Lapadat v. Bondi

FOR PUBLICATION UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT ION LAPADAT; GIOVANI No. 23-1745 BECALI LAPADAT; LAURA Agency Nos. LAPADAT; MIRABELA A208-585-239 LAPADAT; SIMONA LAPADAT, A209-171-053 A208-585-240 Petitioners, A209-171-054 v. A209-171-052 PAMELA BONDI, Attorney General, ORDER AND AMENDED Respondent. OPINION On Petition for Review of an Order of the Board of Immigration Appeals Argued and Submitted May 14, 2024 Pasadena, California Filed February 12, 2025 Amended July 31, 2025 Before: Ronald Lee Gilman, * N. Randy Smith, and Salvador Mendoza, Jr., Circuit Judges. * The Honorable Ronald Lee Gilman, United States Circuit Judge for the Court of Appeals, 6th Circuit, sitting by designation. 2 LAPADAT V. BONDI Order; Opinion by Judge Mendoza; Dissent by Judge N. Randy Smith SUMMARY ** Immigration The panel filed (1) an order amending its opinion filed February 12, 2025, and denying Respondent’s petition for panel rehearing; and (2) an amended opinion granting Ion Lapadat’s petition for review of the Board of Immigration Appeals’ decision affirming the denial of asylum and withholding of removal, and remanding for further proceedings. In the amended opinion, the panel held that the record compelled the conclusion that Ion’s past experiences rose to the level of persecution, and that the BIA erred when it determined that the Roma are not a disfavored group in Romania. The panel concluded that the BIA erred by ignoring Ion’s credible, highly probative, and potentially dispositive testimony when it determined that his mistreatment was insufficiently severe to constitute persecution. Ion’s testimony that he was shot in the back, together with his family’s credible testimony and the remaining record evidence, collectively compelled a finding of serious harm ** This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader.  LAPADAT V. BONDI 3 that rose to the level of past persecution. The panel wrote that this was especially so given the severe assaults, attempted kidnappings, threats, violence, and mistreatment that Ion and his family faced in Romania. The panel noted that it was deciding only whether the Lapadats had established the first prong of the past- persecution claim—serious harm rising to the level of persecution. The panel left it to the agency to determine in the first instance the other two prongs—whether the persecution was motivated on account of a protected ground and was committed by the government, or by forces that the government was unable or unwilling to control. The panel concluded that the BIA also erred when it held that the Roma are not a disfavored group in Romania. Eliding centuries of anti-Roma abuse in Romania, the BIA mistakenly swapped the European Union’s proposed reforms, designed to aid Europe’s Roma population, with the Romanian government’s own documented and persecutory conduct toward the Roma. Moreover, the record unmistakably established that the Roma are a disfavored group in Romania. Given the errors in its past-persecution and disfavored- group analyses, the panel wrote that the BIA likely erred when it determined that Ion lacked a sufficient individualized risk of future persecution to make his fear of return to Romania objectively reasonable. The panel left it to the agency to determine whether Ion established a sufficiently individualized risk of persecution considering the egregious evidence of group persecution. Dissenting, Judge N.R. Smith wrote that the majority incorrectly limited its review to the BIA’s decision, improperly substituted its own view for that of the IJ and the 4 LAPADAT V. BONDI BIA in determining past persecution, and assessed the harms the Lapadats suffered out of context by failing to tie them to persecution on account of a protected ground. Additionally, in remanding for a disfavored group analysis, the majority ignored the BIA’s alternative determination that even if the Roma constitute a disfavored group there was insufficient evidence that any of the individuals who targeted Lapadat were likely to do so in the future. COUNSEL Colyn B. Desatnik (argued), Colyn B. Desatnik APLC, Irvine, California, for Petitioners. Jonathan Needle (argued) and Roberta O. Roberts; Zoe J. Heller and Walter M. Evans, Senior Litigation Counsel; Office of Immigration Litigation, Civil Division; Bernard A. Joseph, Senior Litigation Counsel; Brian M. Boynton, Principal Deputy Assistant Attorney General; United States Department of Justice, Washington, D.C.; for Respondent. ORDER The opinion filed February 12, 2025, is hereby amended. The amended opinion will be filed concurrently with this order. Judges Gilman and Mendoza vote to deny Respondent’s petition for panel rehearing, and Judge N.R. Smith votes to grant the petition for rehearing. The petition for panel rehearing (Dkt. No. 40) is DENIED. No further petitions for rehearing will be considered.  