Skip to main content


Reset A A Font size: Print

United States Court of Appeals, Ninth Circuit.

Emilia VELASQUEZ-GASPAR, Petitioner, v. William P. BARR, Attorney General, Respondent.

No. 17-71964

Decided: September 30, 2020

Before: Richard A. Paez, Consuelo M. Callahan, and Lawrence VanDyke, Circuit Judges. Garish Sarin (argued), Law Offices of Garish Sarin, Los Angeles, California, for Petitioner. Rosanne Perry (argued) and Nelle M. Seymour, Trial Attorneys; Claire L. Workman, Senior Litigation Counsel; Joseph H. Hunt, Assistant Attorney General; Office of Immigration Litigation, Civil Division, United States Department of Justice, Washington, D.C., Washington, D.C.; for Respondent.


Emilia Velasquez-Gaspar, a Guatemalan native and citizen, petitions for review of the Board of Immigration Appeals’ (BIA) dismissal of her appeal of an immigration judge's (IJ) denial of her applications for asylum, withholding of removal, and protection under the Convention Against Torture (CAT). We have jurisdiction under 8 U.S.C. § 1252 and deny the petition.


Velasquez-Gaspar unlawfully entered the United States near Naco, Arizona, in 2010. Later that year the U.S. Department of Homeland Security initiated removal proceedings against her, charging that she was present in the United States without being admitted or paroled. See 8 U.S.C. § 212(a)(6)(A)(i)(I). Velasquez-Gaspar conceded her removability but applied for asylum, withholding of removal, and CAT relief. As the basis for her applications, she claimed that, if deported, she would face persecution and torture at the hands of her abusive ex-boyfriend, Brian Alexander Gonzales.

Velasquez-Gaspar testified at her removal hearing that Gonzales had repeatedly beaten her and that, one night, he and his friends had raped her. She recounted that when she threatened to report Gonzales to the police, he stated that he would kill her if she did so. She added that the police would not have believed her in any event, considering that she was an indigenous woman, and that the police were susceptible to bribes. The Guatemalan government discriminated against indigenous people, she explained, and lacked laws addressing domestic violence.

Velasquez-Gaspar supplemented her testimony with several articles and U.S. Department of State human rights reports for Guatemala, which detailed the country's handling of domestic abuse, rape, and femicide. She also offered written statements from a former employer and neighbor, who were aware of her situation and had urged her to seek help from the police. Fearing retaliation from Gonzales, Velasquez-Gaspar declined this advice.

The IJ denied Velasquez-Gaspar relief on several alternative grounds. She first found Velasquez-Gaspar incredible based on inconsistencies in her testimony. She next determined that Velasquez-Gaspar's proposed social group of “women unable to leave their relationships” was not cognizable. And finally, she found that Velasquez-Gaspar had not established the Guatemalan government's inability or unwillingness to protect her from Gonzales. Velasquez-Gaspar appealed to the BIA, and a divided panel dismissed her appeal. The panel assumed arguendo that she was credible and did not address her proposed social group. It agreed, however, with the IJ that Velasquez-Gaspar failed to establish that governmental authorities would not or could not have protected her. Velasquez-Gaspar timely petitioned for our review of that narrow question.


“We review only the BIA's opinion, except to the extent that it expressly adopted portions of the IJ's decision.” Rayamajhi v. Whitaker, 912 F.3d 1241, 1243 (9th Cir. 2019) (citation omitted). We review the agency's factual findings under the “extremely deferential” substantial-evidence standard, under which we treat such findings as “conclusive unless any reasonable adjudicator would be compelled to conclude to the contrary.” Farah v. Ashcroft, 348 F.3d 1153, 1156 (9th Cir. 2003) (quoting 8 U.S.C. § 1252(b)(4)(B)).


An alien seeking asylum must demonstrate that she is “unable or unwilling” to return to her home country “because of a well-founded fear of persecution on account of race, religion, nationality, membership in a particular social group, or political opinion.” Bringas-Rodriguez, 850 F.3d 1051, 1062 (9th Cir. 2017) (quoting Navas v. INS, 217 F.3d 646, 654 (9th Cir. 2000)). Proving past persecution can satisfy this burden, as it gives rise to a rebuttable presumption of future persecution. Id. But to qualify an alien for asylum, the persecution must have been “committed by the government” or, as relevant here, “by forces that the government was unable or unwilling to control.” Id.; see also Rahimzadeh v. Holder, 613 F.3d 916, 920 (9th Cir. 2010).

