on “legal imagination.” Duenas-Alvarez, 549 U.S. at 193.

However, this Court has clarified that where the petitioner has already demonstrated a clear difference between the statute of conviction and the federal generic offense, the realistic probability requirement is superfluous. Salmoran, 909 F.3d at 82. Indeed, we have found that the statute of conviction “plainly encompasses more conduct than its federal counterpart” based on a comparison of the plain texts of the statute and the federal generic crime, e.g., id.; Singh v. Att’y Gen., 839 F.3d 273, 285–86 (3d Cir. 2016); cf. Jean-Louis v. Att’y Gen., 582 F.3d 462, 481 (3d Cir. 2009), and where the state court has definitively articulated an interpretation of the statute that sweeps in more culpable conduct than that contemplated by the federal offense, e.g., Zhi Fei Liao v. Att’y Gen., 910 F.3d 714, 723–24 (3d Cir. 2018). Having supported the argument with statutory text or state-court interpretations

3 of the law in these cases, we determined it was unnecessary for the petitioner to further demonstrate a realistic probability that the state courts would interpret the statute in the petitioner’s favor; nothing was left to “legal imagination” in these cases. Duenas-Alvarez, 549 U.S. at 193.

II. The Realistic Probability Inquiry Applies

In my view, the outcome-determinative question in this case is whether Cabeda should be required to demonstrate a realistic probability that the Pennsylvania courts would require only recklessness as to the act of deviate sexual intercourse under § 3123(a)(7).3 See 18 Pa. Cons. Stat. § 3101 (defining deviate sexual intercourse as “[s]exual intercourse per os or per anus between human beings and any form of sexual intercourse with an animal. The term also includes penetration, however slight, of the genitals or anus of another person with a foreign object for any purpose other than good faith medical, hygienic or law enforcement procedures.”). For the following reasons, I conclude, unlike my colleagues in the majority, that Cabeda must make that showing.

3 The majority correctly points to two mens rea requirements in the federal generic offense of sexual abuse of minor. Maj. Op. at 18. The second, strict liability with respect to the victim’s age, is not the subject of debate here. Subsequent references to the mens rea at issue will refer only to the culpable mental state as to the conduct of deviate sexual intercourse as defined under § 3127(a)(7) and § 3101.

4 A. The Present Case is Distinguishable from Our Precedent, so Cabeda is not Excused from Demonstrating a Realistic Probability

The lack of a clear difference between the mens rea as to the conduct under § 3123(a)(7) and the federal generic crime’s mens rea of knowledge distinguishes this case from those in the Salmoran line of cases. As such, this is not a case “where the elements of the crime of conviction are not the same as the elements of the generic federal offense.” Salmoran, 909 F.3d at 81.

Cabeda’s statute of conviction does not reference any mens rea, so there is no textual hook for differentiation. Nor have I or any of my colleagues found an occasion where the Pennsylvania courts have definitively spoken to the mens rea of the sexual act under § 3123(a)(7). This case therefore falls outside of those in which we have found it unnecessary to inquire about the realistic probability that the state would interpret the law in the petitioner’s favor. Because we are left without sufficient “guidance as to how the statute applies,” Zhi Fei Liao, 910 F.3d at 723, I would require Cabeda to demonstrate that there is a realistic probability that the Pennsylvania courts would find reckless conduct culpable.

B. We Must Engage in a Realistic Probability Inquiry Because the Majority’s Statutory Interpretation Does Not Definitively Demonstrate a Difference Between the Statute of Conviction and the Federal Generic Offense

In an attempt to avoid this uncertainty, the majority presses to apply a gap-filling provision. This leads my colleagues to the uncomfortable presumption that one could be convicted for recklessly committing the act of deviate sexual

5 intercourse. Maj. Op. at 19–20. The majority then relies on this awkward reading of the statute to differentiate § 3123(a)(7) from the federal generic offense.

Unfortunately, neither the curious conclusion at which my colleagues arrive nor their path in reaching it alleviates the concern that the statutory interpretation they embrace is flawed. The approach I embrace demonstrates that Pennsylvania courts would more likely apply at least a mens rea of knowledge, if not intent. At the very least, this competing approach shows that the majority’s interpretation is far from definitive. Left without sufficient guidance to construe the statute confidently, we cannot forego the application of the realistic probability test.

1. The Majority Fails to Fully Consider Pennsylvania’s Approach to Statutory Interpretation

The majority would apply 18 Pa. Cons. Stat. § 302(c), the default culpability provision, to insert a “recklessness” mens rea as to the conduct. While the majority stops there, Pennsylvania courts have a more involved approach to statutory interpretation. As the Pennsylvania Supreme Court describes, “[t]he object of all interpretation and construction of statutes is to ascertain and effectuate the intention of the General Assembly.” Commonwealth v. Hart, 28 A.3d 898, 908 (Pa. 2011) (citing 1 Pa. Cons. Stat. § 1921(a)). Importantly, “[c]oncomitant with these considerations, the Statutory Construction Act also sets forth certain presumptions regarding the General Assembly’s enactment of statutes which are to be applied when attempting to ascertain its legislative intent.” Id. The Pennsylvania Supreme Court has noted that, “[i]n particular, when interpreting a statutory provision we must presume that the legislature[] does not

6 intend a result that is unreasonable, absurd, or impossible of execution, 1 Pa. Cons. Stat. § 1922(1).” Id. Indeed, Pennsylvania courts have refrained from a strict application of § 302(c). See, e.g., id. at 909–10 (choosing to ascertain a mens rea as to the act of “luring a child into a motor vehicle” from the language of the statute with guidance from the Statutory Construction Act, rather than merely applying § 302(c)); Commonwealth v. Ludwig, 874 A.2d 623, 630 (Pa. 2005) (declining to apply § 302(c) where “[t]he law is clear and well- settled regarding the mens rea for third degree murder”); see also Commonwealth v. Moran, 104 A.3d 1136, 1151–52 (Pa. 2014) (Todd, J., concurring) (“Neither the plain language of Section 302, nor our prior caselaw, requires the automatic application of Section 302’s default culpability requirements in every situation where a criminal statute does not expressly state a requisite level of culpability for the conduct it seeks to penalize. . . . Rather, Section 302 has a narrower application: it applies only where the ‘culpability sufficient to establish a material element of an offense is not prescribed by law.’” (quoting 18 Pa. Cons. Stat. § 302)).

After applying Pennsylvania’s approach to interpreting its own law, it is clear that a mens rea of recklessness does not comport with the legislature’s intent and would produce unreasonable results. We therefore should not apply the default culpability provision in § 302 to supply a mens rea in § 3123(a)(7). Instead, we should find that the sexual act under § 3123(a)(7) requires a mens rea of at least knowledge.

Although the Pennsylvania legislature did not enact a specific statement of legislative intent, the Pennsylvania Supreme Court has noted that § 3123(a)(7) serves a legitimate state interest in “protecting children from sexual predators” and “assuring that a significantly older individual cannot take

7 advantage of a young child's complete lack of knowledge, inexperience or poor judgment.” Commonwealth v. Albert, 758 A.2d 1149, 1153 (Pa. 2000) (“We agree with the Commonwealth that the subject legislation serves a legitimate state interest, i.e., to protect minors younger than 16 years of age from older teenage and adult sexual aggressors. Such an interest recognizes that older, more mature individuals are in a position that would allow them to take advantage of the immaturity and poor judgment of very young minors.”); see also Commonwealth v. Bruner, 527 A.2d 575, 576 (Pa. 1987) (“As to the necessity and purpose of the involuntary deviate sexual intercourse statute, the Commonwealth . . . does have a strong interest in . . . protecting minors from being sexually abused by adults.”). It is clear that a mens rea of at least knowledge, if not intent, complies with § 3123(a)(7)’s legislative purpose.4

Faced with these statements by the Pennsylvania Supreme Court, it is difficult to square the intent to punish sexual predators, sexual abusers, and older individuals who take advantage of children with a reading of the statute that would find reckless sexual acts culpable. § 3123(a)(7).5

4 The Pennsylvania Supreme Court’s definition of the sexual act under § 3123(a)(7) as “oral and anal sex” also supports a mens rea of knowledge or intent. See Commonwealth v. Kelley, 801 A.2d 551, 555 (Pa. 2002); cf. Commonwealth v. Hitchcock, 565 A.2d 1159, 1161 (Pa. 1989) (describing involuntary deviate sexual intercourse as “forcible sexual penetration of a person by another”). 5 Several other circuits have also interpreted similar state statutes to require knowledge or intent. See, e.g., Acevedo

