SUPREME COURT OF THE UNITED STATES
CLEMENTE AVELINO PEREIDA, PETITIONER v.
ROBERT M. WILKINSON, ACTING ATTORNEY GENERAL
on writ of certiorari to the united states court of appeals for the eighth circuit
[March 4, 2021]
Justice Breyer, with whom Justice Sotomayor and Justice Kagan join, dissenting.
This case, in my view, has little or nothing to do with burdens of proof. It concerns the application of what we have called the “categorical approach” to determine the nature of a crime that a noncitizen (or defendant) was previously convicted of committing. That approach sometimes allows a judge to look at, and to look only
at, certain specified documents. Unless those documents show that the crime of conviction necessarily
falls within a certain category (here a “crime involving moral turpitude”), the judge must find that the conviction was not for such a crime. The relevant documents in this case do not show that the previous conviction at issue necessarily was for a crime involving moral turpitude. Hence, applying the categorical approach, it was not. That should be the end of the case.
Mr. Pereida is a citizen of Mexico, not the United States. He has lived in the United States for roughly 25 years. In that time, he and his wife have raised three children. He helped support them by working in construction and cleaning. One child is a U. S. citizen. In 2009 the Department of Homeland Security issued a notice to appear that charged Mr. Pereida with removability because he was never lawfully admitted to the United States. Mr. Pereida conceded that he is removable. But he asked the Attorney General to cancel his removal. The Attorney General has discretion to cancel an order of removal if removal would result in extreme hardship to the noncitizen’s U. S. citizen (or lawful-permanent-resident) spouse, parent, or child.
8 U. S. C. §1229b(b)(1)(D). A noncitizen is ineligible for this discretionary relief, however, if, among other things, he has “been convicted of ” a “crime involving moral turpitude.” §§1229b(b)(1)(C), 1182(a)(2)(A)(i)(I).
Mr. Pereida, in 2010, pleaded nolo contendere
to, and was found guilty of, having committed a Nebraska state crime, namely, attempt to commit criminal impersonation in violation of Neb. Rev. Stat. §28–608. See §28–608 (2008) (since amended and moved to §28–638 (2020)); §28–201(1)(b). The question here is whether this conviction was for a “crime involving moral turpitude.”
I believe we must answer this question by applying what we have called the “categorical approach.” The Immigration and Nationality Act (INA) makes a noncitizen ineligible for cancellation of removal if that noncitizen has been “convicted” of certain “offense[s],”
8 U. S. C. §1229b(b) (1)(C), including “crime[s] involving moral turpitude,” §1182(a)(2)(A)(i)(I). Similarly, the Armed Career Criminal Act (ACCA) increases the sentence of a defendant convicted of possessing a firearm as a felon if that defendant has three or more previous “convictions” for a “violent felony” or “serious drug offense.”
18 U. S. C. §924(e)(1). In ordinary speech, “crime,” “offense,” and “felony” are ambiguous: They might refer to actions that a defendant took on a particular occasion, or they might refer to the general conduct that a criminal statute forbids. So the question arises, shall a judge look to how the noncitizen or defendant behaved on a particular occasion (for example, to see whether he behaved violently)? Or shall a judge look to the statute that the defendant was convicted of violating (to see whether the behavior that it forbids is categorically violent)?
We have answered this question clearly and repeatedly in both the INA and ACCA contexts. We have held that both statutes mandate a categorical approach by asking what offense a person was “convicted
” of, not what acts he “committed
569 U.S. 184
, 191 (2013) (emphasis added) (discussing the INA); see also Taylor
v. United States
495 U.S. 575
, 600 (1990) (discussing ACCA). The categorical approach requires courts to “loo[k ] only to the statutory definitions of the prior offenses, and not to the particular facts underlying those convictions.” Id.