LAPADAT V. BONDI 5 OPINION MENDOZA, Circuit Judge: As appellate judges, we generally defer to the reasoned and expert judgment of our colleagues in the Board of Immigration Appeals (“BIA”), whom we trust to carefully review the record and apply the law. But judges are human, and like all humans, they sometimes make mistakes. Such is the case today. Ion Lapadat, a native of Romania, seeks asylum for himself and his family because he fears persecution on account of their Roma ethnicity. The BIA was unpersuaded by his petition, and it affirmed the Immigration Judge’s (“IJ”) denial of Ion’s application and his family’s derivative applications for asylum, withholding of removal, and protection under the Convention Against Torture (“CAT”). Relevant here, the BIA concluded that Ion had established neither past persecution nor a well-founded fear of future persecution in Romania. Our ruling today does not decide whether the Lapadats have met all the elements required to attain asylum or the withholding of removal. Rather, we write to address only the two errors on which the BIA’s decision rests. First, the BIA ignored Ion’s credible, highly probative, and potentially dispositive testimony when it determined that Ion’s mistreatment was insufficiently severe to constitute persecution. Ion’s testimony that he was shot in the back, together with his family’s credible testimony and the remaining record evidence, collectively compels a finding of serious harm that rises to the level of past persecution. This is especially so given the severe assaults, attempted kidnappings, threats, violence, and mistreatment that the Lapadats faced in Romania. Second, the BIA erred when it 6 LAPADAT V. BONDI held that the Roma are not a disfavored group in Romania. Eliding centuries of anti-Roma abuse in Romania, it swapped the European Union’s proposed reforms, designed to aid Europe’s Roma population, with the Romanian government’s own documented and persecutory conduct toward the Roma. In our view, the record unmistakably establishes that the Roma are a disfavored group in Romania. At base, our opinion today does little more than accomplish Congress’s goal when it passed the Refugee Act of 1980. We give proper deference to the BIA’s analysis and correct its clear missteps, thus preserving our nation’s systematic procedures for admitting refugees who have been persecuted in their countries of origin. Accordingly, we grant Ion’s petition and remand for further proceedings consistent with this opinion. I. Ion, his wife Simona, and their children—Giovani, Laura, and Mirabela—are Roma and natives and citizens of Romania. 1 The Roma—commonly and derogatively known as “gypsies”—are Europe’s largest ethnic minority, identifiable by their darker complexions, historically nomadic lifestyle, and characteristic mode of dress. See Mihalev v. Ashcroft, 388 F.3d 722, 726 (9th Cir. 2004) (“There is no question that Gypsies are an identifiable ethnic group and that being a Gypsy is a protected ground under § 208 of the INA.”). The Lapadats lived in Romania until 2013, moved to France for two years, returned to Romania in 2015, and fled to the United States in 2016. After the 1 The record also contains an application for Mariana Lapadat—Ion’s and Simona’s fourth child. The record reflects that Mariana failed to appear for the merits hearing and was ordered removed in absentia. That decision was not appealed.  LAPADAT V. BONDI 7 Department of Homeland Security served the Lapadats with notices to appear, they timely sought asylum, withholding of removal, and protection under CAT, citing the “racial hate” that they faced living in Romania. Ion submitted the application, and his wife and four children are named as derivative asylum applicants. A. At the Lapadats’ removal hearing, the IJ found each testifying family member—Ion, Simona, and Laura— broadly credible. Over the course of multiple days, they each testified that they endured years of significant hardship while living in Romania on account of their Roma ethnicity. Ion described a particularly harrowing incident from 2007, when he attempted to collect firewood on land that he said belonged to the city hall. While Ion was collecting wood, a man appeared and accused him of trespassing. The man grabbed a rifle from his car and, as Ion apologized and tried to flee in his horse-drawn cart, the man began cursing him for being a “gypsy.” The man then opened fire: Ion heard at least one shot, and bullets struck his back and skull. Bleeding profusely, Ion rushed to the hospital, only for the police to take him in for questioning before he could receive treatment. Once he told the police who shot him, they refused to take his story, threatened to imprison him if he tried to press charges, and left him to find his own way home. As it turns out, the police knew the shooter. Ion luckily made it back to his family, changed his blood-soaked shirt, and returned to the hospital, where a doctor discharged him, casually remarking that he was “not going to die [from] this.” Ion testified that shards of the bullets remain in his body to this day. 8 LAPADAT V. BONDI Simona and Laura credibly shared their own horrifying story. Laura and her sister, Mariana, were each younger than fourteen years old when a group of young Romanian men attempted to abduct them outside of their school, dragging them into a car while threatening to rape them. Men frequently harassed and assaulted the girls at school for being “gypsies,” but had never attempted actual kidnap before. When Simona tried to stop the men from abducting her daughters, the men slashed her back with a knife, threatening to kill her and rape her and her daughters if she called for help. 2 Thankfully, she thwarted the attack. But a few days later, before dawn, the men came to the Lapadats’ home and tried to break in, again threatening to rape Simona and her children if Simona tried to file a complaint with the police. As with Ion’s shooting, Romanian officials were also uninterested in prosecuting this crime—the police just “didn’t believe them” and indicated that they had “no problem with that.” That testimony was just the tip of the iceberg. For one, Ion credibly testified that he was frequently and arbitrarily detained by police, who often beat him. He described one such event in detail. In 2011, an officer detained and handcuffed Ion after he discovered Ion on the streets without identification after 7:00 p.m. He carted Ion to the police station and cursed him for being a “gypsy.” While at the station, the officer did not charge Ion with a crime. Instead, he slapped Ion across the face twice, kicked him in the abdomen, and sent him on his way. Apparently, walking while Roma will get you arrested and beaten, but not formally charged. 