8 Reading recklessness into the mens rea is also in tension with how Pennsylvania courts have so far interpreted the required mental state for committing deviate sexual intercourse. Indeed, to my knowledge, they have never found a defendant culpable for anything less than what appears to be knowing conduct. See, e.g., Commonwealth v. Elia, 83 A.3d 254, 266 (Pa. Super. Ct. 2013) (finding evidence that victim performing oral sex on defendant after previously engaging in other sexual acts was sufficient for conviction under § 3123(a)(7));

v. Barr, 943 F.3d 619, 626–27 (2d Cir. 2019) (affirming BIA’s decision to use § 3509 as the generic offense and finding New York conviction for attempted oral or anal sexual conduct with person under the age of fifteen constituted sexual abuse of a minor); Correa-Diaz v. Sessions, 881 F.3d 523, 529 (7th Cir. 2018) (continuing to apply § 3509 as a guide and concluding a conviction for attempted sexual misconduct with a minor constituted sexual abuse of a minor); Bedolla-Zarate v. Sessions, 892 F.3d 1137, 1141–42 (10th Cir. 2018) (applying § 3509 as a guide for concluding Wyoming conviction for sexual contact with minor constitutes sexual abuse of a minor); Contreras v. Holder, 754 F.3d 286, 295 (5th Cir. 2014) (comparing Virginia statute of conviction for “carnal knowledge of child between thirteen and fifteen years of age” to § 3509 to find conviction qualifies as sexual abuse of a minor); cf. Garcia-Urbano v. Sessions, 890 F.3d 726, 730 (8th Cir. 2018) (applying Esquivel-Quintana’s age requirement to find Minnesota conviction for sexual conduct with person under sixteen years of age qualifies as sexual abuse of a minor); Bahar v. Ashcroft, 264 F.3d 1309, 1313 (11th Cir. 2001) (relying on the BIA’s decision in Rodriguez to find conviction for taking “indecent liberties” with a child under sixteen years of age constitutes sexual abuse of a minor).

9 Commonwealth v. Mawhinney, 915 A.2d 107, 111 (Pa. Super. Ct. 2006), appeal denied 932 A.2d 1287 (Pa. 2007) (finding multiple acts of anal and oral sex sufficient evidence for § 3123(a)(7) conviction); Commonwealth v. Castelhun, 889 A.2d 1228, 1232–33 (Pa. Super. Ct. 2005) (same); see also Commonwealth v. Williams, 439 A.2d 765, 768 (Pa. Super. Ct. 1982) (describing involuntary deviate sexual intercourse as a “general intent crime[]”). The majority’s attenuated reading only drives needless uncertainty into a consistently-applied statute.6 Cf. 1 Pa. Cons. Stat. § 1922(2) (advising that courts

6 As indication that the Pennsylvania courts would apply a recklessness mens rea to § 3123(a)(7), the majority relies on three Pennsylvania cases interpreting two similar statutes in which the courts applied the default culpability provision. But these cases are plainly inapposite; not one speaks to the mens rea of the sexual act, only to the mens rea of other elements that are not present under § 3123(a)(7). Commonwealth v. Cosby, 224 A.3d 372, 419–20 (Pa. Super. Ct. 2019) (applying recklessness to whether the defendant has the “complainant’s consent,” knows that “the complainant is unaware that the penetration is occurring,” or “has substantially impaired the complainant’s power to appraise or control his or her conduct”); Commonwealth v. Torsilieri, No. 2300 EDA 2018, 2019 WL 3854450, at *5 (Pa. Super. Ct. Aug. 16, 2019) (not precedential) (same); Commonwealth v. Carter, 418 A.2d 537, 540-41 (Pa. Super. Ct. 1980) (applying § 302(c) to find there was sufficient evidence showing “a substantial risk that [the victim] was of such mental infirmity as to render her incapable of consent”).

10 should presume that the legislature “intends the entire statute to be effective and certain”).

Importantly, reading recklessness into the mens rea of the sexual act under § 3123(a)(7) plainly violates Pennsylvania’s tenet that courts must presume that the legislature “does not intend a result that is absurd, impossible of execution, or unreasonable.” 1 Pa. Cons. Stat. § 1922(1); see also United States v. Schneider, 14 F.3d 876, 880 (3d Cir. 1994) (“It is the obligation of the court to construe a statute to avoid absurd results, if alternative interpretations are available and consistent with the legislative purpose.”). Certainly, the Pennsylvania General Assembly did not intend to make a mother or father criminally liable for “penetrat[ing]” their baby’s mouth, “however slight,” with a hot spoon, while holding the baby and cooking dinner. 18 Pa. Cons. Stat. § 3101.

2. The Competing Interpretations Present the Precise Challenge in Duenas-Alvarez

Clearly, I have failed to convince my colleagues that knowledge or intent is the appropriate mens rea. So, I humbly proffer that my approach is at least as reasonable as that of the majority. If we accept there are two equally good ways to interpret the mens rea under § 3123(a)(7), then the majority’s conclusion that Pennsylvania would apply recklessness to the statute is far from definitive and cannot be relied upon to show a clear difference between the state statute and the federal generic offense.

This case therefore falls under the same circumstances as in Duenas-Alvarez. In Duenas-Alvarez, the petitioner posited an interpretation of state law, which had been neither

11 confirmed nor denied by the state courts. 549 U.S. at 190–93. Faced with two possible interpretations, the Supreme Court erred on the side of what was realistically probable. Id. at 193. So, too, must this Court.

Instead, the majority favors a theorical possibility and therefore unnecessarily risks superseding the state’s authority as the primary interpreter of its own laws. See Pinho v. Gonzales, 432 F.3d 193, 212 (3d Cir. 2005) (“[T]he authority of state courts to determine state-law questions is clear: ‘[C]omity and respect for federalism compel us to defer to decisions of state courts on issues of state law. That practice reflects our understanding that the decisions of state courts are definitive pronouncements of the will of the States as sovereigns.’” (citations omitted)). In rejecting the realistic probability inquiry, the majority regrettably ignores the Supreme Court’s admonition to avoid “legal imagin[ings].” Duenas-Alvarez, 549 U.S. at 193.

Unrestricted by Salmoran, and for the reasons supplied above, I would require Cabeda to show a realistic probability that the Pennsylvania courts would apply a mens rea of recklessness to the act of deviate sexual intercourse.7

7 The depths of the disagreements between myself and my colleagues in analyzing Pennsylvania cases is further evidence that we should tread lightly in interpreting § 3123(a)(7). So, rather than “put[] the cart before the horse,” Maj. Op. at 26, this application of Salmoran’s realistic probability requirement draws from the Supreme Court’s own cautionary approach.

12 III. Cabeda Cannot Meet the Realistic Probability Requirement

“To show that realistic probability, an offender . . . may show that the statute was so applied in [the petitioner’s] own case. . . . or other cases in which the state courts in fact did apply the statute in the special (nongeneric) manner for which [the petitioner] argues.” Duenas-Alvarez, 549 U.S. at 193. We have not found a Pennsylvania case that has applied a mens rea of recklessness to the conduct criminalized under § 3123(a)(7), so Cabeda could not have met the realistic probability requirement here.

***

I would therefore find Cabeda’s statute of conviction a categorical match to the federal generic offense of an aggravated felony consisting of sexual abuse of a minor. Having found Cabeda was convicted of an aggravated felony pursuant to 8 U.S.C. § 1227(a)(2)(A)(iii), I would deny the petition for lack of jurisdiction, which would preclude Cabeda from applying for cancellation of removal.

13 KRAUSE, Circuit Judge, concurring in part and concurring in the judgment.

I join JUDGE JORDAN’s excellent opinion (the Majority) in its three major respects. First, the generic definition of “sexual abuse of a minor,” 8 U.S.C. § 1101(a)(43)(A), includes a mens rea of knowledge with respect to the proscribed sexual con- duct. Second, Cabeda’s Pennsylvania offense requires a mens rea of only recklessness as to that conduct. Third, under our case law, that is the end of the matter: Where the elements of a state offense differ from and are broader than those of the generic federal offense, there is no categorical match regard- less whether the petitioner can show a “realistic probability” that the state will prosecute the offense in an overbroad way.1 Salmoran v. Att’y Gen., 909 F.3d 73, 81 (3d Cir. 2018) (citation omitted). Based on these three conclusions, I agree that—alt- hough the result is counterintuitive if not bizarre—Cabeda’s crime of conviction does not fit within the generic federal of- fense of “sexual abuse of a minor,” and she can neither be re- moved nor denied cancellation of removal on that basis.