, at 600; see also Esquivel-Quintana
, 581 U. S. ___, ___–___ (2017) (slip op., at 2–3) (applying the categorical approach under the INA); Mellouli
575 U.S. 798, 804–806 (2015) (same); Moncrieffe
, 569 U. S., at 190 (same); Carachuri-Rosendo
560 U.S. 563
, 576 (2010) (same); Gonzales
549 U.S. 183
, 185–186 (2007) (same); Mathis
v. United States
, 579 U. S. ___, ___ (2016) (slip op., at 3) (applying the categorical approach under ACCA); Johnson
v. United States
559 U.S. 133
, 144 (2010) (same); Descamps
v. United States
570 U.S. 254
, 257 (2013) (same); Shepard
v. United States
544 U.S. 13
, 19–20 (2005) (same); Taylor
, 495 U. S., at 600 (same). A judge, looking at a prior conviction, will read the statutory definition of the offense of conviction and decide whether anyone convicted under that offense is necessarily guilty of the type of crime that triggers federal penalties, e.g.,
an enhanced sentence or ineligibility for cancellation of removal. See Mellouli
, 575 U. S., at 805; Taylor
, 495 U. S., at 600.
Consider a hypothetical example of this approach. Suppose a noncitizen’s previous conviction was for violating State Statute §123. Suppose further that the Government argues the noncitizen is ineligible for cancellation of removal because he was “convicted of an offense under” §1227(a)(2), namely, an “aggravated felony.” 8 U. S. C. §§1229b(b)(1)(C), 1227(a)(2)(A)(iii). An immigration judge, looking at the conviction, will simply read §123 and decide whether anyone convicted under §123 is necessarily guilty of an aggravated felony, as that term is defined in the INA. See §1101(a)(43). That is, the judge will decide whether the conduct that §123 prohibits is in general an aggravated felony. The judge will not
look to see whether the defendant’s actual conduct on the relevant occasion was or was not an aggravated felony.
Difficult questions can arise when judges apply the categorical approach. State statutes criminalize many kinds of behavior, often differing in detail one from another. Take burglary, for example, which is an “aggravated felony” under the INA. §1101(a)(43)(G). We can assume that the term “burglary” here, as in ACCA, refers to a specific crime, i.e.
, generic burglary. See Taylor
, 495 U. S., at 599; cf. Duenas-Alvarez
, 549 U. S., at 189 (accepting that the INA’s reference to “theft” in §1101(a)(43)(G) is to generic theft). Generic burglary is “unlawful or unprivileged entry into, or remaining in, a building or structure, with intent to commit a crime.” Taylor
, 495 U. S., at 599. Now suppose that §123 defines “burglary” in a different way (say, by including lawful entry with intent to steal). The sentencing judge then must compare the elements of the state statute and the elements of generic burglary. If the minimum conduct criminalized by the state statute is encompassed by generic burglary, then the conviction is for generic burglary; if not, then the conviction is not for that aggravated felony. See Moncrieffe
, 569 U. S., at 190–191. In our §123 example, the judge would therefore conclude that the conviction is not for an aggravated felony.
And what is a judge to do if a state statute is “divisible” into several different offenses, some of which are aggravated felonies and some of which are not? Suppose, for example, that §123 has three subsections referring to (a) burglary of a dwelling, (b) burglary of a boat, and (c) burglary of a railroad car. Since generic burglary is of a dwelling or structure, only subsection (a) qualifies as an aggravated felony. How is the judge to know which
subsection the defendant was convicted of violating? Simple, we have replied. Under the “modified categorical approach,” the judge can look to a limited set of court records to see if they say which subsection the defendant was convicted of violating. The judge can look at the charging papers and the jury instructions (if there was a jury), see Taylor
, 495 U. S., at 602, and the plea agreement, plea colloquy, or “some comparable judicial record” of the plea (if there was a plea), Shepard
, 544 U. S., at 26; see also Nijhawan
557 U.S. 29
, 35 (2009) (quoting Shepard
, 544 U. S., at 26). If these documents reveal that the previous conviction was for §123(a) (dwelling), then, and only then, can the judge conclude that the conviction is for an aggravated felony. As we explained in Taylor
, the modified categorical approach “allow[s]” “the Government . . . to use [a] conviction” under an overbroad statute to trigger federal penalties (there, ACCA’s sentencing enhancement) if the statute contains multiple offenses and
the permissible documents show that “the jury necessarily had to find” (or the defendant necessarily admitted to) a violent felony. 495 U. S., at 602.
What if, after looking at all the sources we have listed, the judge still does not know which of the three different kinds of burglary was the basis for the conviction? Suppose all the relevant documents that exist speak only
of a violation of §123. Period. What then? As discussed infra
, at 9,
that is the question we face here, and our cases provide the answer. The judge cannot look at evidence beyond the specified court records. See, e.g.