2 The BIA mistakenly found that the men cut Simona’s hand, but that factual finding is directly contradicted by the record.  LAPADAT V. BONDI 9 Ion also testified that he and his family frequently struggled to obtain consistent and prompt government or public services, and that they suffered from pervasive discrimination. He described, for example, seeking treatment at a Romanian hospital for acute appendicitis. Despite the hospital stating that he needed “urgent” care, doctors let him languish in pain for two days before removing his appendix. It is not just hospitals that refused to accommodate Ion. Before he left for France, Ion recalls being denied entrance to a store by a security officer, who pointed to a sign on a window that said “gypsies and dogs” were not allowed inside. In a separate incident, a waitress refused to take his order, explaining the restaurant’s policy of denying services to Roma people. These types of events also occurred after he returned to Romania from France. Ion recalls being denied the simple pleasure of going to a public swimming pool because the man selling tickets refused him entry and punched Ion in the chest when he inquired why. Ion’s family has also experienced discrimination of a similar ilk. Ion’s brother, for instance, tried to purchase a house in a non- Roma neighborhood. The Romanian residents of that neighborhood threatened to set his house on fire rather than let him live there. As for employment, Ion testified that he had to rely on the kindness of friends, and like others in his community, he could not obtain official employment because he is Roma and has “no right to work in Romania.” No company would hire him, no office would consider his application, and he lacked the right documents to secure formal employment. He testified that hiring personnel turned him away because he is a “gypsy.” Accordingly, Ion worked odd jobs for other Romani people—building their fences and houses. When 10 LAPADAT V. BONDI the Lapadats moved to France, Ion tried to find stable work but, just as in Romania, he lacked the necessary paperwork to get a job. So his wife resorted to begging to support them. Ion also testified at length about his fear of a local police officer, whom he knows by name and who, like the officer from the 2011 incident, has repeatedly harassed and beaten him. This officer made a habit of kicking, slapping, and threatening Ion with physical harm and false imprisonment, all because Ion is a “gypsy” and “[y]ou gypsies are idiots.” That harassment continued when the Lapadats briefly returned to Romania from France, before making their way to the United States. One day before the Lapadats left for the United States, that same officer recognized Ion on the street. As before, the officer physically assaulted Ion, slapping him over the head, kicking him, and demanding to know why he had “crawl[ed]” back to Romania. It seems that this incident cemented in Ion’s mind that he and his family were not and would never be safe in their homeland. The beatings, the shootings, the threatened rape, the harassment, the discrimination, the unemployment, the poverty, and the lack of medical attention had taken their toll on the Lapadats. B. The Lapadats’ testimony reflects the experiences of many Romani people living in Europe and Romania. The Roma have been marginalized and vilified throughout Europe for centuries, and are frequently derided as swindlers, petty criminals, and scum. According to the European Union (“EU”), anti-Roma “discrimination continues to be widespread across the EU and is present in all societies, and in all key areas.” Romania is no exception. The EU classifies Romania as a “Cluster 1” country, which  LAPADAT V. BONDI 11 means it has one of the largest Roma communities in Europe and faces the “most acute challenges” relating to “antigypsyism.” Defined as “the specific racism towards Roma who are stigmatized as ‘gypsies’ in the public imagination,” antigypsyism animates severe mistreatment of the Roma in Romania. The EU has recently taken measures to combat antigypsyism in Europe. Starting in 2011, the EU developed and implemented a “Framework” for national Roma integration strategies across the continent. Seeking to “close the gap between Roma and non-Roma in four key areas: education, employment, healthcare and housing,” the EU’s Framework “invited Member States to design national Roma integration strategies.” The record is silent as to whether Romania has designed or otherwise proposed such a strategy. As of 2020, the EU’s Framework had not accomplished its goals. Relying on survey data, the EU reports that “[a]n overwhelming majority of [stakeholders] think that the situation of Roma is worse than that of non-Roma” in all target areas. And the EU concedes that its Framework has failed to