I write separately because I reach the first of those conclu- sions by a different path. The Majority rejects the argument that Esquivel-Quintana v. Sessions, 137 S. Ct. 1562 (2017), re- quires us to revisit Restrepo v. Attorney General, 617 F.3d 787 (3d Cir. 2010), which, in interpreting “sexual abuse of a

1 That is, I join in full not only JUDGE JORDAN’s analysis of the contents of Pennsylvania law, Maj. Op. 19–22, but also his application of our holdings in the Salmoran line of cases, id. at 22–24, and his rejection of the alternative lines of reasoning offered in dissent by JUDGE GREENAWAY, id. at 24–30.

1 minor,” deferred to the Board of Immigration Appeals’s (BIA) decision in In re Rodriguez-Rodriguez, 22 I. & N. Dec. 991 (BIA 1999) (en banc). Its reasoning is that Esquivel-Quintana reaches no further than its context—fixing the age of consent for statutory rape offenses based solely on the participants’ ages—and gives us no basis to depart from Restrepo. Yet once the Majority turns to defining the generic federal offense, it does not rely on Restrepo, Rodriguez-Rodriguez, or even 18 U.S.C. § 3509(a)(8), the statutory provision on which those decisions were based. Instead, its analysis turns on 18 U.S.C. § 2243—a distinct provision that both Restrepo and Rodriguez- Rodriguez rejected as an interpretive guide—as well as statu- tory and contextual clues that those decisions discounted or ig- nored. So in the end, the Majority looks beyond Restrepo and finds a knowledge requirement by adopting the interpretive ap- proach of Esquivel-Quintana.

I would reach that result more directly by holding that Es- quivel-Quintana abrogates the bases on which Restrepo rested. Under the approach set out in Esquivel-Quintana, before deem- ing a provision ambiguous under Chevron, courts must engage in vigorous textual and contextual statutory analysis tailored to the precise interpretive question presented. That approach is impossible to square with Restrepo, which ignored or dis- counted sources of statutory meaning on which Esquivel-Quin- tana relied, assessed ambiguity at the broadest level, and re- quired deference to Rodriguez-Rodriguez for all future inter- pretive questions involving § 1101(a)(43)(A). Restrepo is a relic of an era when deference was far more reflexive and au- tomatic than it is today and, after Esquivel-Quintana, we can no longer follow Restrepo in uncritically relying on Rodriguez- Rodriguez or § 3509(a)(8).

2 I tackle these issues in three steps. I first highlight the fun- damental inconsistencies between Esquivel-Quintana and Re- strepo, which reveal that the latter decision’s assessment of ambiguity was erroneous. I then explore the case law on Ro- driguez-Rodriguez after Esquivel-Quintana to show why I view the issue as unsettled. Finally, I explain why the conclu- sion that § 1101(a)(43)(A) requires knowledge—with which I wholeheartedly agree—flows not from Restrepo or Rodriguez- Rodriguez, but from Esquivel-Quintana.

I. The Majority ably summarizes the contours of Esquivel- Quintana’s holding. I agree Esquivel-Quintana resolved only one aspect of the generic federal definition of “sexual abuse of a minor”: the meaning of “minor” “in the context of statutory rape offenses that criminalize sexual intercourse based solely on the age of the participants.” 137 S. Ct. at 1568. The Court did not set down an all-encompassing definition of “sexual abuse of a minor”; in fact, it expressly reserved interpretive questions not before it. See, e.g., id. at 1572. So I join the Majority in rejecting the argument, which the BIA endorsed,2

2 The BIA “agree[d] with [Cabeda]” about the contents of the new definition and cited the pages of Cabeda’s brief laying out that definition. A.R. 4. As those pages reveal, the defini- tion Cabeda put forward, and that the BIA accepted, was “the engaging in sexual contact with a person who is below a spec- ified age or who is incapable of giving consent because of age

3 that Esquivel-Quintana created a new, comprehensive generic federal definition to be applied in future cases. So far, so easy.

What is not so easy, though, is the distinct question whether Esquivel-Quintana’s analysis—the sources and evidence the Court considered and the order in which the Court considered them—undermines Restrepo’s analysis of § 1101(a)(43)(A) or its choice to defer to the BIA’s interpretation in Rodriguez-Ro- driguez. The Majority acknowledges that more complicated question, observing that there is “some tension between” Re- strepo and Esquivel-Quintana. Maj. Op. 12; accord id. at 13 n.4. I see that tension as irreconcilable. In multiple respects, the statutory analysis of Esquivel-Quintana is entirely incon- sistent with that of Restrepo. Indeed, Esquivel-Quintana has

or mental or physical incapacity.” A.R. 15 (emphasis omitted) (quoting Esquivel-Quintana, 137 S. Ct. at 1569).

But the idea that the quoted language established a new ge- neric definition borders on the fatuous. That language, which came from Merriam-Webster’s Dictionary of Law, see Es- quivel-Quintana, 137 S. Ct. at 1569, was just one of many pieces of evidence the Court considered en route to deciding the age of consent for a subset of statutory rape offenses. See id. at 1568–72. Nothing in the Court’s opinion even hints at an endorsement of that language as a definitive encapsulation of the generic federal definition. Nor, contrary to Cabeda’s argu- ment, is there any evidence to suggest that by quoting the phrase “sexual contact” as part of the dictionary definition, the Court meant to silently adopt the definition of that phrase from 18 U.S.C. § 2246(3).

4 so undermined that analysis that I believe Restrepo is no longer good law.

Under the first step of Chevron, we may “defer to [an] agency’s reasonable interpretation of a statute only if the text of the statute is unclear and we cannot discern congressional intent by utilizing various tools of statutory construction.” Quinteros v. Att’y Gen., 945 F.3d 772, 784 (3d Cir. 2019) (in- ternal quotation marks and citation omitted). But Restrepo put the cart before the horse, signaling that deference was appro- priate before engaging in statutory interpretation. What’s more, Restrepo’s eventual statutory analysis cannot withstand scrutiny after Esquivel-Quintana, which demanded a much more targeted approach and which took a contrary position on essentially all the sources of statutory meaning Restrepo con- sidered. Viewed through the lens of Esquivel-Quintana, Re- strepo erred in concluding that § 1101(a)(43)(A) was across- the-board ambiguous under Chevron.

A. Perhaps the most noteworthy aspect of Esquivel-Quintana cannot be gleaned only from reading the Court’s opinion. As it came to the Court, Esquivel-Quintana was very much a case about Chevron deference in general and about the wisdom of deferring to the BIA’s reliance on a procedural statute like § 3509(a)(8) in particular.3 See, e.g., Brief for the Petitioner at 35–48, Esquivel-Quintana, 137 S. Ct. 1562 (No. 16-54), 2016

3 As noted by the Majority, § 3509 deals with the rights of child victims and witnesses in federal criminal proceedings. See 18 U.S.C. § 3509. It does not criminalize any conduct.

5 WL 7384847; Brief for the Respondent at 36–55, Esquivel- Quintana, 137 S. Ct. 1562 (No. 16-54), 2017 WL 345128; Re- ply Brief for Petitioner at 17–23, Esquivel-Quintana, 137 S. Ct. 1562 (No. 16-54), 2017 WL 632460. Yet for all the space it commanded in the briefing, Chevron ended up relegated to a short paragraph at the tail end of the Court’s opinion. After reviewing the text along with dictionary definitions, statutory context, related federal statutes, and analogous state laws, the Court found “no need” to resort to Chevron deference at all “because the statute, read in context, unambiguously fore- close[d] the [BIA]’s interpretation.” Esquivel-Quintana, 137 S. Ct. at 1572. In other words, when faced with a question about the phrase “sexual abuse of a minor,” the Court treated Chevron as a canon of last resort, to be used if—but only if— the Court could not dispel ambiguity through a robust applica- tion of all the tools in its statutory toolkit. See also Epic Sys. Corp. v. Lewis, 138 S. Ct. 1612, 1630 (2018) (noting that “Chev- ron leaves the stage” where the “court, employing traditional tools of statutory construction,” can discern the meaning of the contested provision (citations omitted)).

Not so with Restrepo. In analyzing “sexual abuse of a mi- nor,” we first noted that, unlike other aggravated felonies listed in the Immigration and Nationality Act (INA), § 1101(a)(43)(A) contains neither cross-references to other statutory provisions nor explanatory parentheticals about the nature of the offense. Restrepo, 617 F.3d at 792. Then—before looking to any other textual or contextual clues—we suggested that “Congress ha[d] explicitly left a gap for the agency to fill,” “intend[ing] for the BIA to utilize its expertise to define the phrase.” Id. at 793 (quoting Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 843 (1984)). And—once again without

6 further statutory analysis—we concluded that because § 1101(a)(43)(A) lacks an express definition and cross-refer- ences, its “precise definition . . . is most assuredly not clear and unambiguous.”4 Id.