, 579 U. S., at ___ (slip op., at 18). Instead, in such a case, the judge is to determine what the defendant necessarily admitted (or what a jury necessarily found) in order for a court to have entered a conviction under §123, since that is the conviction reflected in the permissible documents.
The purpose of the modified categorical approach, like the categorical approach it helps implement, is to compare what “was necessarily found or admitted” to the elements of the generic federal offense. Id.
, at ___ (slip op., at 4).
If the record materials do not specify that the defendant was convicted of §123(a) (dwelling) rather than §123(b) (boat) or §123(c) (railroad car), or if the record materials do not exist at all, then the sentencing judge cannot say that generic burglary was necessarily found or admitted. The Court has said as much before. In Shepard
, the Court acknowledged that both the “vagaries of abbreviated plea records” and the destruction of “stenographic notes” of a jury charge would preclude the application of ACCA. 544 U. S., at 22. In Mathis
, the Court explained that
if the “record materials” do not “speak plainly,” then “a sentencing judge will not be able to satisfy ‘Taylor
’s demand for certainty’ when determining whether a defendant was convicted of a generic offense.” 579 U. S., at ___ (slip op., at 18). And we applied this principle in Johnson
, holding that a prior conviction did not count as a “violent felony” under ACCA because the statute of conviction swept more broadly than a “violent felony” and “nothing in the record of [the] conviction permitted the District Court to conclude that it rested upon anything more than the least of th[e] acts” prohibited by the state statute. See 559 U. S., at 137; see also id.,
(“[I]n many cases state and local records from” state convictions “will be incomplete” and “frustrate application of the modified categorical approach”).
That is to say, if (as far as the available, listed documents reveal) the judge could have entered the conviction without the noncitizen admitting to burglarizing a dwelling, then the immigration judge cannot hold that the conviction is necessarily for an aggravated felony. Applying the categorical approach, the judge must find the conviction is not for an aggravated felony at all.
Why would Congress have chosen such a seemingly complicated method? The method would appear sometimes to lead to counterintuitive results. After all, if the prior crime is for burglary and the offense occurred in a small town near the Mojave Desert, it seems unlikely that the conviction was based on burglary of a boat. Yet, in the absence of an indication from the permissible documents that the conviction necessarily was for burglary of a dwelling, the judge cannot classify the crime of conviction as an “aggravated felony.”
The primary reason for choosing this system lies in practicality. Immigration judges and sentencing judges have limited time and limited access to information about prior convictions. See Mellouli
, 575 U. S., at 806; Moncrieffe
, 569 U. S., at 200–201; Shepard
, 544 U. S., at 23, n. 4. The vast majority of prior convictions reflect simple guilty pleas to the crime charged, and, where the record papers are silent, efforts to uncover which of several crimes was “really” at issue can force litigation that the guilty plea avoided. Suppose that the defendant in the Mojave Desert pleaded guilty to a violation of §123 and there is no indication in the relevant record documents which subsection was the basis for the conviction. To find out which of the several provisions was the basis for the conviction, it might be necessary to call as witnesses the defendant, the prosecutor, or even the judge, and question them about a criminal proceeding that perhaps took place long ago. To make his case, the defendant might now deny that the provision involving a dwelling was at issue, and he might seek the opportunity to prove that. As a result, the immigration judge or sentencing judge now might have to conduct the very fact-based proceeding that the earlier guilty plea was designed to avoid. See id.
, at 21–23.
I do not know how often this kind of counterintuitive example will arise. But I do know that, in such a case, there is a safeguard against the harms that the “prior conviction” provisions are designed to stop. In the INA context, if a noncitizen is eligible for cancellation of removal, the Attorney General has discretionary
power to cancel the removal order. Where he believes the noncitizen in fact previously burgled a dwelling (or worse), he can simply deny relief. And in the ACCA context, a sentencing judge, even where ACCA is inapplicable, has some discretion in determining the length of a sentence. If he finds that the present defendant in fact burgled, say, a dwelling and not a boat, he can take that into account even if the sentencing enhancement does not apply.