meaningfully counteract antigypsyism. Since its inception, EU stakeholders report that while there have been modest improvements in Roma access to education and health outcomes, discrimination against Romani populations in housing and employment has worsened since 2011, and Roma are generally discriminated against more than they were before in Europe. Accordingly, the EU projects that “real impact may not be seen for at least a generation.” C. After considering this documentary evidence and Ion’s, Simona’s, and Laura’s credible testimony, the IJ denied 12 LAPADAT V. BONDI Ion’s application and his family’s derivative applications for relief. The BIA affirmed the IJ’s decision. First, although the BIA concurred with the IJ that Ion had faced discrimination, it found that the mistreatment that he suffered in the past “did not rise to the level of persecution.” Second, the BIA affirmed that the Lapadats could not establish a well-founded fear of future persecution. It held that the Roma are not a disfavored group in Romania, given the EU’s steps to combat anti-Roma sentiment in Europe through its Framework. And it agreed with the IJ that Ion’s evidence of past harm was insufficient to show that he would be “single[d] out and target[ed] for persecution in the future.” 3 This appeal timely followed. II. We have jurisdiction under 8 U.S.C. § 1252(a). When, as here, “the BIA conducts its own review of the evidence and law, rather than 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.” 4 Rodriguez v. Holder, 683 3 The BIA also determined that Ion waived his challenge to the IJ’s denial of CAT protection. In his briefing before this court and at oral argument, Ion failed to raise any issues or advance any arguments in favor of relief under CAT. Accordingly, “petitioners waived review of their CAT claim.” Tampubolon v. Holder, 610 F.3d 1056, 1058 n.3 (9th Cir. 2010). 4 The dissent suggests that we have ignored the record. Dissent at 37. Not so. We review the IJ’s decision only to the extent that the agency has expressly adopted the IJ’s reasoning. See Rodriguez, 683 F.3d at 1169; Hosseini v. Gonzales, 471 F.3d 953, 957 (9th Cir. 2006); Cordon-Garcia v. I.N.S., 204 F.3d 985, 990 (9th Cir. 2000); Ghaly v. I.N.S., 58 F.3d 1425, 1430 (9th Cir. 1995). Here, although it might be a close call because the BIA agreed with the IJ’s conclusion that the harm did not rise to the level of persecution, the BIA conducted its own review  LAPADAT V. BONDI 13 F.3d 1164, 1169 (9th Cir. 2012) (citation and quotation marks omitted). “We review factual findings for substantial evidence and legal questions de novo.” Guerra v. Barr, 974 F.3d 909, 911 (9th Cir. 2020). We consider factual findings “conclusive unless any reasonable adjudicator would be compelled to conclude to the contrary.” 8 U.S.C. § 1252(b)(4)(B). “While this standard is deferential, ‘deference does not mean blindness.’” Parada v. Sessions, 902 F.3d 901, 909 (9th Cir. 2018) (quoting Nguyen v. Holder, 763 F.3d 1022, 1029 (9th Cir. 2014)). III. To qualify for asylum, Ion must demonstrate that he “is unable or unwilling” to return to Romania “because of persecution or a well-founded fear of persecution on account of race, religion, nationality, membership in a particular social group, or political opinion.” Melkonian v. Ashcroft, 320 F.3d 1061, 1064 (9th Cir. 2003) (quoting 8 U.S.C. § 1101(a)(42)(A)). To be well-founded, Ion’s “fear of persecution must be both subjectively genuine and objectively reasonable.” Sael v. Ashcroft, 386 F.3d 922, 924 (9th Cir. 2004). As is the case here, “[a]n applicant ‘satisfies the subjective component by credibly testifying that [he] genuinely fears persecution.’” Id. (quoting Mgoian v. I.N.S., 184 F.3d 1029, 1035 (9th Cir. 1999)). The objective component can be established in two different ways: either of the evidence and relied on the IJ’s reasoning regarding only four out of the five harms discussed. The shooting—the most severe harm—was absent from the BIA’s consideration. Thus, we review only the BIA’s decision and disregard the IJ’s discussion of the shooting, as the BIA did not expressly adopt—or even mention—that discussion. Regardless, even if we look to the IJ’s reasoning, it is clear that the IJ also erred. See infra Section III.A.i.2. 14 LAPADAT V. BONDI by demonstrating (1) past persecution or (2) a well-founded fear of future persecution. See Ratnam v. I.N.S., 154 F.3d 990, 994 (9th Cir. 1998). If an applicant demonstrates past persecution, “then fear of future persecution is presumed.” Deloso v. Ashcroft, 393 F.3d 858, 863 (9th Cir. 2005). Here, the BIA affirmed the IJ’s holding that Ion failed to establish either past persecution or a well-founded fear of future persecution. Ion raises challenges to both objective- component determinations, and we consider each in turn. See Flores Molina v. Garland, 37 F.4th 626, 630, 637 (9th Cir. 2022). A. We acknowledge that our standard of review for past- persecution determinations appears to be in flux. In Kaur v. Wilkinson, we held that “[w]hether particular acts constitute persecution for asylum purposes is a legal question reviewed de novo.” 986 F.3d 1216, 1221 (9th Cir. 2021) (alterations omitted) (quoting Boer-Sedano v. Gonzales, 418 F.3d 1082, 1088 (9th Cir. 2005)). But we have also held that we “review for substantial evidence the BIA’s particular determination that a petitioner’s past harm ‘does not amount to past persecution.’” Sharma v. Garland, 9 F.4th 1052, 1060 (9th Cir. 2021) (alteration omitted) (quoting Villegas Sanchez v. Garland, 990 F.3d 1173, 1179 (9th Cir. 2021)). Other circuits’ decisions reflect this intra-circuit inconsistency. See Xue v. Lynch, 846 F.3d 1099, 1105 n.11 (10th Cir. 2017) (explaining that “[t]he circuits are split as to the standard of review applicable to the question [of] whether an undisputed set of facts constitute persecution” and collecting cases). Since 2021, our colleagues have thoughtfully explained why we ought to consider reviewing past-persecution determinations de novo, seeing as our review generally  LAPADAT V. BONDI 15 requires the “application of a legal standard to settled facts.” Flores Molina, 37 F.4th at 640 (Korman, J., concurring). 