We conclude that substantial evidence supports the agency's determination that, had Velasquez-Gaspar reported her abuse,1 the Guatemalan government could have protected her from Gonzales. The State Department reports show that Guatemala is working to curb violence against women. For example, Guatemalan law criminalizes rape and domestic abuse, and officials investigate and prosecute cases under those laws. Granted, the conviction rates are exceptionally low, and officers often face a lack of resources and training, but we do not limit our analysis to whether the government can “control the attackers”; we also look to whether it can “protect the attacked.” Bringas-Rodriguez, 850 F.3d at 1066. On this point the reports convey that justices of the peace issued restraining orders and ordered police protection for abuse victims in an unspecified number of cases. In addition, Guatemala has established programs, offices, and shelters for female victims of physical and sexual assault. One such shelter operates in Quetzaltenango, where Velasquez-Gaspar lived. Thus, although the State Department reports make clear that Guatemala still has a long way to go in addressing domestic violence, the country's efforts, coupled with the pleas of Velasquez-Gaspar's acquaintances, suggest that she could have obtained help.2 As a result, we cannot say that the record compels a finding contrary to the agency's.

Indeed, we denied an analogous petition in Castro-Perez v. Gonzales, 409 F.3d 1069 (9th Cir. 2005). There, the petitioner credibly testified that she did not report being raped because the police would not have protected her and because she feared violent reprisal from her father. Id. at 1072. Looking to the State Department report for Honduras, we acknowledged the country's widespread domestic violence “despite attempts to strengthen domestic abuse law.” Id. We nonetheless held that a reasonable trier of fact would not be “compelled to find that the Honduran government must bear some responsibility” for the abuse. Id. In the end, the country-report evidence “did not conclusively show that the Honduran government would have ignored the report of rape.” Bringas-Rodriguez, 850 F.3d at 1066 n.9 (discussing Castro-Perez).

As in Castro-Perez, the evidence here falls short of compelling the conclusion that Guatemalan authorities would have been unable or unwilling to help Velasquez-Gaspar.3 See Lianhua Jiang v. Holder, 754 F.3d 733, 738 (9th Cir. 2014) (“[T]he petitioner must establish that the evidence not only supports that conclusion, but compels it.” (quotation marks and citation omitted)). Accordingly, she has failed to demonstrate her eligibility for asylum. And because withholding of removal also turns on this factor, substantial evidence likewise supports the agency's denial of that claim. Castro-Perez, 409 F.3d at 1072. Finally, Velasquez-Gaspar waived any argument as to her CAT claim by failing to “specifically and distinctly” discuss the matter in her opening brief. See id. (quoting Arpin v. Santa Clara Valley Transp. Agency, 261 F.3d 912, 919 (9th Cir. 2001)); Lopez-Vasquez v. Holder, 706 F.3d 1072, 1079–80 (9th Cir. 2013). Such an argument would have failed in any event, as she has not shown a likelihood of torture by or with the acquiescence of public officials. See Bromfield v. Mukasey, 543 F.3d 1071, 1079 (9th Cir. 2008). We therefore deny her petition.


I join Judge Callahan's opinion in full. Given the considerable deference this court owes to the BIA's weighing of the evidence, the evidence here certainly does not “compel[ ]” reversal. I.N.S. v. Elias-Zacarias, 502 U.S. 478, 481 n.1, 112 S.Ct. 812, 117 L.Ed.2d 38 (1992) (“To reverse the BIA finding we must find that the evidence not only supports that conclusion, but compels it ․”); 8 U.S.C. § 1252(b)(4). Looking to the 2014 Country Report, the BIA identified Guatemalan progress in protecting domestic violence victims.1 The BIA also weighed the uncontested relevant, but non-dispositive, fact that Petitioner never sought police assistance. Citing to and affirming the Immigration Judge's findings on both counts, the BIA reasonably concluded that Petitioner failed to carry her burden of proof.

Notwithstanding slower-than-preferred progress in the 2014 Country Report, the BIA's conclusion is clearly supported by “reasonable, substantial, and probative evidence on the record considered as a whole.” Elias-Zacarias, 502 U.S. at 481, 112 S.Ct. 812. The BIA and IJ properly weighed the mixed evidence within the Country Reports, and the court should not second-guess that appraisal unless the evidence compels it. Aden v. Holder, 589 F.3d 1040, 1046 (9th Cir. 2009) (“Our standard of review, though, does not enable us to substitute our judgment ․ for the BIA's” when it comes to reasonably weighing the persuasiveness of administrative record evidence.). “We have repeatedly recognized that the IJ and the BIA are entitled to rely on country reports that contain mixed messages, ambiguities, or inconsistencies.” Singh v. Holder, 753 F.3d 826, 831 (9th Cir. 2014) (“[S]low and uneven progress in prosecuting those responsible for past persecution is troubling. But it does not mean that the Board's decision is not supported by substantial evidence.”). It is of course true, as the dissent notes, that “deference does not mean blindness.” Li v. Ashcroft, 356 F.3d 1153, 1158 (9th Cir. 2004) (en banc). But deference does mean deference, and Li’s truism cannot empower this court to substitute its judgment for that of the BIA's. Deference only counts when judges uphold a decision they disagree with. And agree with it or not, Castro-Perez v. Gonzales, 409 F.3d 1069 (9th Cir. 2005), is directly applicable and controls this case.