And most importantly, whatever the costs and benefits of the categorical approach, it is what Congress has long chosen with respect to both statutes. The categorical approach has a particularly “long pedigree in our Nation’s immigration law,” tracing back to 1913. Moncrieffe
, 569 U. S., at 191. As the majority acknowledges, “Congress could have (and sometimes has) used statutory language requiring courts to ask whether the defendant’s actual conduct meets certain specified criteria.” Ante
, at 8, n. 2. But it has not done so in the INA provision here. See ante
, at 8. Thus, here, as in the case of ACCA, a judge must ask whether “a conviction
of the state offense ‘ “necessarily” involved . . . facts equating to’ ” the kind of behavior that the relevant federal statute forbids. Moncrieffe
, 569 U. S., at 190 (emphasis added). Only if it did does that conviction trigger federal penalties.
Now, let us apply the categorical approach to the conviction here at issue. The criminal complaint says that Mr. Pereida “intentionally engage[d] in conduct which . . . constituted a substantial step in a course of conduct intended to culminate in his commission of the crime of CRIMINAL IMPERSONATION R.S. 28–608, Penalty: Class IV Felony.” App. to Brief for Petitioner 7a. It then quotes the entire criminal-impersonation statute, including all of its parts. See id.,
at 7a–8a. The complaint does not say which part of the statutory provision the State accuses Mr. Pereida of violating. And the majority, like the Government, concedes that some of the provisions set forth crimes that are not crimes involving moral turpitude. See ante
, at 10; Brief for Respondent 15.
The journal entry and order related to the charge do not help. They say only that Mr. Pereida pleaded “no contest” to the crime charged, identifying the relevant statute as Neb. Rev. Stat. §28–201 (the attempt provision) and describing the charge as “[a]ttempt of a class 3A or class 4 felo[ny].” App. to Brief for Petitioner 3a. They do not narrow down the possible offenses because all the criminal- impersonation offenses can be a Class III or Class IV felony. See Neb. Rev. Stat. §§28–608(2)(a), (b). We cannot look to jury instructions because there was no jury. Nor is there any plea agreement, plea colloquy, or “comparable judicial record” of the plea that might help determine what Mr. Pereida admitted.
As far as we know, all appropriate documents that exist were before the Immigration Judge. None shows that Mr. Pereida’s conviction necessarily
involved facts equating to a crime involving moral turpitude. He may have pleaded guilty to a crime involving moral turpitude or he may not have. We do not know.
The Immigration Judge thus cannot characterize the conviction as a conviction for a crime involving moral turpitude. That resolves this case.
How does the majority argue to the contrary? The majority says that this case is different because which crime was the basis of a prior conviction is a factual question that the categorical approach cannot answer and a noncitizen seeking cancellation of removal, unlike a criminal defendant, bears the burden of proof on that factual question.
First, the majority says that what the defendant’s “actual
offense of conviction was,” is a “threshold factual” question that a court must resolve before tackling the categorical approach’s “hypothetical
question” (could someone complete the offense of conviction without committing a crime involving moral turpitude). Ante
, at 8–9, 14. In my view, there is no unresolved “threshold factual” question in this case since there is no dispute that Mr. Pereida has a prior conviction. We have made clear that unless the offense of conviction, as determined from the statute and the specified documents, is necessarily
a crime involving moral turpitude, the judge must rule that the conviction was not for a crime involving moral turpitude. The method for determining the offense of conviction (the modified categorical approach) “acts not as an exception, but instead as a tool,” retaining “the categorical approach’s central feature.” Descamps
, 570 U. S., at 263.
Here, looking at the pertinent documents, we can conclude only that Mr. Pereida pleaded guilty to the minimum conduct necessary to complete an offense under Neb. Rev. Stat. §28–608. Thus, the issue is whether someone could complete that
offense without committing a crime involving moral turpitude.
This question is the central question the categorical approach resolves, not a threshold question. And it is a legal question, not a factual one. To answer it, the judge is to examine the state statute and limited portions of the record that our cases specify and determine from those documents whether the crime of conviction was a crime involving moral turpitude. There is nothing at all unusual about referring to a question that a judge must answer based on specified legal documents before him as a “question of law.” To the contrary, construction of written instruments such as deeds, contracts, tariffs, or patent claims “often presents a ‘question solely of law.’ ” Teva Pharmaceuticals USA, Inc.
v. Sandoz, Inc.