5 Like many panels before us, 6 we need not resolve this question today. Here, the record compels our conclusion that Ion experienced persecutory treatment under either standard of review. i. To establish past persecution, Ion must demonstrate (1) serious harm “ris[ing] to the level of persecution”; (2) that “the persecution was committed by the government, or by forces that the government was unable or unwilling to 5 In their concurring opinions in Fon v. Garland, Judge Graber and Judge Collins present competing concerns about resolving this standard-of- review issue. 34 F.4th 810 (9th Cir. 2022). Judge Graber recommends “a middle way” for this mixed question of law and fact, in which we adapt our review of a past-persecution determination depending on whether the question “entails primarily legal or factual work.” Id. at 816–17 (Graber, J., concurring). Judge Collins rejoins that “the question is actually quite a bit more complicated” than Judge Graber suggests, and he reasons that her path is foreclosed by our and the Supreme Court’s precedent. Id. at 819–20 (Collins, J., concurring). 6 See, e.g., Singh v. Garland, 97 F.4th 597, 603 (9th Cir. 2024) (“We need not address which standard should apply because the harm suffered by Singh rose to the level of persecution even under the substantial evidence standard, which affords greater deference to the BIA’s determinations.”); Flores Molina, 37 F.4th at 633 n.2 (applying the “more deferential ‘substantial evidence’ standard” to avoid “discuss[ing] the nuances” of the competing standards); Singh v. Garland, 57 F.4th 643, 652 (9th Cir. 2023) (declining to address the competing standards because “the harm [Singh] suffered rose to the level of persecution under the more deferential ‘substantial evidence’” standard (alteration in original) (quoting Flores Molina, 37 F.4th at 633 n.2)); Fon, 34 F.4th at 823 (Collins, J., concurring) (“[G]iven that the Petitioner here prevails even if we apply a more deferential standard of review to the agency’s decision, this is not [the] case [to clarify the standard of review].”). 16 LAPADAT V. BONDI control”; and (3) that “the persecution was on account of one or more protected grounds.” Kaur, 986 F.3d at 1221. Like in Flores Molina, here the BIA reached only the first element, finding that the Lapadats failed to establish a past- persecution claim because the harms they suffered did not rise to the level of persecution. See 37 F.4th at 633. The BIA erred in concluding so and in dismissing the past- persecution claim. Accordingly, we address only the first element of the past-persecution claim. We assess a past-persecution finding by “looking at the cumulative effect of all the incidents [that the] [p]etitioner has suffered.” Singh v. I.N.S., 134 F.3d 962, 967 (9th Cir. 1998). In Sharma v. Garland, we identified seven factors to guide a past-persecution analysis: physical injury or violence; “isolated” or “ongoing pattern[s] of serious maltreatment”; length and quality of detention; evidence of “threats” that “are repeated, specific[,] and combined with confrontation or other mistreatment”; harm to family and friends; “substantial economic deprivation that constitutes a threat to life or freedom”; or “political and social turmoil.” 9 F.4th at 1060–63 (citations and quotation marks omitted). No single factor is necessarily dispositive, and the list is non- exhaustive; but Sharma’s factors provide a helpful set of relevant indicia to assess whether mistreatment rises to the level of persecution, which is “an extreme concept” that eclipses mere harassment or discrimination. Ghaly v. I.N.S., 58 F.3d 1425, 1431 (9th Cir. 1995). 1. The record makes clear that Ion demonstrated harm rising to the level of persecution when he credibly testified that he was shot in the back for collecting firewood. This is not a close case. As we noted in Sharma, “physical  LAPADAT V. BONDI 17 violence” is “often a significant consideration” to a past- persecution analysis, and “when we have granted petitions for review because the record compelled a finding of past persecution,” we have frequently done so because the petitioner “experienced serious physical violence.” 9 F.4th at 1061. “There is no question” that “assaults on one’s life, family and business”—like, say, shooting someone—is serious physical violence, “ris[ing] to the level of persecution.” Singh v. I.N.S., 94 F.3d 1353, 1360 (9th Cir. 1996); see also Kaur, 986 F.3d at 1223 (“[W]e have held that attempted murder constitutes persecution.” (citing Lopez v. Ashcroft, 366 F.3d 799, 803 (9th Cir. 2004))); Madrigal v. Holder, 716 F.3d 499, 504 n.2 (9th Cir. 2013). 