I write separately to respond to the dissent. It argues that Castro-Perez “provides little guidance” in deciding this case, but ultimately offers only one basis to distinguish it.2 The dissent relies extensively on Velasquez-Gaspar's own testimony as supplying “concrete evidence that Castro-Perez did not.” Only by pointing to Velasquez-Gaspar's testimony can the dissent attempt to distinguish Castro-Perez to conclude that “the Guatemalan government was unable to protect Velasquez-Gaspar ․”

But the Immigration Judge in this case expressly found that Velasquez-Gaspar was “not a credible witness.” The BIA never rejected that factual finding nor concluded it was clearly erroneous. Therefore, the IJ's adverse credibility finding is part of the “record evidence” that the dissent insists “compels” a different conclusion than that reached in Castro-Perez. The dissent never explains how non-credible testimony can do that. Nor does it attempt to show that the IJ's credibility determination was clearly erroneous. Rather, it simply treats Velasquez-Gaspar's testimony as wholly credible, literally ignoring the fact-finder's contrary conclusion.

I think that is wrong—obviously as a matter of statutory law and the Supreme Court's caselaw; perhaps less obviously as a matter of our circuit's confused statements. So I write to clarify what the law requires of this court when reviewing an administrative record containing an IJ's unrebutted adverse credibility finding.

The REAL ID Act significantly altered testimonial credibility standards in immigration cases. See generally Shrestha v. Holder, 590 F.3d 1034, 1039–45 (9th Cir. 2010). One example is the language in 8 U.S.C. § 1158(b)(1)(B)(iii): “There is no presumption of credibility, however, if no adverse credibility determination is explicitly made, the applicant or witness shall have a rebuttable presumption of credibility on appeal.” As Dai v. Sessions points out, the BIA—and not the Court of Appeals—hears an alien's appeal from an adverse ruling by an Immigration Judge. 884 F.3d 858, 868–69 (2018). So if the IJ makes no adverse credibility determination, the BIA presumes the testimony is credible; or the BIA can make its own adverse credibility determination if it concludes the “totality of the circumstances” rebut the presumption. 8 U.S.C. § 1158(b)(1)(B)(iii). That's simple enough.

The question is, what do we do when the IJ makes an explicit adverse credibility determination, but the BIA affirms the IJ's conclusion without directly addressing the IJ's adverse credibility determination? In other words, what do we do in a case like this?

I think the rule is straightforward and apparent in our “substantial evidence” standard of review. Our review of the agency's findings is “extremely deferential.” Farah v. Ashcroft, 348 F.3d 1153, 1156 (9th Cir. 2003) (citing 8 U.S.C. § 1252(b)(4)(B)). Indeed, “[t]he BIA's determination [that an alien is ineligible for asylum] must be upheld if supported by reasonable, substantial, and probative evidence on the record considered as a whole.” Elias-Zacarias, 502 U.S. at 481, 112 S.Ct. 812 (internal quotation marks omitted & emphasis added). Because we are reviewing an administrative decision—and not a district court decision—“we must decide whether to grant or deny the petition for review based on the Board's or IJ's reasoning rather than our own independent analysis of the record.” Ali v. Holder, 637 F.3d 1025, 1029 (9th Cir. 2011) (alteration marks and internal quotation marks omitted) (quoting Azanor v. Ashcroft, 364 F.3d 1013, 1021 (9th Cir. 2004)).

“That does not mean, however, that our review may consider only evidence expressly identified in the BIA's decision.” Ramirez-Villalpando v. Holder, 645 F.3d 1035, 1039 (9th Cir. 2011). Rather, while our review is limited “only [to] the administrative record,” 8 U.S.C. § 1252(b)(4)(A), and “we consider only the grounds relied upon by” the BIA, Andia v. Ashcroft, 359 F.3d 1181, 1184 (9th Cir. 2004) (per curiam), we must determine whether the record evidence “considered as a whole”—as the BIA encountered it—“compels” us to reverse the BIA's decision. Elias-Zacarias, 502 U.S. at 481 & n.1, 112 S.Ct. 812; Melkonian v. Ashcroft, 320 F.3d 1061, 1065 (9th Cir. 2003). Properly understood, our role is to review the basis for the BIA's decision by meeting all the relevant evidence as the BIA found it and then decide whether a “reasonable BIA” would be “compelled” to reach a different conclusion.

Under our case law then, “grounds” and “evidence” are two different things.3 Accordingly, we may affirm the BIA on its decisional grounds by looking to the entirety of the evidence—including record evidence it may not have expressly relied on. Ramirez-Villalpando, 645 F.3d at 1039. This principle has a logical inverse corollary: we may not reverse the BIA by ignoring or recasting the evidence that was before it.

