Miguel A. ROSILES-CAMARENA, Petitioner, v. Eric H. HOLDER, Jr., Attorney General of the United States, Respondent, 735 F.3d 534

EASTERBROOK, Chief Judge. Miguel Rosiles-Camarena, a citizen of Mexico, was admitted to the United States for permanent residence in 1977, when he was ten years old. He did not use his opportunities to become a citizen. Following his felony conviction for indecent solicitation of a minor, his permanent-residence status was revoked, and he has been ordered removed to Mexico. Rosiles-Camarena is homosexual and HIV positive. He contends that gays are persecuted in Mexico (at least outside of cosmopolitan Mexico City) and that gays infected by HIV face extra risk.

Although he is not eligible for asylum (the deadline for seeking that relief expired long ago), he applied for withholding of removal under 8 U.S.C. § 1231(b)(3), and relief under the Convention Against- Torture, implemented by 8 C.F.R. §§ 1208.16 to .18. To be eligible for either benefit, an alien must show a clear probability that persecution (for withholding of removal) or torture (for the Convention) is more likely than not in the alien’s native country. The immigration judge and the Board of Immigration Appeals disagree about whether Rosiles-Camarena satisfies these requirements.

The IJ initially granted his application for relief under both the statute and the Convention, finding on the basis of statistics and expert testimony that Rosiles-Camarena probably would be killed or injured in Mexico as a result of his sexuality and disease. The BIA remanded, but the IJ adhered to his position on remand. The BIA then reversed and, after a remand (by consent) from this court, adhered to its position.

The most recent decision states that “[t]he probability of future harm is a legal question that we review de novo” and that, “[i]n assessing the probability of harm de novo, we may give different weight to the evidence than did the Immigration Judge.” The BIA proceeded to do just that. It accepted all of the IJ’s findings of historical fact but disagreed with the IJ about the risk implied by those facts. For example: the IJ found that Rosiles-Camarena is at substantial risk because 148 persons were murdered in Mexico, between 1995 and 2006, because of then-sexual orientation.

But the Board observed that this amounts to 12 or 13 killings a year in a population exceeding 110 million, at least 2% of which is homosexual, making it unlikely (a risk of no more than 1 in 100,000) that any given gay man would be killed any given year. Expert testimony establishing that “attacks on homosexuals are frequent” does not show the magnitude of risks, any more than expert testimony that “auto accidents are frequent” would imply that a given driver (even one in a high-risk group, such as men under 25) is more likely than not to be injured.

The Board stated that the IJ did not commit clear error in crediting the statistics and the expert’s testimony but added: “as atrocious as it is to have 12 or 13 such killings per year, that fact does not show a clear probability that [Rosiles-Camarena] will be killed or otherwise persecuted.” The Board treated the risk of future harm as a matter of legislative fact, and it took the view that decisions on mixed (or “ultimate”) questions are open to plenary decision. Rosiles-Camarena contends that the Board made a legal error by engaging in this kind of review.

He also contends that the Board’s decision lacks substantial evidence in the record, but we lack jurisdiction to address that subject. He has been convicted of an aggravated felony, and as a result 8 U.S.C. § 1252(a)(2)(

C) forbids judicial review of the removal decision, except to the extent that the alien presents legal arguments (statutory or constitutional). See 8 U.S.C. § 1252(a)(2)(D). A contention that the agency’s decision is not supported by enough evidence is not a “legal” argument for this purpose. See Jiménez Viracacha v. Mukasey, 518 F.3d 511 (7th Cir.2008); Paez Restrepo v. Holder, 610 F.3d 962 (7th Cir.2010).

Section 1252(a)(2)(

C) applies to applications for relief based on § 1231(b)(3). See Moral-Salazar v. Holder, 708 F.3d 957 (7th Cir.2013). Although Moral-Salazar expresses a reservation for CAT claims, we need not explore in this litigation what sort of arguments under CAT § 1252(a)(2)(

C) allows us to consider. A regulation specifies the extent to which the Board may review or supplement factual decisions by immigration judges. It provides: (

i) The Board will not engage in de novo review of findings of fact determined by an immigration judge.

Facts determined by the immigration judge, including findings as to the credibility of testimony, shall be reviewed only to determine whether the findings of the immigration judge are clearly erroneous. (ii) The Board may review questions of law, discretion, and judgment and all other issues in appeals from decisions of immigration judges de novo. (iv) Except for taking administrative notice of commonly known facts such as current events or the contents of official documents, the Board will not engage in factfinding in the course of deciding appeals.