Put plainly, in Restrepo we waved the white flag of ambi- guity far too readily, and without performing the rigorous anal- ysis Esquivel-Quintana demands. But that is not Restrepo’s only flaw.

B. Esquivel-Quintana also clarified that in assessing ambigu- ity under step one of Chevron, courts must define the interpre- tive question narrowly. Said differently, courts must ask not whether a statute is ambiguous in general, but whether it is am- biguous as to the specific legal issue in the case. That, too, is impossible to square with Restrepo, which approached the is- sue of ambiguity practically in the abstract.

In analyzing the phrase “sexual abuse of a minor,” the Court in Esquivel-Quintana did not trouble itself with whether

4 We revealed we were jumping the gun on ambiguity in other ways, too. Among Restrepo’s brief statutory analysis is a footnote discussing the legislative history, which we under- stood to suggest Congress had intended § 1101(a)(43)(A) to be read “expansive[ly].” 617 F.3d at 795 n.6. We considered that reading consistent “with the structure of § 1101(a)(43)(A)”— but also “with the BIA’s interpretation” of the phrase. Id. Ac- cordingly, it appears Rodriguez-Rodriguez colored what should have been an independent analysis of statutory meaning at step one of Chevron.

7 every element of the federal generic offense was clear from the text. Instead, it limited its interpretive task to the specific cat- egory of crimes it faced: “statutory rape offenses that criminal- ize sexual intercourse based solely on the age of the partici- pants.” Esquivel-Quintana, 137 S. Ct. at 1568. Because it did so, the Court could assess § 1101(a)(43)(A) through an appro- priately tailored lens and conclude that the generic age of con- sent, at least where the statutory rape offense is based only on age (rather than, say, a relationship of trust between the adult and the minor), is sixteen. Id. at 1569–72. That narrow focus allowed the Court to avoid the broader dispute about whether it was appropriate to defer to a BIA interpretation based on a procedural statute like § 3509. See id. at 1572.

Since Esquivel-Quintana, the Court has reaffirmed its en- dorsement of this approach. Take Pereira v. Sessions, 138 S. Ct. 2105 (2018), a case involving the INA’s “stop-time rule,” which if triggered ends a noncitizen’s period of continuous physical presence for purposes of cancellation of removal. Id. at 2109. As in Esquivel-Quintana, in Pereira the issue of Chevron deference was hotly contested and commanded sub- stantial space in the briefing. See Brief for Petitioner at 24–55, Pereira, 138 S. Ct. 2105 (No. 17-459), 2018 WL 1083742; Brief for the Respondent at 21–52, Pereira, 138 S. Ct. 2105 (No. 17-459), 2018 WL 1557067; Reply Brief for Petitioner at 3–23, Pereira, 138 S. Ct. 2105 (No. 17-459), 2018 WL 1792078. Yet in its opinion, the Court criticized the parties for approaching the interpretive question too “broadly.” Pereira, 138 S. Ct. at 2113. Rather than asking whether “all ‘items listed’ in [8 U.S.C.] § 1229(a)(1)” must be included for a doc- ument to qualify as a “notice to appear” triggering the stop- time rule, the Court narrowed the question presented to the one

8 item missing from Pereira’s notice: “the date and time of [the] removal proceedings.” Id. Once it had sufficiently “narrowe[d] [the] question” presented, the Court found it “need not resort to Chevron deference” at all. Id. Instead, after reviewing the statutory text, neighboring provisions, and “common sense,” id. at 2114–16, it concluded the stop-time provision “ha[d] sup- plied a clear and unambiguous answer to the interpretive ques- tion at hand,” id. at 2113 (emphasis added). Once again, there- fore, the Court made clear that statutory ambiguity—without which deference to an agency is inappropriate—must be as- sessed through the lens of the precise dispute. See also Niel- sen v. Preap, 139 S. Ct. 954, 963–65 (2019) (resolving an in- terpretive dispute related to an INA provision governing the apprehension and detention of noncitizens based on a close analysis of the text and without citing Chevron, even though the parties had sharply litigated the deference issue); SAS Inst. Inc. v. Iancu, 138 S. Ct. 1348, 1358 (2018) (declining to ad- dress the parties’ Chevron arguments because, “after applying traditional tools of interpretation” to the specific question pre- sented, the Court was “left with no uncertainty that could war- rant deference”); Wis. Cent. Ltd. v. United States, 138 S. Ct. 2067, 2074 (2018) (similar); Sturgeon v. Frost, 139 S. Ct. 1066, 1080 n.3 (2019) (similar).

Now consider Restrepo. The specific question presented was a narrow one: whether inappropriate contact with a mi- nor’s sensitive areas “through her clothing” qualified as sexual abuse. See 617 F.3d at 800. But you would hardly know it from reading the opinion because the first time we grappled with that specific issue occurred well over halfway in, id. at 799—and long after we had deemed § 1101(a)(43)(A) ambig- uous, see id. at 793, 795. Instead, we analyzed the statute at

9 the highest level of generality, focusing on the lack of cross- references and explanatory parentheticals, id. at 792–93, and on general clues drawn from surrounding provisions and con- temporaneous legislation, id. at 793–95. It is no surprise that such a general review of the statute “le[ft] us in a state of inter- pretive uncertainty,” id. at 795: Without the benefit of deci- sions like Esquivel-Quintana and Pereira, we hadn’t defined the precise question we were meant to answer.

We have, since Restrepo, gotten it right, stating that we must assess the ambiguity of a statute “with respect to the spe- cific issue of law in the case.” Da Silva v. Att’y Gen., 948 F.3d 629, 634 (3d Cir. 2020) (citation omitted); accord, e.g., S.E.R.L. v. Att’y Gen., 894 F.3d 535, 549 (3d Cir. 2018) (Chev- ron step one requires courts to narrow in on “the precise ques- tion at issue” (citation omitted)). Had we followed that path in Restrepo, we may or may not have found ambiguity over whether “sexual abuse of a minor” encompasses “inappropriate touching of a minor through the clothing,” 617 F.3d at 799— but we would not have concluded that § 1101(a)(43)(A) was ambiguous across the board, and consequently we could not have purported to require “defin[ing] sexual abuse of a minor by reference to § 3509(a)” in all future cases, id. at 796. After Esquivel-Quintana, we must revisit that misstep.

C. Finally, and apart from those issues about the timing and framing of the ambiguity inquiry, Esquivel-Quintana has re- vealed that Restrepo’s statutory analysis was deeply flawed. In five respects, we either misinterpreted or ignored key clues about § 1101(a)(43)’s meaning.

10 First, Restrepo discounted a longstanding “rule of statutory construction”: that “identical words used in different parts of the same act are intended to have the same meaning.” Dep’t of Revenue of Or. v. ACF Indus., Inc., 510 U.S. 332, 342 (1994) (citation omitted). Congress added “sexual abuse of a minor” to the INA’s list of aggravated felonies in 1996. Illegal Immi- gration Reform and Immigrant Responsibility Act of 1996, Pub. L. No. 104-208, div. C, § 321(a)(1), 110 Stat. 3009, 3009- 627 (codified at 8 U.S.C. § 1101(a)(43)(A)). As part of the same omnibus legislation, Congress also amended the code provision criminalizing sexual abuse of a minor. Amber Hagerman Child Protection Act of 1996, Pub. L. No. 104-208, div. A, tit. I, § 121, subsec. 7(c), 110 Stat. 3009, 3009-31 (cod- ified at 18 U.S.C. § 2243(a)). So Restrepo invoked the same- act canon, urging us to interpret “sexual abuse of a minor” un- der § 1101(a)(43)(A) in line with § 2243. See Restrepo, 617 F.3d at 793. But we rejected that canon as “inapplicable” to omnibus legislation, concluding—without citing any prece- dent for this proposition—that terms used “in separate and dis- tinct statutes” were not subject to the rule. Id. at 793–94.

Esquivel-Quintana took a contrary approach. There, the Court leaned heavily on § 2243, which it described as “[a] closely related federal statute” providing valuable “evidence [of] the generic federal definition of sexual abuse of a minor.” 137 S. Ct. at 1570. It did so not just because § 2243 “contains the only definition of that phrase in the United States Code,” but also because Congress had amended § 2243 “in the same omnibus law that added sexual abuse of a minor to the INA.” Id. As the Court explained, the temporal proximity between revisions to § 1101(a)(43)(A) and § 2243 “provide[s] stron[g] suppor[t]” for the argument that the two should be interpreted

11 consistently. See id. (second and third alterations in original) (internal quotation marks and citation omitted). Here, the con- flict could hardly be more evident: Restrepo held inapplicable a canon of statutory construction that Esquivel-Quintana em- ployed in interpreting the same provision.