When the BIA affirms an IJ's denial of relief but does not explicitly address the IJ's adverse credibility finding, the BIA is not tacitly reversing that finding; on the contrary, it remains a “conclusive” finding in the administrative record. 8 U.S.C. § 1252(b)(4)(B); 8 C.F.R. § 1003.1(d)(3)(i). That same BIA could affirm the IJ's conclusion that an alien failed to meet her evidentiary burdens while “[a]ssuming arguendo that the [alien] testified credibly.” But it defies logic (and the governing statute and regulation) to assume that this banal method of appellate decision-making is tantamount to the BIA's repudiation of the IJ's explicit adverse credibility finding or constitutes a positive credibility finding. In other words, the evidentiary record remains just where the IJ left it, regardless of whether the BIA relies upon each and every IJ fact finding to reach its conclusion.4

Momentarily setting aside the credibility issue that has bedeviled our immigration jurisprudence, an illustration proves the point. Assume, for example, that an IJ properly makes five findings and relies upon them all in denying an asylum request on a single ground. On appeal, assume the BIA affirms on the same ground as the IJ, but only relies upon two of the five findings without discussing the other three. While this court must limit its review to the ground of the BIA's decision, it is not required to consider only the two evidentiary findings that the BIA considered. Nor can this court on review simply assume the BIA reversed those three unaddressed findings. Certainly, this court on review cannot construe the facts to be diametrically opposite of the IJ's undisturbed findings. To the contrary, this court is statutorily obligated to consider all five conclusive findings, together with the rest of the “administrative record,” to determine whether “the evidence ․ considered as a whole ․ compels” the conclusion that the BIA's ground of decision was wrong. Elias-Zacarias, 502 U.S. at 481 & n.1, 112 S.Ct. 812 (emphasis in original); 8 U.S.C. § 1252(b)(4)(A)–(B). Practically then, if the court concludes that the two findings relied upon are themselves insufficient to support the BIA's conclusion, this court must still uphold the BIA's decision if the other three findings support the BIA's decision by “substantial evidence.”

This straightforward method of review does not somehow mystically change just because part of the evidence before the BIA is the IJ's adverse credibility finding. The dissent argues—with nary a single citation in support—that “the IJ's credibility determination is not itself record evidence.” It is a factual finding that is part of the evidentiary record before the BIA and us.5 “The same standard applies to the IJ's credibility findings” as to the IJ's other factual findings. Farah, 348 F.3d at 1156; 8 C.F.R. § 1003.1(d)(3)(i) (“Facts determined by the immigration judge, including findings as to the credibility of testimony, shall be reviewed only to determine whether ․ clearly erroneous.”) (emphasis added).

Accordingly, where the BIA has not expressly reversed the IJ's explicit adverse credibility finding, we, who encounter the evidence just as the BIA did, must consider that finding as a fixed feature of the record. 8 U.S.C. § 1252(b)(1)(4)(A).6 In determining whether “the evidence on the record considered as a whole ․ compels” a different conclusion than that reached by the BIA, we cannot ignore the part of that evidence that included the fact-finder's adverse credibility finding. Elias-Zacarias, 502 U.S. at 481 & n.1, 112 S.Ct. 812 (emphasis in original). To conclude otherwise is to ignore the Supreme Court's and Congress's clear admonition to accept the BIA's conclusion unless all of the evidence before the BIA—including credibility findings—“compels” a different conclusion than the BIA. Id.; 8 U.S.C. § 1252(b)(4). Ignoring the fact-finder's adverse credibility finding attributes an energetic potency to the BIA's silence that finds no basis in the statutory text, forces this court to blink reality by counter-factually rearranging the deck chairs of the evidentiary record, and jettisons the deference we owe the fact-finder under the REAL ID Act. 8 U.S.C. § 1252(b)(1)(4)(B) (“the administrative findings of fact are conclusive unless any reasonable adjudicator would be compelled to conclude to the contrary”).

This method of review is not inconsistent with the “distinct rule” this court relied on in Dai “that in the absence of an adverse credibility finding by either the IJ or the BIA, we are required to treat the petitioner's testimony as credible.” 884 F.3d at 869 n.8. There, this court credited the alien's testimony “[b]ecause neither the IJ nor the BIA made an adverse credibility determination ․” Id. at 870 (emphasis added);7 see also Ornelas-Chavez v. Gonzales, 458 F.3d 1052, 1054 n.2 (9th Cir. 2006) (same); Zheng v. Ashcroft, 332 F.3d 1186, 1189 n.4 (9th Cir. 2003) (same). Of course, that is a very different situation than the one presented in this case, where the IJ did clearly make an adverse credibility finding supported by a lengthy analysis, and the BIA never rejected that finding as “clearly erroneous.” In a case like Dai, there is no adverse credibility finding in the record that this court must consider when deciding whether the whole record “compels” reversal of the BIA. Elias-Zacarias, 502 U.S. at 481 & n.1, 112 S.Ct. 812.8