2. The BIA affirmed the IJ’s denial of Ion’s petition on past-persecution grounds, determining that the abuse he suffered did not reach the level of persecution. It committed legal and factual errors in the process. First, the BIA declined to consider or mention the shooting in its past-persecution analysis, despite Ion’s clear testimony that he was shot in the back. For its part, the IJ not only minimized the shooting, but mischaracterized it by noting that “[Ion] was physically harmed in the sense that he was shot at by an individual for trespassing onto property,” and that the shooting was “isolated.” The statement “in the sense that he was shot at,” is vague and makes unclear the extent of the harm Ion suffered. Moreover, saying that he was merely “shot at” is incorrect. Ion was struck with a bullet in the back and the head, where bullet fragments remain today. He was clearly physically harmed. Considering that the IJ had found Ion’s testimony credible, 18 LAPADAT V. BONDI the IJ knew this. And pretending otherwise, or minimizing it, is not supported by substantial evidence. The BIA did not mention, much less explicitly adopt, this aspect of the IJ’s decision when it affirmed the IJ’s past- persecution finding. Accordingly, we are bound to consider the BIA’s decision on this point, see Rodriguez, 683 F.3d at 1169, and its oversight was legal error, see Flores Molina, 37 F.4th at 632 (“Where the BIA does not consider all the evidence before it, either by ‘misstating the record [or] failing to mention highly probative or potentially dispositive evidence,’ its decision is legal error and ‘cannot stand.’” (alteration in original) (quoting Cole v. Holder, 659 F.3d 762, 772 (9th Cir. 2011))). Second, the IJ held, and the BIA agreed, that the Lapadats’ other evidence of maltreatment “did not rise to the level of persecution.” After considering and rejecting the “isolated” shooting, the IJ characterized the attempted kidnapping of the Lapadats’ daughters as an abhorrent-but- not-persecutory crime under our precedent. The IJ then rejected as insufficiently persecutory a laundry list of abuse (like being denied healthcare, work, and access to public pools and restaurants), as well as criminal acts, police beatings, and threats—each stretching across years. This too was error. 3. The shooting, the attempted kidnapping and the rape and death threats—accompanied by the Lapadats’ credible testimony regarding an ongoing pattern of serious maltreatment—collectively compel a finding of harm rising  LAPADAT V. BONDI 19 to the level of persecution. 7 See Gu v. Gonzales, 454 F.3d 1014, 1020 (9th Cir. 2006). We have held as much in similar cases. In Korablina v. I.N.S., for example, we considered a petition by a Jewish woman who suffered a physical attack on her life, watched other Jewish individuals suffer physical attacks, struggled to secure employment, and received threatening phone calls. 158 F.3d 1038, 1044–45 (9th Cir. 1998). In reversing the BIA’s past-persecution finding, we held that “[w]here evidence of a specific threat on [a petitioner]’s life . . . is presented in conjunction with evidence of political and social turmoil, the [petitioner] has succeeded in establishing a prima facie eligibility for 7 The dissent disagrees that the attempted kidnapping, and the rape and death threats rise to the level of persecution, but in doing so, the dissent conflates prong one with prong three. Dissent at 40–48. To meet prong one and establish a qualifying harm, a petitioner does not need to also prove prong three—whether the persecution was motivated on account of a protected ground. In Gormley v. Ashcroft, for example, we did not consider whether the perpetrators “victimized [the petitioner] on account of his race” or whether they were “groups that the government is unwilling or unable to control” until the analysis of the second and third prongs. 364 F.3d 1172, 1177 (9th Cir. 2004); see also Navas v. I.N.S., 217 F.3d 646, 658–61 (9th Cir. 2000) (identifying two distinct issues— whether the petitioner had demonstrated persecution and “whether the persecution was on account of [his] political opinion”—and proceeding to analyze them separately). We decide today only that the Lapadats have established the first prong of the past-persecution claim—serious harm “ris[ing] to the level of persecution.” See Kaur, 986 F.3d at 1221. We leave the other two prongs—including whether the persecution was motivated on account of a “protected ground”—to the agency to determine in the first instance. See id. We do not dispute that the Lapadats must satisfy all three prongs in order to prevail on their asylum claim, but we rebut the assertion that prong three subsumes the first prong. 20 LAPADAT V. BONDI asylum.” Id. at 1045. Likewise, in Guo v. Sessions, we considered a Christian Chinese citizen’s petition for asylum. 897 F.3d 1208, 1213 (9th Cir. 2018). Drawing on our decision in Korablina, we reversed the BIA, and we held that the record “compel[led] a finding of past persecution” because police beat the petitioner, even though the beating resulted in “no permanent injuries”; the beating required a hospital stay; and the petitioner received a short detention, was forced to disavow his faith, and was impeded from practicing it. Id. at 1215. Or take our decision in Baballah v. Ashcroft, where we considered an Israeli citizen’s petition for asylum based on mistreatment that he received for his parents’ “mixed marriage.” 