574 U.S. 318
, 326 (2015). And legal questions are not affected by a burden of proof. See, e.g., Microsoft Corp.
v. i4i L. P.
564 U.S. 91
, 100, n. 4 (2011).
The majority points out that we have occasionally referred to the “ ‘fact of a prior conviction.’ ” Ante
, at 13. The majority reads too much into that reference. All that we have seriously referred to as a fact is the “mere
fact of conviction.” Taylor
, 495 U. S., at 602 (emphasis added). Establishing that basic fact is, of course, a prerequisite to application of the categorical approach at all. It goes to “the validity of a prior judgment of conviction.” Apprendi
v. New Jersey
530 U.S. 466
, 496 (2000). But the mere fact of conviction is not at issue here. Instead, the question here (and the question the categorical approach asks) is “what [that] conviction necessarily
, 575 U. S., at 806. We have referred to that
question as a “legal question.” Ibid.
And rightly so. Thus, if the majority applies the categorical approach, it should agree that there is no factual dispute in this case for any burden of proof to resolve. If the majority does not apply the categorical approach, it does not explain that or why.
Second, the majority points to statutory language stating that an applicant for relief from removal “has the burden of proof to establish” that he “satisfies the applicable eligibility requirements,” §1229a(c)(4)(A), which includes the requirement that he not have been convicted of a crime involving moral turpitude. See ante
, at 5.
But burdens of proof have nothing to do with this case. As just discussed, because the categorical approach conclusively resolves the ambiguity as to which offense was the basis for the conviction, there is no role for the burden of proof to play. Indeed, the Government agreed at argument that the burden of proof would not apply “if this were just a categorical approach case.” Tr. of Oral Arg. 53. That this case implicates the modified categorical approach rather than the categorical approach does not make a difference. The modified categorical approach, like the categorical approach, provides a conclusive answer without any resort to burdens of proof. It does so not by “treating [a] (divisible) statute as if it states a single offense,” ante
, at 11, n. 4, but by permitting courts to look at only certain conclusive records of a conviction to determine what that conviction necessarily involved.
This conclusion is consistent with the text. The statutory text itself “singl[es] out this lone requirement for special treatment,” ante
, at 6, by using a term (“conviction”) that requires application of a categorical rather than factual analysis. The burden-of-proof provision does not require departing from our settled understanding of the meaning of that term. That the categorical approach applies does not mean that the burden of proof is entirely irrelevant to the requirement that a noncitizen not have a disqualifying prior conviction. The burden of proof may be relevant when “the existence of [a] conviction” is in doubt. See §§1229a(c)(3)(B)(iii), (iv), (vi). Such doubt may have arisen, for example, if Mr. Pereida had contested that a complaint submitted by the Government actually resulted in a conviction or contended that the conviction is against a different Clemente Avelino Pereida. See ante
, at 12. There is no such doubt in this case. No one disputes that Mr. Pereida has a prior conviction. The parties apparently presented the judge with all the existing relevant documentary material of that conviction. This case concerns a different question: Given the fact of Mr. Pereida’s conviction, was it necessarily for a crime involving moral turpitude? The law instructs the judge how to determine, looking at only a limited set of material, whether the crime of conviction is or is not a crime involving moral turpitude. Because of the categorical approach, there is nothing left for a party to prove.
In my view, the “textual clues” and “statutory signals” relied on by the majority further demonstrate that burdens of proof are not relevant to the question at hand. See ante
, at 7, 11, n. 5. As the majority points out, the INA sets forth a list of particular materials that, the INA says, “shall constitute proof of a criminal conviction.” §1229a(c)(3)(B). They include an “official record of judgment and conviction,” an “official record of plea, verdict, and sentence,” a “docket entry from court records that indicates the existence of the conviction,” court minutes of a “transcript . . . in which the court takes notice of the existence of the conviction,” an official “abstract of a record of conviction” that indicates “the charge or section of law violated” (among certain other things), and any other “document or record attesting to the conviction” prepared or kept by the court or by a “penal institution.” Ibid.
The majority also notes that the INA authorizes an immigration judge to make “credibility determination[s]” about a noncitizen’s written and oral proof and determine whether “testimony is credible, is persuasive, and refers to specific facts sufficient to demonstrate that the applicant has satisfied the applicant’s burden of proof.” §§1229a(c)(4)(B), (C). As the majority concedes, this evidence is broader than what we have permitted in our modified categorical approach cases. See ante
, at 16–17.