A party asserting that the Board cannot properly resolve an appeal without further factfinding must file a motion for remand. If further factfinding is needed in a particular case, the Board may remand the proceeding to the immigration judge or, as appropriate, to the Service. 8 C.F.R. § 1003.1(d)(3). An argument that the Board has exceeded the scope of review permissible under this regulation is a legal one, for the purpose of § 1252(a)(2)(D). See Rotinsulu v. Mukasey, 515 F.3d 68, 72 (1st Cir.2008). Matter of V- K-, 24 I. & N.

Dec. 500 (2008), on which the Board relied here, concludes that § 1003.1(d)(3)® does not prevent it from disagreeing with an IJ’s predictions about the likelihood of future harm. V- K- gives two principal reasons. First, clause (ii) authorizes the Board to “review questions of law, discretion, and judgment” — and if the probability of harm is an issue of fact, it is also one of “law” (to the extent the Board must choose “how probable is probable enough?”) and of “judgment” (because evaluating the probability of harm requires the application of judgment to historical facts).

Second, the Board observed that the explanation issued with the adoption of § 1003.1(d)(3) reveals that the resolution of a mixed question of law and fact is not itself a “fact” for the purpose of clause (i). The commentary stated that clause (ii) covers “judgments as to whether the facts established by a particular alien amount to ‘past persecution’ or a ‘well founded fear of persecution.’ ” 67 Fed.Reg. 54,878, 54,890 (Aug. 26, 2002). The first circuit has held that the approach articulated in V- K- is within the Board’s authority. See Rotinsulu, 515 F.3d at 73; Sicaju-Diaz v.

Holder, 663 F.3d 1, 5 (1st Cir.2011). The third circuit, by contrast, set aside V- K- on petition for review. Kaplun v. Attorney General, 602 F.3d 260, 269-71 (3d Cir.2010) (CAT). In En Hui Huang v. Attorney General, 620 F.3d 372, 381-87 (3d Cir.2010), it applied Kaplun to applications for withholding of removal. Four other circuits have agreed with the third. See Hui Lin Huang v. Holder, 677 F.3d 130 (2d Cir.2012); Turkson v. Holder, 667 F.3d 523 (4th Cir.2012); Ridore v. Holder, 696 F.3d 907 (9th Cir.2012); Zhou Hua Zhu v. Attorney General, 703 F.3d 1303 (11th Cir.2013).

The third circuit concluded that the Board is entitled to adopt an independent view about whether a potential harm identified by an IJ amounts to “persecution” or “torture,” but that an IJ’s predictions (which it called the “present probability of a future event”)- — -such that a particular harm is “likely” should an alien return to his native land — are “facts” under clause (i), and the Board’s role is limited to identifying clear error by the IJ. The Board’s decision in our case adds some rationales in the course of explaining why it finds Kaplun and En Hui Huang unpersuasive.

The Board’s principal concern is that its legal views won’t have much significance if all predictions are facts.

It observed that “ ‘predictive’ findings, particularly regarding the level of harm that is likely to be inflicted, may preordain resolution of the legal quéstion regarding whether such harm rises to the level of persecution or torture.” The Board added: “any such predictive findings dre likely to be based on written reports of country conditions over which a trier of fact has no particular expertise or advantage, in contrast to issues of credibility, resolution of conflicting testimony, or questions of historical fact.” In other words, the Board thought that the category of predictions identified by the third circuit often concerns legislative rather than adjudicative facts.

A sound prediction depends on country conditions, not (necessarily) on facts unique to the alien. For example, Rosiles-Camarena has lived in the United States since he was 10 and has visited Mexico only briefly. Many of the IJ’s predictions concern conditions in Mexico. The Board thinks it unacceptable to have one IJ assert that conditions in Mexico are horrible, while another deems them fine, and to have both conclusions immune to effective review. Immigration judges display substantial disparity in evaluating claims for asylum or withholding of removal. See Jaya Ram-ji-Nogales, Andrew I. Schoenholtz & Philip G.

Schrag, Refugee Roulette: Disparities in Asylum Adjudication, 60 Stan. L.Rev. 295 (2007). The Board thinks that it is entitled to curtail IJs’ divergent approaches and believes that it can do so by determining whether particular countries are, or are not, hostile to particular political or social groups. Indeed, we have urged the Board to make categorical decisions. E.g., Banks v. Gonzales, 453 F.3d 449, 453-55 (7th Cir.2006); Xiu Ling Chen v. Gonzales, 489 F.3d 861, 862 (7th Cir.2007); Chun Hua Zheng v. Holder, 666 F.3d 1064, 1068 (7th Cir.2012).