Second, Restrepo also neglected another longstanding canon: noscitur a sociis, under which “a word is known by the company it keeps.” Yates v. United States, 135 S. Ct. 1074, 1085 (2015). Here, the words are “sexual abuse of a minor,” and the company they keep is sinister indeed: “murder” and “rape.” 8 U.S.C. § 1101(a)(43)(A). Invoking noscitur a sociis, Restrepo argued that the severity of murder and rape should inform our interpretation of sexual abuse of a minor. See Re- strepo, 617 F.3d at 794. We rejected that argument as incon- sistent with “the broader context of the legislation at issue,” id., which—looking beyond § 1101(a)(43)(A) to unrelated parts of the INA—we characterized as an effort “to expand both the protections afforded to minors and the penalties applicable to aliens who commit crimes against minors.” Id. (More on that in a moment.) We otherwise did not explore how the phrase’s proximity to murder and rape should affect our analysis.

Once again, Esquivel-Quintana revealed the error in our thinking. The Court emphasized that “the INA lists sexual abuse of a minor in the same subparagraph as ‘murder’ and ‘rape,’” which are “among the most heinous crimes it defines as aggravated felonies.” Esquivel-Quintana, 137 S. Ct. at 1570 (quoting 8 U.S.C. § 1101(a)(43)(A)). For the Court, the nosci- tur a sociis canon was highly useful because it revealed “that sexual abuse of a minor encompasses only especially egre- gious felonies.” Id. (emphasis added). That insight is not just

12 missing from Restrepo—it is directly contrary to how Restrepo characterized § 1101(a)(43)(A).

Third, Restrepo ignored another valuable contextual clue about the contours of § 1101(a)(43)(A). Restrepo emphasized that when Congress added sexual abuse of a minor to the list of aggravated felonies, “it simultaneously amended the INA to classify crimes of domestic violence, stalking, child abuse, child neglect, and child abandonment as grounds for deporta- bility.” 617 F.3d at 794. We took that to mean that “Congress intended to implement comprehensive protections for minor victims that were expansive in nature.” Id. at 795 n.6. That “expansive” reading was critical to our analysis, as it supported our conclusion that it “would be counterintuitive” to adopt an interpretation of § 1101(a)(43)(A) that excluded certain state offenses. See id. at 795.

Restrepo’s expansive reading is nowhere to be found in Es- quivel-Quintana. In fact, after a thorough review of the statu- tory text and context, the Court concluded the opposite. Sec- tion 1101(a)(43)(A), the Court emphasized, is “an ‘aggravated’ offense,” meaning “one made worse or more serious by [the] circumstances.” Esquivel-Quintana, 137 S. Ct. at 1570 (inter- nal quotation marks and citation omitted). And unlike the broader grounds for removal discussed in Restrepo, sexual abuse of a minor is an aggravated felony carrying the addi- tional, and severe, sanction of ineligibility for cancellation of removal. That signal was a key reason the Court concluded that § 1101(a)(43)(A) “encompasses only especially egregious felonies,” id., but unfortunately it is one we missed in Restrepo.

Fourth, Restrepo misunderstood the role state criminal of- fenses should play in the analysis of a federal generic offense

13 like § 1101(a)(43)(A). Restrepo was convicted of a New Jer- sey statute criminalizing “an act of sexual contact,” N.J. Stat. Ann. § 2C:14-3(a), which includes intentional touching of in- timate areas “through clothing,” id. § 2C:14-1(d). But § 2243’s actus reus is a “sexual act,” 18 U.S.C. § 2243(a), which is de- fined as intentional touching “not through the clothing,” id. § 2246(2)(D). In Restrepo, we signaled our concern that rely- ing on § 2243 would exclude the New Jersey offense from the generic federal definition, something we felt would conflict with the “expansive” reading discussed above. See 617 F.3d at 794–95 nn.6 & 7. So instead we turned to § 3509, whose definition includes intentional touching “through clothing,” 8 U.S.C. § 3509(a)(9)(A). See Restrepo, 617 F.3d at 799–800.

Esquivel-Quintana reveals we were wrong to let the state crime drive our interpretation of the generic federal offense. Indeed, the government tried something similar in Esquivel- Quintana: Faced with a state statute criminalizing consensual sex between 21- and a 17-year-old, 137 S. Ct. at 1567, it of- fered up a definition broad enough to capture the statute, namely that sexual abuse covers sexual activity “directed at a person younger than 18.” Id. at 1569 (citation omitted). But the Court criticized the government for “turn[ing] the categor- ical approach on its head by defining the generic federal of- fense of sexual abuse of a minor as whatever is illegal under the particular law of the State where the defendant was con- victed.” Id. at 1570. Under that conception of the categorical approach, the Court explained, “there is no ‘generic’ definition at all.” Id.

Further, although the Court ultimately “look[ed] to state crim- inal codes for additional evidence,” it did so only after developing

14 an independent view on the meaning of § 1101(a)(43)(A) and, just as critical, only after a comprehensive cross-jurisdictional survey. Esquivel-Quintana, 137 S. Ct. at 1571–72; see id. at 1573 app. That survey allowed the Court to confirm its under- standing of the statute’s meaning by reference to a majority of jurisdictions—even though its conclusion excluded over a dozen states’ laws from the generic definition, id. at 1576 app. Although we referred to a few states’ laws in Restrepo, 617 F.3d at 795 n.7 (Pennsylvania and Delaware); see id. at 799 n.16 (citing, but not analyzing, sexual-abuse-of-a-minor analogs from five states and the District of Columbia), we stopped far short of the cross-jurisdictional analysis used in Es- quivel-Quintana.

We are not alone in having misunderstood the import and role of state crimes in analyzing a generic federal offense. See, e.g., Bedolla-Zarate v. Sessions, 892 F.3d 1137, 1141 n.4 (10th Cir. 2018) (noting that Esquivel-Quintana abrogated circuit precedent on looking to state criminal statutes). But after Es- quivel-Quintana, we should not hesitate to ask whether we in- appropriately “turn[ed] the categorical approach on its head,” 137 S. Ct. at 1570, in crafting § 1101(a)(43)(A) to capture the offense before us.

Fifth, Restrepo unnecessarily painted itself into a corner. Restrepo portrayed an interpretive dilemma in which it would have to choose either § 2243 or § 3509 as the definitive lodestar for “sexual abuse of a minor.” Choosing § 2243, we con- cluded, would exclude too many state offenses from the ge- neric federal definition. Restrepo, 617 F.3d at 795. So instead we chose § 3509(a)—and purported to do so for all future cases involving § 1101(a)(43)(A). See id. at 796.

15 But Esquivel-Quintana has shown that to be a false choice. There, although the Court relied heavily on § 2243, it did so only because it judged § 2243 to be a sensible guide for the problem it was facing. And it made clear it was not adopting § 2243 “as . . . the complete or exclusive definition” of § 1101(a)(43)(A). Esquivel-Quintana, 137 S. Ct. at 1571. If we had had the benefit of Esquivel-Quintana’s guidance, we would have understood that our choice between the two sources of statutory meaning was neither as absolute nor as consequential as we made it seem.

D. Although the Majority recognizes that “Esquivel-Quintana reflects . . . a more searching and nuanced approach” than the one we took in Restrepo, Maj. Op. 13 n.4, it attributes those differences to “shifting interpretive methodologies [that] are not usually viewed as carrying the force of stare decisis.” Id. I disagree in four respects.

First, I cannot help but note the irony in embracing this prop- osition when the delta between my and my colleagues’ views comes down to “interpretive methodologies” on both sides. If all that were binding on us were Restrepo’s substantive hold- ing—namely, that a statute criminalizing the “intentional touch- ing . . . , either directly or through clothing,” of a minor’s sensi- tive areas “for the purpose of degrading or humiliating the vic- tim or sexually arousing or sexually gratifying the actor” quali- fies as “sexual abuse of a minor” under § 1101(a)(43)(A), see 617 F.3d at 800 & n.18 (citation omitted)—we would be in complete alignment about how to resolve this appeal. It is only what Restrepo had to say about its interpretive method—i.e., that in analyzing “sexual abuse of a minor,” we would “refer[]

16 to § 3509(a)” as “a guide,” id. at 796 n.10 (citation omitted)— that is at issue between us. In essence, the Majority dismisses Esquivel-Quintana’s contributions as merely providing inter- pretive guidance but declines to apply the same lens to Re- strepo. But “what is sauce for the goose is normally sauce for the gander,” Heffernan v. City of Paterson, 136 S. Ct. 1412, 1418 (2016), and there is no reason why our own prior views on interpretive methods—especially views that are incon- sistent with current doctrine, see Maj. Op. 13 n.4—would con- tinue to demand adherence when the Supreme Court’s more recent views would not.