I agree with the majority that bearing the burden of proof goes hand in hand with being able to introduce this evidence. But in my view, Mr. Pereida cannot introduce this evidence because it goes beyond the limited record our precedents allow. Hence, he must not bear the burden of proof. The majority’s response is that there is no limitation on the documents an immigration judge can look at when applying the categorical approach. That is because, the majority says, the limitation was adopted in the criminal context out of a concern for
Sixth Amendment rights that is not present in the immigration context. Ibid.
That was not, however, our only, or even primary, reason for adopting the limitation. Rather, we limited the documents that a judge can review in order “to implement the object of the statute and avoid evidentiary disputes.” Shepard
, 544 U. S., at 23, n. 4. To be sure, we were there referencing ACCA, not the INA. But the statutes share the relevant object (tying federal penalties to certain convictions, not certain conduct) signaled by the same statutory text (“conviction”). See Taylor
, 495 U. S., at 600; Mellouli
, 575 U. S., at 806. The “central feature” of this statutory object is “a focus on the elements, rather than the facts, of a crime.” Descamps
, 570 U. S., at 263. Allowing review of a broad array of evidence is incompatible with this statutory object, even if the judge looks at the evidence only to determine the nature of the offense of which a noncitizen was convicted. See Shepard
, 544 U. S., at 21–23. I see no reason for the categorical approach to apply differently under the INA than under ACCA given their shared text and purpose. The “ ‘long pedigree’ ” of the categorical approach in our immigration law further counsels against departing from how we have long understood that approach to work. Mellouli
, 575 U. S., at 805–806. Al- though this Court first applied the categorical approach in the criminal context, see ante
, at 7, courts examining the federal immigration statutes concluded that Congress intended a categorical approach decades before Congress even enacted ACCA. See Mellouli
, 575 U. S., at 805–806.
At a minimum, I would not hold, in this case, that the categorical approach’s limitation on the documents a judge can consult is inapplicable in immigration proceedings. That argument was neither raised nor briefed by the parties. The Government confirmed several times at oral argument that it had not argued that a judge should be allowed to look at a broader array of evidentiary materials because, in its view, that issue was not implicated since no other documents exist. See Tr. of Oral Arg. 34, 46, 56. Without the benefit of briefing and argument, we cannot fully anticipate the consequences of today’s decision.
The majority does not
apply the categorical approach as our cases have explained it and used it. So what happens now? I fear today’s decision will result in precisely the practical difficulties and potential unfairness that Congress intended to avoid by adopting a categorical approach.
allowing parties to introduce a wide range of documentary evidence and testimony to establish the crime of conviction may undermine the “judicial and administrative efficiency” that the categorical approach is intended to promote. Moncrieffe
, 569 U. S., at 200.
As we have recognized before, “[a]sking immigration judges in each case to determine the circumstances underlying a state conviction would burden a system in which ‘large numbers of cases [are resolved by] immigration judges and front-line immigration officers, often years after the convictions.’ ” Mellouli
, 575 U. S., at 806 (alterations in original). The same is true here. In cases where noncitizens are able to introduce evidence of their crime of conviction, immigration judges now may have to hear and weigh testimony from, for example, the prosecutor who charged the noncitizen or the court reporter who transcribed the now-lost plea colloquy. Given the vast number of different state misdemeanors, plea agreements made long ago, cursory state records, and state prosecutors or other officials who have imperfect memories or who have long since departed for other places or taken up new occupations, there is a real risk of adding time and complexity to immigration proceedings. Such hearings may add strain to “our Nation’s overburdened immigration courts.” Moncrieffe
, 569 U. S., at 201.