But in a case like this, there is. This court is ill-suited to make credibility findings in the first instance—adverse or otherwise. Shrestha, 590 F.3d at 1041 (the deference required by the REAL ID Act “makes sense because IJs are in the best position to assess demeanor and other credibility cues that we cannot access on review”). And where such findings are part of the record, we cannot presume to ignore or reverse them while pretending to properly review the “record ․ as a whole.” Elias-Zacarias, 502 U.S. at 481, 112 S.Ct. 812.

Beyond cases like Dai where the fact-finder made no adverse credibility finding, I am aware of two post-REAL ID Act cases from this court that ostensibly suggest we take a remarkable evidentiary tack: that we tacitly reverse an IJ's adverse credibility finding when reviewing the record as a whole, unless the BIA has explicitly adopted that finding. See Zhiqiang Hu v. Holder, 652 F.3d 1011 (9th Cir. 2011); Singh v. Holder, 753 F.3d 826 (9th Cir. 2014). This suggestion runs directly counter to the REAL ID Act, itself; moreover, the statement in both cases was unnecessary and non-binding.

In Hu, the court determined “the IJ's adverse credibility finding is not supported by substantial evidence.” 652 F.3d at 1016 n.4. A fact-finder's adverse credibility finding unsupported by substantial evidence is reversable error, and thus cannot properly be relied on by this court as part of the evidentiary record “as a whole,” regardless of what the BIA did with it. Therefore, Hu’s statement—that “[t]he BIA's decision is silent on the issue of credibility, despite the IJ's explicit adverse credibility finding, so we may assume that the BIA found Hu to be credible”—is, in addition to being an obvious logical fallacy, pure dictum. That statement made no difference to the case's disposition. 652 F.3d at 1016. To support this dictum, Hu cites a pre-REAL ID Act case, Krotova v. Gonzales, 416 F.3d 1080, 1084 (9th Cir. 2005). Krotova says that “[w]hen the BIA's decision is silent on the issue of credibility, despite an IJ's explicit adverse credibility finding, we may presume that the BIA found the petitioner to be credible.” Id. But just like in Hu, Krotova’s statement too is dictum; in its very next sentence, Krotova declared: “the IJ's adverse credibility finding is not supported by substantial evidence.” Id.9

Singh similarly recited the Krotova dictum but did not rely upon it. 753 F.3d at 828, 835. Like Hu and the present case, the IJ in Singh determined the alien's testimony was not credible, and the BIA did not expressly adopt that finding. Id. at 828. Yet both the IJ and the BIA weighed the alien's testimony against other record evidence and denied relief. Id. at 830, 83–37. On one hand, the panel in Singh purportedly credited the testimony (pursuant to Krotova), but on the other, affirmed the BIA's and IJ's decision to weigh and discount that testimony, which was contradicted by compelling record evidence. Id. at 835–36. Ultimately, Singh merely reaffirmed the agency's right to weigh conflicting evidence. Id. at 837. According to the court, the IJ weighed the very testimony it found non-credible and reached a conclusion supported by substantial evidence. Id. at 835. Like the other cases that purport to embrace the Krotova dictum, it made no difference to Singh’s outcome. Id. at 837.

Admittedly, that is a lot of unnecessary recitation of dicta. One might suppose that if dicta is repeated enough, even as dicta, it eventually becomes law. But that isn't the rule. See In re Magnacom Wireless, LLC, 503 F.3d 984, 993–94 (9th Cir. 2007) (“In our circuit, statements made in passing, without analysis, are not binding precedent.”); see also Estate of Magnin v. C.I.R., 184 F.3d 1074, 1077 (9th Cir. 1999) (“When a case assumes a point without discussion, the case does not bind future panels.”).

That is a good thing—at least in this case—because the Krotova dictum, if it ever had merit, clearly has none in the post-REAL ID Act world. “[D]icta distilled from dicta,” Poventud v. City of New York, 715 F.3d 57 (2d Cir. 2013) (Jacobs, C.J., dissenting), even copiously imbibed, remains too low-proof a spirit to wheedle our court into ignoring clear statutory and Supreme Court directives.

I've already noted how such a rule would not be consistent with 8 U.S.C. § 1252(b)(4) and the Supreme Court's clear command that we must not reverse the BIA unless the evidentiary record “considered as a whole ․ compels it.” Elias-Zacarias, 502 U.S. at 481 & n.1, 112 S.Ct. 812. Moreover, it is obvious that actual application of the Krotova dictum in a case where it mattered would undermine either the BIA's proper role on appeal or ours on review, or both. The BIA's “Scope of Review” regulations bear this out:

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.