Second, the rule on which the Majority relies is, in my view, far narrower than as described. My colleagues assert, for in- stance, that “the Supreme Court ‘typically avoids methodolog- ical stare decisis,’” Maj. Op. 14 n.4 (quoting Am. Farm Bureau Fed’n v. U.S. EPA, 792 F.3d 281, 307 n.8 (3d Cir. 2015)), and to be sure, some Justices have suggested that statements in [the Court’s] opinions about . . . generally applicable interpretive methods, like the proper weight to afford historical practice in constitutional cases or legislative history in statutory cases,” are not entitled to the “full force” of stare decisis, e.g., Kisor v. Wilkie, 139 S. Ct. 2400, 2444 (2019) (Gorsuch, J., concurring in the judgment). But that principle addresses only the extent to which “future Justices” are bound by methodology in the Supreme Court’s prior decisions—that is, it addresses only “horizontal” stare decisis. Id. at 2444. And that distinction is critical: Whatever might be said of how the Court chooses to treat “its own precedents,” “[b]y contrast, vertical stare decisis is absolute, as it must be in a hierarchical system with ‘one su- preme Court.’” Ramos v. Louisiana, 140 S. Ct. 1390, 1416 n.5

17 (2020) (Kavanaugh, J., concurring in part) (emphasis added and omitted) (citation omitted).

Nor have we held that, as a Court of Appeals, we are free to discard the Supreme Court’s instructions on “interpretive methodology” as not a “form of ‘law’” that binds us, Maj. Op. 14 n.4 (quoting Am. Farm Bureau, 792 F.3d at 307 n.8). What my colleagues omit in discussing American Farm Bu- reau is that there, although we outlined an academic debate about so-called methodological stare decisis, we concluded we had “no occasion to explore further the contours of th[at] de- bate.” 792 F.3d at 307 n.8. Instead, we applied the interpretive method called for by precedent. See id. The Majority has ap- parently opted to embrace one side of that debate, but we as a Court did not do so in American Farm Bureau or any other decision of which I am aware.

Indeed, the “sea change in our jurisprudence,” Maj. Op. 15 n.6, if anything, would be if the Courts of Appeals were sud- denly free to discard as nonbinding the Supreme Court’s in- structions on “interpretive methodologies,” id. at 13 n.4 (cita- tion omitted)—for instance, that we may defer to an agency’s interpretation only if the statute “is ambiguous” and the inter- pretation “reasonable,” Nat’l Cable & Telecomms. Ass’n v. Brand X Internet Servs., 545 U.S. 967, 980 (2005), or that we may apply the rule of lenity “only . . . after consulting tradi- tional canons of statutory construction,” United States v. Shabani, 513 U.S. 10, 17 (1994). What are those if not “inter- pretive methodologies” addressing how to understand civil and criminal enactments?—and yet we follow them just as we fol- low all other binding statements from the High Court, see, e.g., Mejia-Castanon v. Att’y Gen., 931 F.3d 224, 232–33 (3d Cir.

18 2019) (Chevron); United States v. Barbosa, 271 F.3d 438, 455 (3d Cir. 2001) (lenity).

Third, perhaps searching for a limiting principle, the Ma- jority suggests we can ignore methodological guidance from the Supreme Court “when the decisions employing [such guid- ance] do not purport to overrule past precedent.” Maj. Op. 13– 14 n.4. Yet it is unclear exactly what “past precedent” the Es- quivel-Quintana Court was supposed to note it was overruling. After all, Esquivel-Quintana was the Court’s first occasion to address § 1101(a)(43)(A). And surely the Majority does not mean to suggest that the Supreme Court, to guarantee that its dictates will be followed, must canvas the case law of this Cir- cuit and every other Court of Appeals and identify each possi- ble point of abrogation. That suggestion would all but erase the very concept of abrogation, restricting changes in circuit law to those occasions when our decisions are expressly over- ruled. But that has never been our practice. To the contrary, “[a] panel of this Court may reevaluate the holding of a prior panel which conflicts with intervening Supreme Court prece- dent,” even if only “impliedly.” In re Krebs, 527 F.3d 82, 83– 84 (3d Cir. 2008). Here, however, the Majority turns vertical stare decisis on its head, elevating a concededly outmoded panel decision over a contrary and more recent statement from the Supreme Court.

Fourth, the fundamental inconsistencies between Esquivel- Quintana and Restrepo cannot be swept aside as if they were purely “methodological,” Maj. Op. 13 n.4. As discussed in de- tail above, see supra 10–16, Esquivel-Quintana has revealed Restrepo to be flawed not only in its methodology, but also in its substantive conclusions about § 1101(a)(43)(A)’s context

19 and meaning, particularly the expansive reading of the statute on which Restrepo (and Rodriguez-Rodriguez) so heavily de- pended. Those substantive conflicts fall beyond the methodo- logical-stare-decisis argument, whatever its merits, and go un- addressed by my esteemed colleagues.

* * * In brief, a close look at Esquivel-Quintana reveals that Re- strepo found ambiguity in § 1101(a)(43)(A) too quickly, too generally, and by ignoring or misinterpreting valuable evi- dence of statutory meaning. And as a result, Restrepo errone- ously suggested that under Chevron step one, it would always be appropriate to refer to the BIA’s interpretation in Rodriguez- Rodriguez.5

5 Because in my view § 1101(a)(43)(A) unambiguously re- quires a mens rea of knowledge with respect to the proscribed sexual conduct, I take no position on whether the BIA’s view in Rodriguez-Rodriguez is “arbitrary, capricious, or manifestly contrary to the statute,” Egan v. Del. River Port Auth., 851 F.3d 263, 271 (3d Cir. 2017) (citation omitted), either in terms of the statute’s text or its “design and structure . . . as a whole,” Util. Air Regulatory Grp. v. EPA, 134 S. Ct. 2427, 2442 (2014) (citation omitted).

I note, though, that Rodriguez-Rodriguez suffers from sev- eral of the flaws that Esquivel-Quintana has revealed in Re- strepo. Like Restrepo, Rodriguez-Rodriguez hinged on an “ex- pansi[ve]” view of § 1101(a)(43)(A), see 22 I. & N. Dec. at 994, one that ignored multiple pieces of evidence “suggest[ing] that sexual abuse of a minor encompasses only especially

20 As a result, I cannot agree this is simply a matter of “Chev- ron deference[’s] [being] unnecessary in one specific instance” and not in another, Maj. Op. 12. The Court’s analysis in Es- quivel-Quintana not only implicates, but directly undermines, everything we said and did in Restrepo. That is more than enough to “‘undercut the decisional basis’ of Restrepo,” id. at 12–13 (quoting West v. Keve, 721 F.2d 91, 93 (3d Cir. 1983)), and accordingly we are not bound by that decision’s suggestion that we must uncritically defer to § 3509(a)(8) in all cases in- volving § 1101(a)(43)(A).

egregious felonies,” Esquivel-Quintana, 137 S. Ct. at 1570. And, again like Restrepo, Rodriguez-Rodriguez seems to have reverse-engineered the generic federal definition to fit the state offense, see 22 I. & N. Dec. at 995–96, thereby “turn[ing] the categorical approach on its head,” Esquivel-Quintana, 137 S. Ct. at 1570. Although it is possible that Rodriguez-Rodriguez’s bottom-line conclusion, 22 I. & N. Dec. at 993, that “sexual abuse of a minor” encompasses non-contact indecent exposure offenses was correct, future panels of this Court faced with genuine ambiguity may have to decide whether Rodriguez-Ro- driguez offers up a reasonable interpretation. If so, they will also have to confront the fact that Rodriguez-Rodriguez looked to § 3509(a)(8) only as a “guide,” id. at 996, and only in one specific context. Cf., e.g., Amos v. Lynch, 790 F.3d 512, 520 (4th Cir. 2015) (holding that Rodriguez-Rodriguez is not enti- tled to deference “[b]eyond [its] limited holding” on indecent exposure offenses).

21 II. My learned colleagues view the idea that Restrepo’s defer- ence to Rodriguez-Rodriguez survives Esquivel-Quintana as something of a fait accompli, which we and other Courts of Appeals have already decided. In my view, neither we nor, with one exception, any other circuit has reached that conclu- sion, and in fact most of our sister circuits’ case law suggests we must revisit Restrepo in light of Esquivel-Quintana.