today’s decision may make the administration of immigration law less fair and less predictable. One virtue of the categorical approach is that it “enables aliens ‘to anticipate the immigration consequences of guilty pleas in criminal court,’ and to enter ‘ “safe harbor guilty” pleas [that] do not expose the [alien defendant] to the risk of immigration sanctions.’ ” Mellouli
, 575 U. S., at 806 (alterations in original). The majority’s approach, on the other hand, may “deprive some defendants of the benefits of their negotiated plea deals.” Descamps
, 570 U. S., at 271. A noncitizen may agree to plead guilty to a specific offense in a divisible statute because that offense does not carry adverse immigration consequences. But in many lower criminal courts, misdemeanor convictions are not on the record. See Brief for National Association of Criminal Defense Lawyers et al. as Amici Curiae
7–9 (NACDL Brief ); Brief for United States in Johnson
v. United States
, O. T. 2008, No. 08–6925, p. 43 (“[P]lea colloquies . . . are not always transcribed or otherwise available”). In jurisdictions where misdemeanor convictions are on the record, such records frequently omit key information about the plea and may be destroyed after only a few years. See NACDL Brief 10–16; see also Brief for United States in Voisine
v. United States
, O. T. 2014, No. 14–10154, p. 45 (“[R]ecords from closed misdemeanor cases are often unavailable or incomplete”). And even where complete records do exist, noncitizens, who often are unrepresented, detained, or not fluent English speakers, may not have the resources to offer more than their own testimony. See Brief for Immigrant Defense Project et al. as Amici Curiae
11–19. Thus, under the majority’s approach, noncitizens may lose the benefit of their plea agreements unless their testimony persuades the immigration judge that they pleaded guilty to the lesser offense.
, today’s decision risks hinging noncitizens’ eligibility for relief from removal on the varied charging practices of state prosecutors. In some cases (perhaps even this one), state prosecutors and state courts may treat statutes that list multiple offenses as if they list only one, whether inadvertently or as a matter of practice. See NACDL Brief 13 (explaining that “[a]cross many states and localities, the records of misdemeanor pleas often do not include the statutory subsection or factual basis underlying the conviction”). It sometimes can be challenging to determine whether a fact is an element or a means (and so whether a statute is divisible or not). If a prosecutor mistakes a divisible statute for an indivisible one, she may well not identify which particular offense was the basis for the charge. Some States, including Nebraska, do not require a pleading to identify the alternative means of committing a crime—as opposed to the alternative crimes—on which a conviction is based. See 5 W. LaFave, J. Israel, N. King, & O. Kerr, Criminal Procedure §19.3(a), p. 263 (3d ed. 2007); State
, 265 Neb. 214, 221, 655 N.W.2d 876
, 884 (2003) (“[T]his court has made clear that certain crimes are single crimes that can be proved under different theories, and that because each alternative theory is not a separate crime, the alternative theories do not require that the crime be charged as separate alternative counts”). When a divisible statute is wrongly treated as indivisible, for whatever reason, records will be “inconclusive” because the defendant was not, as a matter of fact, convicted of any particular alternative crime. It would be unfair for mandatory deportation to result from inconclusive records in these cases.
The Court dismisses these “policy” concerns on the ground that Congress has chosen “to conclude that uncertainty about an alien’s prior conviction should not redound to his benefit.” Ante
at 16. But Congress made precisely the opposite choice by tying ineligibility for relief to a noncitizen’s “conviction.” That text mandates a categorical approach in which uncertainty about a conviction redounds to a noncitizen or defendant’s benefit. The approach is underinclusive by design, and the majority’s “objection to th[e categorical approach’s] underinclusive result is little more than an attack on the categorical approach itself.” Moncrieffe
, 569 U. S., at 205.
Finally, it makes particularly little sense to disregard this core feature of the categorical approach here. See id.
, at 203–204. As already noted, cancellation of removal is discretionary. Thus, when a conviction is not disqualifying under the categorical approach, the Government may still deny the noncitizen relief. If it turns out that an individual with a record like the one here in fact violated the statute in a reprehensible manner, that can be accounted for during the discretionary phase of the proceedings, when the categorical approach does not apply.
* * *
In my view, the Court should follow Congress’ statute. Congress has long provided that immigration courts applying the INA provision here, like sentencing courts applying ACCA, must follow the categorical approach. See Mellouli
, 575 U. S., at 805–806. Our cases make clear how that approach applies in a case like this one. We should follow our earlier decisions, particularly Taylor
, and Johnson
. And, were we to do so, ineluctably they would lead us to determine that the statutory offense of which Mr. Pereida was “convicted” is not “necessarily” a “crime involving moral turpitude.”
Because the Court comes to a different conclusion, with respect, I dissent.