8 C.F.R. § 1003.1(d)(3)(i) (emphasis added). In this case, the BIA inherited the IJ's explicit adverse credibility finding, and never found it “clearly erroneous.” That adverse credibility finding thus remains part of the record, even though the BIA ruled in reliance on evidence that did not hinge upon the non-credibility of the petitioner's testimony. Likewise, that finding remains part of the record this court inherits on review. “[T]he court of appeals shall decide the petition only on the administrative record on which the order of removal is based, [and] the administrative findings of fact are conclusive unless any reasonable adjudicator would be compelled to conclude to the contrary.” 8 U.S.C. § 1252(b)(1)(4)(A)–(B).

It follows that when this court sua sponte “assumes” or “presumes” the truthfulness of testimony the IJ has explicitly found non-credible (which the BIA has not found “clearly erroneous”), we flagrantly disregard 8 U.S.C. § 1252 and 8 C.F.R. § 1003.1. That is precisely what the dissent has done here. Velasquez-Gaspar's testimony, according to the dissent, is the dispositive factor that pushes her petition over the evidentiary hurdle and distinguishes this case from Castro-Perez.10 But we may not revive and credit noncredible testimony by presumptive fiat.

The dissent's willingness to do so here ignores the law by permitting the court to rewrite conclusive findings of the administrative record through the power of presumption rather than substantial evidence review. But the only presumption we may legitimately draw here is that the law means what it says. This court does not deem testimony credible, for that would be doing what even the BIA may not: “engag[ing] in de novo review of findings of fact.” 8 C.F.R. § 1003.1(d)(3)(i). Rather, the court reviews the BIA's decision in the context of the entire record as the BIA received it, adverse credibility findings and all.

Although our standard of review is deferential, as the majority states, “deference does not mean blindness.” Li v. Ashcroft, 356 F.3d 1153, 1158 (9th Cir. 2004) (en banc). Velasquez-Gaspar presented unrebutted evidence that the Guatemalan government systematically fails to hold perpetrators of gender-based violence accountable for their crimes. Because this evidence and her personal experiences compel the conclusion that the Guatemalan government was unable or unwilling to protect her from further abuse, I would grant the petition for review and remand for further proceedings.

The agency relied principally on the 2014 State Department Country Report (“2014 Report”) to find that the Guatemalan government is able and willing to protect victims of domestic violence. The 2014 Report found that Guatemalan law authorizes prison sentences for perpetrators of gender-based violence and that the government secured a number of convictions under those statutes.

In principle, Guatemalan law criminalizes rape and domestic violence. In practice, it does not. Contrary to the agency's interpretation of the 2014 Report, the State Department was clear on this point: it noted that police have “minimal training or capacity to investigate sexual crimes or assist survivors of such crimes, and the government did not enforce the law effectively.” This systemic failure has created an environment in which victims “frequently d[o] not report crimes due to lack of confidence in the justice system, social stigma, and fear of reprisal.”

The agency, in my view, mischaracterized the contents of the 2014 Report. Specifically, as evidence that “the Guatemalan government does investigate and prosecute domestic violence crimes,” the IJ noted that the government had secured “976 convictions in cases of intrafamily violence against women and children as of the end of September [of 2014].”

The IJ failed to note, however, that the 2014 Report also indicated that the government received 48,132 official reports of these crimes over the same time period. Assuming that each survivor of domestic violence reported their abuser to the police, these statistics indicate that ninety-eight percent of abusers faced no consequences for their crimes, even when their victims went to the police for help. The IJ likewise observed that the Guatemalan government successfully prosecuted 304 sexual assault cases. But again, the IJ failed to include context for this number: during the same period of analysis, the government received 8,871 reports of these crimes, indicating that over ninety-six percent of reported sexual assaults went unpunished. The same page of the 2014 Report detailing these statistics also notes that homicides of women resulted in convictions “1 to 2 percent” of the time.

Ultimately, this evidence compels the conclusion that the Guatemalan government is utterly unable to protect the overwhelming majority of women from men like Gonzalez. To be sure, the BIA also noted that the Guatemalan government has created specialized law-enforcement units and programs to address the violence against women. But these efforts suggest only that the Guatemalan government is willing to address these crimes. The hard evidence compels the conclusion that it is unable to do so.

The majority's reliance on Castro-Perez v. Gonzales is misplaced. 409 F.3d 1069 (9th Cir. 2005). In Castro-Perez, the applicant—a survivor of rape at the hands of her boyfriend—testified that she did not report the attack to the Honduran police because she did not think they would do anything to help her. Id. at 1072. She further feared that her father would beat her if she reported her boyfriend to the police. Id. We held that the BIA did not err in finding that the petitioner failed to meet the unable-or-unwilling element of proving past persecution. Id.