A. Relying on Mondragon-Gonzalez v. Attorney General, 884 F.3d 155 (3d Cir. 2018), the Majority infers, as a general matter, that “we have already taken the position . . . that Es- quivel-Quintana speaks specifically to the question of statutory rape, not more broadly to the definition of the generic offense of sexual abuse of a minor.” Maj. Op. 15. I do not believe Mondragon-Gonzalez sweeps so broadly; indeed, that case did not involve the aggravated felony of “sexual abuse of a minor” at all.

In Mondragon-Gonzalez, the issue was whether the peti- tioner’s Pennsylvania conviction of unlawful contact with a minor constituted a “crime of child abuse” under 8 U.S.C. § 1227(a)(2)(E)(i). 884 F.3d at 157. After Esquivel-Quintana came down, the petitioner argued that “a child for purposes of . . . the term ‘crime of child abuse’ means someone under the age of 16,” and not (as the BIA had concluded) anyone un- der eighteen. See id. at 160. In rejecting that argument, we emphasized Esquivel-Quintana’s limits, noting that the Court had addressed only the “sexual abuse of a minor” aggravated felony provision. Id. Unsurprisingly, we concluded that “Es- quivel-Quintana has no application . . . at all,” id., in a case

22 turning on the meaning of the word “child” in the statutory phrase “crime of child abuse.” There is little that is “notable,” Maj. Op. 15 n.6, in that decision; to the contrary, the surprise would have been if a panel addressing an entirely unrelated INA provision had launched into an unnecessary analysis of whether Esquivel-Quintana had abrogated unrelated prece- dent.

Because nothing in Mondragon-Gonzalez addressed the ef- fect of Esquivel-Quintana’s reasoning on Restrepo’s statutory analysis, that issue remains open in our Circuit.

B. To the extent we may draw wisdom from our sister circuits, they generally favor revisiting Restrepo’s deference to Rodri- guez-Rodriguez in the aftermath of Esquivel-Quintana.

Many of those circuits reject outright the notion that Rodri- guez-Rodriguez is entitled to deference. Several reached that conclusion before Esquivel-Quintana. See Estrada-Espi- noza v. Mukasey, 546 F.3d 1147, 1157–58 (9th Cir. 2008) (en banc) (holding that Rodriguez-Rodriguez, which looked to § 3509(a)(8) only as a “guide” and which did not identify de- fined elements of the generic federal offense, did not set down a definitive interpretation entitled to Chevron deference), over- ruled on other grounds by United States v. Aguila-Montes de Oca, 655 F.3d 915 (9th Cir. 2011) (en banc); Amos v. Lynch, 790 F.3d 512, 519–20 (4th Cir. 2015) (following Estrada-Es- pinoza); Rangel-Perez v. Lynch, 816 F.3d 591, 598–601 (10th Cir. 2016) (same), abrogated in other part by Esquivel-Quin- tana, 137 S. Ct. 1562; see also Contreras v. Holder, 754 F.3d 286, 292–93 (5th Cir. 2014) (suggesting that Rodriguez-

23 Rodriguez’s definition may “not [be] a reasonable one”). An- other has implicitly joined the fold since. See Garcia-Ur- bano v. Sessions, 890 F.3d 726, 728 (8th Cir. 2018) (noting that the BIA has “interpreted the phrase [‘sexual abuse of a minor’] through case-by-case adjudication,” not a definitive interpreta- tion). In the Fourth, Fifth, Eighth, Ninth, and Tenth Circuits, therefore, there is simply no Restrepo analog for those courts to reconsider after Esquivel-Quintana.

And in the few circuits with a Restrepo analog—the Sec- ond, see Mugalli v. Ashcroft, 258 F.3d 52, 60 (2d Cir. 2001), Sixth, see Esquivel-Quintana v. Lynch, 810 F.3d 1019, 1026– 27 (6th Cir. 2016), rev’d in part, 137 S. Ct. 1562 (2017), and Seventh, see Velasco-Giron v. Holder, 773 F.3d 774, 776 (7th Cir. 2014)—the reaction to Esquivel-Quintana has been any- thing but uniform. To be sure, in Correa-Diaz v. Sessions, 881 F.3d 523 (7th Cir. 2018), the court adhered to pre-Es- quivel-Quintana precedent deferring to Rodriguez-Rodriguez, reasoning that the Supreme Court’s decision made Chevron deference inappropriate only “as to [the] one precise question” presented there. Id. at 527–28. So the Seventh Circuit, at least, is firmly in the Majority’s camp.

The same cannot be said, though, for the Second Circuit. In Acevedo v. Barr, 943 F.3d 619 (2d Cir. 2019), the court noted that although Esquivel-Quintana “avoided any issue of Chevron deference with respect to . . . § 3509(a)(8)” as to the issue before it, “it also did not foreclose the BIA’s use of that statute in other instances,” and thus circuit precedent deferring to Rodriguez-Rodriguez “survives Esquivel-Quintana.” Id. at 623. That is fair, as far as it goes. But Acevedo then side- stepped § 3509(a)(8) altogether and instead performed a wide-

24 ranging analysis of “the structure of the INA, the inherent egre- gious nature of an aggravated felony, and closely-related stat- utes,” id. at 624—including § 2243, which Esquivel-Quintana had cited and which Rodriguez-Rodriguez had rejected. In es- sence, ongoing deference to Rodriguez-Rodriguez was more honored in the breach than in the observance.

The Sixth Circuit’s view is yet unclear. But in Keeley v. Whitaker, 910 F.3d 878 (6th Cir. 2018), a case involving the nearby aggravated felony of “rape,” 8 U.S.C. § 1101(a)(43)(A), the court showed both sensitivity to the need for rigorous stat- utory analysis and reluctance to give unwarranted deference to the BIA. It rejected the BIA’s interpretation that “rape” in- cludes digital penetration, reasoning that the BIA had given short shrift to statutory language in order to sweep more state laws into the aggravated felony category. See Keeley, 910 F.3d at 883–84. That criticism echoed Esquivel-Quintana’s caution that the government cannot “turn[] the categorical approach on its head by defining the generic federal offense . . . as whatever is illegal under the particular law of the State where the defend- ant was convicted,” 137 S. Ct. at 1570. And, as in Esquivel- Quintana, Keeley found “no need” to resort to Chevron defer- ence after fully exhausting the tools of statutory construction. 910 F.3d at 885–86 (citation omitted). So although Rodriguez- Rodriguez’s vitality in the Sixth Circuit is uncertain, Keeley ar- guably laid the groundwork for independent statutory interpre- tation beyond § 3509(a)(8).

To review: The Seventh Circuit’s decision in Correa-Diaz, to date, remains an outlier. Of the three Courts of Appeals to have followed Restrepo’s path before Esquivel-Quintana, one has stayed the course; another has swept § 3509(a)(8) to the

25 side; and the third has yet to return to the issue but has signaled a muscular view of courts’ responsibility to engage in statutory interpretation before deferring to the BIA. And in five other circuits, Rodriguez-Rodriguez commands no deference at all. That is the conclusion compelled by Esquivel-Quintana in this context and the position we should adopt today.

III. To follow the course the Court has charted, we must assess § 1101(a)(43)(A) through the lens of “the least of the acts crim- inalized by the state statute.” Esquivel-Quintana, 137 S. Ct. at 1568. The Majority does so by focusing on the fact that invol- untary deviate sexual intercourse under Pennsylvania law has a minimum mens rea of recklessness as to the proscribed sex- ual conduct.6 18 Pa. C.S.A. §§ 3101, 3123(a)(7); see id.

6 Although I take no issue with this approach, I note another feature of the Pennsylvania offense: that it can be committed by “penetration, however slight, of the genitals or anus of an- other person with a foreign object for any purpose other than good faith medical, hygienic or law enforcement procedures.” 18 Pa. C.S.A. § 3101; see id. § 3123(a). That feature sets Penn- sylvania law apart from that of other states, almost all of which define such inappropriate sexual contact in terms of a specific “intent to abuse, humiliate, harass, or degrade a person or to arouse or gratify the sexual desire of a person.” E.g., Alaska Stat. § 26.05.890(h)(5)(B); accord Ala. Code § 13A-6-60(3); Ariz. Rev. Stat. Ann. § 13-1419(D)(1)(a); Ark. Code Ann. § 5- 14-101(11); Cal. Penal Code § 288(a); Colo. Rev. Stat. § 18-6- 403(2)(c); Conn. Gen. Stat. § 53a-65(3); Del. Code Ann. tit. 11, § 1356(4); D.C. Code § 22-3001(9); Fla. Stat. § 827.071(1)(h); Ga. Code Ann. § 16-6-4(a)(1); Haw. Rev.