The Board fears that, under Kaplun and similar decisions, every IJ may maintain a personal view about the risks in each nation. Both the Board and the courts of appeals (reviewing it) would have their hands tied. Rosiles-Camarena would benefit from such a limit on review, but aliens whose claims are denied by other IJs would lose; neither the Board nor a court of appeals could conclude that country conditions are more hazardous than the IJ found. The Board has interpreted § 1003.1(d)(3)(

i) in a way that it believes makes clauses (

i) and (ii) harmonious by relying on the longstanding distinctions between adjudicative and legislative facts, and between historical (case-specific) facts and those aspects of discretion or judgment that concern country-wide conditions — subjects on which the Board thinks that the United States should be able to speak with one voice, rather than through a cacophony of immigration judges.

The problem is that the Board’s arguments would be better as reasons to revise the regulation than as reasons to interpret it differently from the similar language that governs relations between federal trial and appellate courts. Appellate courts are bound by Fed. R.Civ.P. 52(a)(6), which says that district judges’ findings of fact must not be set aside unless clearly erroneous. See also Anderson v. Bessemer City, 470 U.S. 564, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985). Rule 52(a)(6) may have been the inspiration for § 1003.1(d)(3)®.

The Rule does not allow plenary appellate review of district judges’ characterizations based on historical facts. A finding on an “ultimate” fact (a conclusion based on the application of legal rules to historical facts) often is treated as one of fact. See, e.g., Pullman-Standard v. Swint, 456 U.S. 273, 102 S.Ct. 1781, 72 L.Ed.2d 66 (1982) (whether the plaintiff is a victim of “discrimination” is a question of fact); Icicle Seafoods, Inc. v. Worthington, 475 U.S. 709, 106 S.Ct. 1527, 89 L.Ed.2d 739 (1986) (whether the plaintiff is a “seaman” is a question of fact).

When a decision is person-specific, there is rarely a good reason for having three judges substitute their views for the assessment of the trial judge. See, e.g., Mars Steel Corp. v. Continental Bank N.A., 880 F.2d 928, 933-34 (7th Cir.1989) (en banc); Mucha v. King, 792 F.2d 602, 605-06 (7th Cir.1986). When a decision depends on the characteristics of non-litigants, however, and establishes a rule with broad scope, the appellate role can be more substantial. For example, in American Needle, Inc. v.

National Football League, 560 U.S. 183, 130 S.Ct. 2201, 176 L.Ed.2d 947 (2010), the Supreme Court gave no weight to a district judge’s conclusion that the National Football League should be characterized as a single firm for the purpose of antitrust law. But even that sort of review, which concerns legal consequences, differs from an entitlement to make an independent appellate decision on facts just because they are “legislative” in nature. Kaplun observed that many predictions are facts, in the sense that they rest on subsidiary facts and can be true or false.

It gave this example: “It is likely that it will take less than 3 hours to drive the 100 miles to grandmother’s house next week.” 602 F.3d at 269. Likewise, a medical prediction about whether a victim of injury-will recover is factual, even though it rests on the application of medical knowledge to subsidiary facts. Id. at 270. These illustrations show how person-specific circumstances (adjudicative facts) can give rise to predictions that also are sensibly treated as facts. That is as true when a prediction depends on country conditions as when it depends on what happened to a particular alien.

We therefore agree with Kaplun and similar decisions. Perhaps the Board’s view that it may make independent decisions about predictions is a consequence of giving itself too little leeway to find IJs’ predictions clearly erroneous. A federal court of appeals would be inclined to think it a clear error, eorrectible under Rule 52, for a district judge to say that a l-in-100,000 chance of death meets a more-likely-than-not burden of persuasion. But that’s not what the Board did. Instead it claimed a right to substitute its judgment for that of the IJ mthout finding a clear error.

That is a mistake under the regulation. A court is limited to the agency’s stated grounds of decision and cannot enforce an order on a basis that the agency did not include among its reasons. Our task, having corrected a legal error, is to remand to the Board rather than make our own decision. See, e.g., Gonzales v. Thomas, 547 U.S. 183, 126 S.Ct. 1613, 164 L.Ed.2d 358 (2006). To say that the regulation leaves the Board free to declare an IJ’s findings clearly erroneous is not at all to say that it would be appropriate for the Board to do so in this case.

For although we have mentioned so far only the statistical risk of death for homosexuals as a group, Rosiles-Camarena contends that he is at greater risk. He is not only gay and HIV positive but also “out” and planning to live openly with his partner. He contends, and the IJ found, that his family has disowned him and will not offer any' support. He adds that, because he has lived in the United States most of his life and does not know contemporary Mexican customs, he will find it hard to avoid attracting attention from persons who might do him harm.

And he stresses that injuries (and deprivations of economic opportunities) short of death may amount to persecution. The question for the Board on remand is thus not whether aggregate data imply that Rosiles-Camarena is likely to be killed, but whether the IJ clearly erred in finding that he is more likely than not to be persecuted. That question is for the Board in the first instance; we do not express an opinion on it. The petition for review is granted, and the matter is remanded to the Board for proceedings consistent with this opinion.