367 F.3d 1067, 1071 (9th Cir. 2004). The Baballah petitioner presented evidence that Israeli citizens and officials “shot bullets in the air over” his fishing boat; sprayed him, his brother, and his boat with “pressurized water in freezing weather” in an apparent effort to sink it; called him religious and ethnic slurs; “made it impossible” for the petitioner “to earn a living”; and “refused” to employ his family members. Id. at 1071–72. We reversed the BIA because, “[w]hen analyzed in the aggregate, the physical assaults and economic harassment endured by [the petitioner] compel a finding of persecution.” Id. at 1076. Our recent decision in Kaur is also instructive. 986 F.3d at 1219. There, we addressed a petition by an Indian petitioner largely based on an attempted rape connected to her political beliefs, and we determined that the BIA’s rejection of her petition was “marred by legal error.” Id. at 1219, 1222. First, we held that “[s]imilar to attempted murder and attempted kidnapping, attempted rape almost always constitutes persecution.” Id. at 1224. After all, “in evaluating whether past treatment rises to the level of  LAPADAT V. BONDI 21 persecution, we do not look to the level of harm experienced by the petitioner.” Id. at 1226. It is the “treatment” that counts. Id. (quoting Mihalev, 388 F.3d at 729). Second, we held that it was “plain on the record” that petitioner “suffered past persecution.” Id. at 1227. In addition to the attempted rape, petitioner “endured death threats,” and her parents “were attacked on multiple occasions.” Id. We found such evidence sufficient to compel a finding of mistreatment rising to the level of persecution. 8 Id. Trying to compare harms between past and present petitioners is often a perverse task. But we conclude that the Lapadats’ persecution is plain on the record and falls squarely within the bounds established by our precedent. “It is, of course, ‘well established that physical violence is persecution under 8 U.S.C. § 1001(a)(42)(A).’” Parada, 902 F.3d at 909 (quoting Li v. Holder, 559 F.3d 1096, 1107 (9th Cir. 2009)). Like the Korablina and Baballah 8 The dissent invokes Beskovic v. Gonzales, 467 F.3d 223, 226 (2d Cir. 2006), for the proposition that the degree of harm suffered “must be assessed with regard to the context in which the mistreatment occurs.” Dissent at 43. In other words, if the harm suffered is not connected to a protected ground, it necessarily cannot amount to harm rising to the level of persecution. But that is not what Beskovic stands for. Beskovic elaborated that “even mistreatment that, in other contexts, could fairly be characterized as the mere annoyance and distress of harassment can take on an entirely different character when officially inflicted on an individual while detained on account of [a] protected ground[].” Id. at 226 (internal quotation marks and citation omitted). First, being shot and receiving death and rape threats can hardly be considered a “mere annoyance and distress.” Second, Beskovic explains a one-way ratchet— harassment and mistreatment that would typically not be considered a harm rising to the level of persecution might be considered such a harm when it is motivated by a protected ground. Beskovic does not provide a general rule allowing courts to downgrade or disregard severe harms (like being shot) simply because prong three is not satisfied. 22 LAPADAT V. BONDI petitioners, each of whom recounted assaults that could have turned deadly, Ion too experienced a violent physical attack on his life. 9 His account of being cursed and shot—and then dragged by the police from the hospital, blood soaking his clothes, only to have those same police threaten to imprison him for reporting the crime—is chilling. Indeed, he must relive this experience every time he contemplates the shrapnel still embedded in his body. In our opinion, this shooting eclipses the harm we deemed sufficient in Guo, where petitioner testified to “repeated baton blows” to parts of his body by police, which required a hospital examination but left no lasting injuries. 897 F.3d at 1215. Likewise, the attempted kidnapping of Laura and Mariana, and the threatened rape and threatened murder of the girls and their mother, echo the events described in Kaur. As we explained there, “some forms of physical violence,” like “attempted murder and attempted kidnapping,” “are so extreme that even attempts to commit them constitute persecution.” 10 See Kaur, 986 F.3d at 1223–24; see also 9 The dissent criticizes our reliance on Korablina, Baballah, and Kaur because, in those cases, the persecutory conduct was connected to a protected ground. Dissent at 53. The dissent’s concern is unwarranted. Of course, when you discuss all three prongs of a past-persecution claim, like in Korablina and Baballah, prongs one and three may bleed together. And in Kaur, it is unsurprising that the court repeatedly mentioned that the persecutory conduct was motivated on a protected ground because it was “uncontested that Kaur had suffered past physical abuse and death threats on account of her political opinion.” 986 F.3d at 1222. Thus, we do not read those cases as precluding us from analyzing only prong one—persecutory conduct—particularly when that is the only prong addressed by the BIA here. 10 The dissent’s quibbling with the Lapadats’ characterization of the shooting as “attempted murder,” dissent at 46, misses the mark. What is  LAPADAT V. BONDI 23 Sangha v. I.N.S., 103 F.3d 1482, 1487 (9th Cir. 1997) (attempting to forcibly conscript a petitioner, which was tantamount to attempted kidnapping, constituted persecution). Indeed, those moments appear seared in Laura’s mind: she still cries each time she recalls this incident, making tangible the Kaur court’s thoughtful explanation that loss of “bodily autonomy” causes severe psychological damage. 