In so holding, we specifically noted that the country report in that case was “not particularly enlightening.” Id. Indeed, the report did not include any information on the efficacy of Honduran law enforcement; it merely noted that Honduran law criminalized rape. Id. Because the report did not suggest that the Honduran government failed to enforce the law, Castro-Perez could not meet her burden of proof. Id.

Velasquez-Gaspar presented the concrete evidence that Castro-Perez did not. In contrast to the Honduran Country Report in Castro-Perez, the 2014 Report for Guatemala establishes that the government fails to protect women like Velasquez-Gaspar. And Velasquez-Gaspar supplemented this evidence with her own testimony (assumed credible by the BIA) about her abuse and why she did not report it to the police—she feared specific, targeted threats of reprisal from her abuser. Velasquez-Gaspar further testified that she believed the police would likely be of little help given their susceptibility to bribes and their prejudice against indigenous people like herself. For these reasons, Castro-Perez provides little guidance.

Bringas-Rodriguez v. Sessions instead controls. 850 F.3d 1051, 1072 (9th Cir. 2017) (en banc). In that case, we made clear that Castro-Perez did not establish a brightline rule applicable to all or even most survivors of rape at the hands of an intimate partner. 850 F.3d at 1066 n.9. We explained that because Castro-Perez did not offer any concrete evidence that the Honduran government failed to enforce its rape laws, her “reasons for not reporting ․ [were] insufficient to establish governmental inability or unwillingness to protect her.” Id. (citing Ornelas-Chavez v. Gonzales, 458 F.3d 1052, 1057–58 (9th Cir. 2006)). Bringas-Rodriguez stands for the principle that on-the-ground conditions—not official policies or programs—are what matter. Id. at 1072 (declining to afford weight to Mexico's “laud[able] ․ efforts” to protect gay and lesbian people because “it is well recognized that a country's laws are not always reflective of actual country conditions”). The BIA thus erred when it “falsely equated legislative and executive enactments prohibiting persecution with on-the-ground progress.” Id. at 1075. Just so here.1

The concurrence makes much of the fact that the IJ found Velasquez-Gaspar not credible. But her credibility is not properly before us. When the BIA issues its own decision instead of simply adopting the IJ's decision, “our review ‘is limited to the BIA's decision, except to the extent the IJ's opinion is expressly adopted.’ ” Hosseini v. Gonzales, 471 F.3d 953, 957 (9th Cir. 2006) (quoting Cordon-Garcia v. INS, 204 F.3d 985, 990 (9th Cir. 2000)). We “cannot affirm the BIA on a ground upon which it did not rely.” Arrey v. Barr, 916 F.3d 1149, 1157 (9th Cir. 2019) (quoting Navas v. I.N.S., 217 F.3d 646, 658 n.16 (9th Cir. 2000)).

Although Velasquez-Gaspar challenged the IJ's adverse credibility finding on appeal, the BIA did not address her argument. Instead, the BIA explicitly assumed that Velasquez-Gaspar testified credibly. Accordingly, we review the BIA's decision based on its assumption that Velasquez-Gaspar was a credible witness.

The concurrence errs in classifying the IJ's credibility finding as part of the “evidentiary record before the BIA” which we must consider in deciding the petition for review. Concurrence 1068. But the IJ's credibility determination is not itself record evidence. Rather, it reflects the weight the IJ ascribed to certain record evidence—Velasquez-Gaspar's testimony. And it is subject to clear-error review by the BIA. Ridore v. Holder, 696 F.3d 907, 911 (9th Cir. 2012); 8 C.F.R. § 1003.1(d)(3)(i).

In my view, the record evidence as a whole compels the conclusion that the Guatemalan government was unable to protect Velasquez-Gaspar from Gonzalez. Parada v. Sessions, 902 F.3d 901, 908–09 (9th Cir. 2018) (quoting Afriyie v. Holder, 613 F.3d 924, 931 (9th Cir. 2010)). The evidence relied upon by the agency in concluding otherwise, when fully considered, actually shows that almost all perpetrators of gender-based violence in Guatemala carry out their crimes undisturbed by law enforcement, even when their victims reach out for help. Accordingly, substantial evidence does not support the BIA's decision dismissing Velasquez-Gaspar's appeal. I respectfully dissent.


1.   Because she did not report Gonzales's abuse, Velasquez-Gaspar needed to show that doing so would have been futile or dangerous. Ornelas-Chavez v. Gonzales, 458 F.3d 1052, 1058 (9th Cir. 2006). And “[w]hile private threats may explain an applicant's reluctance to go to the authorities, the question in an asylum case is whether the police could and would provide protection.” Rahimzadeh, 613 F.3d at 923.

2.   We note that, during her testimony, Velasquez-Gaspar could not recall any particular instance of the police ignoring women or taking bribes.