26 Stat. § 712-1210; Idaho Code § 18-1507(1)(c); 720 Ill. Comp. Stat. 5/11-0.1; Ind. Code § 35-42-4-3(b); Iowa Code § 709.8(1); Kan. Stat. Ann. § 21-5506(a)(1); Ky. Rev. Stat. Ann. § 510.010(7); La. Stat. Ann. § 14:81(A); Me. Stat. tit. 17- A, § 251(1)(C)(3); Md. Code Ann., Crim. Law § 3- 301(d)(1)(v); Mich. Comp. Laws § 750.145c(1)(g); Miss. Code Ann. § 97-5-23(1); Mo. Rev. Stat. § 566.010(6); Mont. Code Ann. § 45-5-701(9); Neb. Rev. Stat. § 28-318(5); Nev. Rev. Stat. § 201.520(4)–(5); N.H. Rev. Stat. Ann. § 632-A:1(IV); N.J. Stat. Ann. § 2C:14-1(d); N.Y. Penal Law § 130.00(3); N.C. Gen. Stat. § 14-202.1(a)(1); N.D. Cent. Code § 12.1-20- 02(5); Ohio Rev. Code Ann. § 2907.01(B); Okla. Stat. tit. 44, § 920(F)(1)(c); Or. Rev. Stat. § 163.305(6); 11 R.I. Gen. Laws § 11-37-1(7); S.C. Code Ann. § 16-3-655(C); S.D. Codified Laws § 22-22-7.1; Tenn. Code Ann. § 39-15-501(13); Tex. Pe- nal Code Ann. § 21.01(2); Utah Code Ann. § 76-5-404.1(2); Vt. Stat. Ann. tit. 13, § 2602(a)(1); Va. Code Ann. § 18.2- 67.10(6); Wash. Rev. Code § 9A.44.010(2); W. Va. Code § 61- 8B-1(6); Wis. Stat. § 948.01(5)(a); Wyo. Stat. Ann. § 6-2- 301(a)(vii)(A); see 18 Pa. C.S.A. § 3101 (defining “[i]ndecent contact” as “[a]ny touching of the sexual or other intimate parts of the person for the purpose of arousing or gratifying sexual desire, in any person”); id. § 3126(a)(8) (criminalizing inde- cent contact by certain defendants with a “complainant . . . less than 16 years of age”); see also 9 Guam Code Ann. § 25.10(a)(8); V.I. Code Ann. tit. 14, § 1699(d). And many of our sister circuits, interpreting § 1101(a)(43)(A) or similarly worded sentencing enhancements, have concluded that “sexual abuse, consistent with its common meaning, connotes the use or treatment of so as to injure, hurt, or damage for the purpose

27 § 302(c). It concludes, quite rightly, that the generic federal offense of “sexual abuse of a minor” requires knowledge as to the sexual conduct. Under our precedent, this facial difference in the elements of the state and generic federal offenses

of sexual or libidinal gratification.” E.g., United States v. Ma- teen, 806 F.3d 857, 861 (6th Cir. 2015) (collecting cases).

I therefore would approach this case by asking two ques- tions: (i) whether “sexual abuse of a minor,” 8 U.S.C. § 1101(a)(43)(A), in the context of a statute criminalizing de minimis penetration with a foreign object, requires an intent to humiliate, harass, degrade, or arouse or gratify the libidinal de- sire of any person; and, if so, (ii) whether Pennsylvania’s carveout for “good faith medical, hygienic or law enforcement procedures,” 18 Pa. C.S.A. § 3101, is sufficiently broad or in- stead leaves less culpable conduct subject to prosecution, cf., e.g., Alaska Code § 11.81.900(b)(61)(B) (excluding from the definition of “sexual contact” acts “that may reasonably be construed to be normal caretaker responsibilities for a child” or that are “performed for the purpose of administering a recog- nized and lawful form of treatment that is reasonably adapted to promoting the physical or mental health of the person being treated”); Ariz. Rev. Stat. § 13-1401(A)(3)(b) (excluding from the definition of “sexual contact” any “direct or indirect touch- ing or manipulating during caretaking responsibilities, or inter- actions with a minor or vulnerable adult that an objective, rea- sonable person would recognize as normal and reasonable un- der the circumstances”). Still, I join the Majority in analyzing the Pennsylvania statute through the lens of the more general mens rea requirement and leave for another day the question of any libidinal-gratification-or-degradation requirement.

28 “leave[s] nothing to the ‘legal imagination,’” Zhi Fei Liao v. Att’y Gen., 910 F.3d 714, 724 (3d Cir. 2018) (quoting Gonza- les v. Duenas-Alvarez, 549 U.S. 183, 193 (2007)), and, because the state offense sweeps in less culpable conduct than the ge- neric federal offense, our categorical approach analysis is at an end. Again, I agree in all respects. See supra 1 & n.1.

We part company on the basis for concluding that § 1101(a)(43)(A) contains a knowledge requirement, a conclu- sion my esteemed colleagues view as consistent with ongoing fealty to Restrepo. Although this approach mirrors the Second Circuit’s in Acevedo, I find it unconvincing. In my view, that conclusion is more consistent with the open-ended statutory construction that Esquivel-Quintana demands than it is with the reflexive deference that Restrepo calls for.

The key analytical move in Acevedo is that § 3509(a)(8) “does not . . . provide guidance as to the mens rea a defendant must possess.” 943 F.3d at 623–24; accord Maj. Op. 16 (“Sec- tion 3509(a)(8) does not specify a mens rea requirement, and we cannot defer to a nullity . . . .”); id. at 19 (asserting that § 3509(a)(8) addresses “the contours of the substance of the offense” but not “the necessary mental state”); id. at 18 n.7 (again asserting that “§ 3509(a)(8) fails to provide the neces- sary guidance”). There are two things to note about that ana- lytical move. First, it recognizes what many of our sister cir- cuits have acknowledged: that Rodriguez-Rodriguez did not adopt § 3509(a)(8) “as a definitive standard or definition” in interpreting “sexual abuse of a minor” in all cases. Amos, 790 F.3d at 519–20 (quoting Rodriguez-Rodriguez, 22 I. & N. Dec. at 996); see Rangel-Perez, 816 F.3d at 601 (holding that § 3509(a)(8) is not “the exclusive touchstone for defining the

29 INA’s generic ‘sexual abuse of a minor’ [offense]”). Rather, § 3509(a)(8) is merely “a guide,” Rodriguez-Rodriguez, 22 I. & N. Dec. at 996, and not a comprehensive one. At very least, therefore, Restrepo’s statement that “we will define sexual abuse of a minor by reference to § 3509(a),” 617 F.3d at 796, is not as absolute as it may seem, and there may be times when § 3509(a)(8) offers limited guidance or no guidance at all. See id. at 796 n.10 (recognizing that § 3509 is only “a guide” (cita- tion omitted)).

Second, Acevedo’s analytical move begs the question whether the lack of language on mens rea in § 3509(a)(8) re- flects open-ended silence on the required mental state—rather than, say, an indication that the definition has no required men- tal state. Answering that question is ultimately a matter of stat- utory interpretation,7 but Acevedo assumes it away. In my view, the reason it does so is plain: “Given the inherent seri- ousness of an aggravated felony and the harsh immigration consequences that come from that categorization”—not to mention other indications of statutory meaning, including the “closely related” provision in 18 U.S.C. § 2243—it is simply untenable that any mens rea less than knowledge could form part of the generic federal definition. Acevedo, 943 F.3d at

7 It might, for instance, depend on the extent to which “‘mere omission from a criminal enactment of any mention of criminal intent’ [can] be read ‘as dispensing with it,’” United States v. Elonis, 135 S. Ct. 2001, 2009 (2015) (citation omit- ted)—an issue further complicated by whether § 3509(a) counts as a “criminal enactment” and whether recklessness can be read into such an enactment, see id. at 2012 (reserving this question).

30 624. And because § 3509(a)(8) does not require knowledge, it is cast aside as not “provid[ing] guidance.” Id. at 623. That approach may allow for avoiding an impression of revisiting a prior panel decision in light of intervening precedent, but it does so at the cost of logical coherence.

To be clear, I take no issue with the Majority’s statutory analysis, which cogently explains why “sexual abuse of a mi- nor,” read in context, unambiguously embodies a knowledge requirement. But I would ground that analysis in an acknowl- edgment that at least here, deferring to Rodriguez-Rodriguez is inappropriate and, to the extent Restrepo suggests otherwise, it is no longer good law.

IV. For now, Restrepo limps on. Yet there may come a day when the conflict between § 3509(a)(8) and other sources of statutory meaning is less easily avoided. Should that day come, I would recognize that § 3509(a)(8) is but one of many sources we must consider in analyzing, using all the tools of statutory construction at our disposal, whether a particular as- pect of § 1101(a)(43)(A) is ambiguous.