3.   Considering that substantial evidence supports the agency's findings even crediting Velasquez-Gaspar's testimony, we need not consider the IJ's adverse credibility finding.

1.   Importantly, neither the IJ nor the BIA ignored Guatemala's lingering problems in preventing and punishing domestic violence, but rather weighed those difficulties against the Report's indications of improvement.

2.   The dissent suggests that both Velasquez-Gaspar's testimony and the 2014 Country Report distinguish this case from Castro-Perez. But there is no meaningful difference between the Country Report in Castro-Perez and the one the BIA considered here. Castro-Perez v. Gonzales, 409 F.3d at 1072 (“[D]omestic violence is widespread despite attempts to strengthen domestic abuse law.”) (discussing U.S. Dep't of State's Country Reports on Human Rights Practices for 2000, Honduras (“2000 Honduras Report”)). Both Reports reveal government efforts and de jure reforms intended to protect women from violence. And both reveal that public funding, training, and other factors have metered progress toward that goal and failed to remediate the problem as quickly as desired. See 2000 Honduras Report (“In March the Public Ministry reported that it receives an average of 200 allegations of domestic violence each month, but that many cases remain pending because the Government has not yet created the special courts authorized by the Law Against Domestic Violence.”). The two Reports paint very similar pictures, so ultimately the dissent is left with just Velasquez-Gaspar's testimony.

3.   The dissent simply ignores this legal and logical distinction, arguing that, like the BIA, we “must assume that Velasquez-Gaspar testified credibly” because we “ ‘cannot affirm the BIA on a ground on which [the BIA] did not rely.’ ” (Emphasis added and citation omitted.) But the latter has nothing to do with the former. Concluding that Velasquez-Gaspar's noncredible testimony does not undermine the BIA's stated ground of its decision is hardly upholding the BIA's decision on a different ground. It is classic substantial evidence review, where our court reviews the entire evidentiary record—including the IJ's adverse credibility finding—and determines whether that entire evidentiary record compels the conclusion that the BIA's ground of decision was wrong.

4.   The BIA may only repudiate the IJ's findings if clearly erroneous. 8 C.F.R. § 1003.1(d)(3)(i). It follows that we may not interpret the BIA's silence as a reversal of the IJ's findings.

5.   The dissent's attempt to paint an IJ's credibility determination as something that we can just ignore in our substantial evidence review (but only, of course, if the IJ made an adverse credibility finding) is particularly odd since the dissent acknowledges, literally one sentence later, that the BIA must review the IJ's credibility findings for clear error—just as the BIA does with all factual findings by the IJ.

6.   Of course, just like any other factual finding, we are not barred from reversing an adverse credibility finding if it is unsupported by substantial evidence. See Shrestha, 590 F.3d at 1042; see also Zhiqiang Hu v. Holder, 652 F.3d 1011, 1016 n.4 (9th Cir. 2011). But we must actually determine that it isn't supported by substantial evidence. We don't get to just pretend the adverse credibility finding doesn't exist.

7.   Dai’s unsupported stray remark that “when the BIA has on appeal [not] affirmed an adverse credibility finding made by the IJ ․ we may not deny the petition for review based on lack of credibility,” 884 F.3d at 869, is dictum divorced from that case's facts and disposition. In Dai, “the IJ did not make an adverse credibility finding.” Id. at 865.

8.   Indeed, this court in Dai acknowledged that “the IJ's decision not to make an explicit adverse credibility finding means that there is no finding to which we can defer.” 884 F.3d at 868 (citations, alteration marks, and internal quotation marks omitted).

9.   It is also worth emphasizing that Krotova is a pre-REAL ID Act case. And the earlier cases Krotova cites for its dictum do not, themselves, support it. 416 F.3d at 1084 (citing Maldonado-Cruz v. I.N.S., 883 F.2d 788, 789 (9th Cir. 1989)) (neither the IJ nor the BIA made an adverse credibility finding) and Damaize-Job v. I.N.S., 787 F.2d 1332, 1337 (9th Cir. 1986) (finding the IJ's adverse credibility finding unsupported by substantial evidence).

10.   It is notable that in Castro-Perez, “the IJ expressly determine[d] the petitioner [wa]s credible.” 409 F.3d at 1071.

1.   The concurrence misunderstands my position on Castro-Perez. According to the concurrence, I treat Velasquez-Gaspar's testimony as “the dispositive factor that pushes her petition over the evidentiary hurdle and distinguishes this case from Castro-Perez.” Concurrence 1071 (emphasis in original). That characterization is plainly wrong. Velasquez-Gaspar's testimony alone does not differentiate her case from Castro-Perez. Rather, it is the 2014 Report's statistical evidence on Guatemalan law enforcement that does so.

CALLAHAN, Circuit Judge:

Concurrence by Judge VanDyke; Dissent by Judge Paez

Copied to clipboard