986 F.3d at 1224–25 (explaining that attempted rape and sexual assault impose “severe psychological effects,” like shame, clouded memory, self- blame, and a “pervasive feeling of loss of control”). And the Lapadats’ fear of further violence from this incident is justified—the men, after all, returned to the Lapadats’ home, threatening to kill and rape them if Simona or the girls tried to go to the police. These events, when considered in the context of the ongoing mistreatment that the Lapadats experienced, compel a finding of maltreatment rising to the level of persecution. As we noted in Sharma, “serious maltreatment that is sustained and recurring is more likely to compel the conclusion of past persecution” when it is “[i]n combination with other indicia of persecution.” 9 F.4th at 1061; see also Korablina, 158 F.3d at 1045 (holding that the conjunction of social unrest and a credible threat on a petitioner’s life satisfied the past-persecution standard). Ion’s shooting and his daughters’ attempted kidnapping did not occur in relevant is that Ion was shot in the back and the head. Whether the shooting can be characterized as an attempted murder is not dispositive because, regardless of the verbiage used, as we outline above, “serious physical violence,” such as being shot, is sufficient to establish persecution. See Sharma, 9 F.4th at 1061; see also Guo, 897 F.3d at 1215 (noting that “it would be a strange rule if the absence or presence of a broken arm were the dispositive fact”). 24 LAPADAT V. BONDI isolation, contrary to what the IJ held. 11 Far from it. That “contention disregards the reality that . . . persecution was ongoing”: the Lapadats, like many Romani people, had been mistreated by the police and Romanian public for years because of their ethnicity. Guo, 897 F.3d at 1216. Ion testified to repeated harassment and beatings by Romanian police and his inability to secure meaningful or formal employment, basic healthcare, or public services. And both he and his family recounted the death, rape, and arson threats that they have endured. Such pernicious maltreatment clearly satisfies the past-persecution criteria established in our caselaw. See Ahmed v. Keisler, 504 F.3d 1183, 1194 (9th Cir. 2007) (“Where an asylum applicant suffers [physical harm and threats] on more than one occasion . . . , the cumulative effect of the harms is severe enough that no reasonable fact-finder could conclude that it did not rise to the level of persecution.”). For its part, the government asks us to discount the persecution that the Lapadats suffered because it produced no “lasting injuries.” Our law has never imposed such a requirement. See Guo, 897 F.3d at 1215 (“[A] beating ‘may constitute persecution, even when there are no long-term effects and the [petitioner] does not seek medical attention.’” (second alteration in original and citation omitted)). Indeed, we’ve routinely held the opposite. In Kaur, for example, we considered and rejected the BIA’s attempts to require a 11 And even if the shooting and attempted kidnappings were the only persecutory acts in the record, those standing alone would be sufficient. It is clear beyond peradventure that a single, “isolated,” persecutory act can—and often does—amount to harm reaching the level of persecution if that act is sufficiently severe. See, e.g., Kaur, 986 F.3d at 1224 (noting that “attempted murder[,] attempted kidnapping, [and] attempted rape almost always constitutes persecution”).  LAPADAT V. BONDI 25 petitioner to produce “evidence of ongoing trauma or psychological treatment to establish a claim to past persecution on account of attempted rape.” 986 F.3d at 1225–26. Holding that such a view is legal error, we determined that, when evaluating “whether past treatment rises to the level of persecution, we do not look to the level of harm experienced by the petitioner.” Id. at 1226 (emphasis added). Rather, we look at the “conduct of the persecutor.” Id. Or, as we put it more colloquially in Mihalev, “it would be a strange rule if the absence or presence of a broken arm were the dispositive fact” for a past-persecution finding. 388 F.3d at 730 (overturning agency’s past-persecution finding, despite petitioner’s lack of “serious bodily injury”). Here, the “treatment” Ion suffered was extreme: he was shot for being a “gypsy.” The “treatment” his wife and children suffered was also extreme: Simona was slashed for attempting to protect her daughters from an attempted kidnapping and threatened rape. We would “miss[] the forest for the trees,” Guo, 897 F.3d at 1215, if we required the Lapadats to prove that they suffered lasting injuries from these assaults in order to demonstrate persecution. Additionally, the underlying conduct described in the three decisions cited by the BIA to support its findings bears no resemblance to the violence, threats, and mistreatment that the Lapadats have experienced. In Fisher v. I.N.S., for example, we denied an Iranian petitioner’s appeal based in part on inadequately persecutory police conduct. 79 F.3d 955, 959 (9th Cir. 1996). In that case, Iranian police questioned and detained the petitioner for attending a party at a male friend’s home where that friend wore a bathing suit; ordered her, albeit at gunpoint, to cover her hair beneath her chador; and visited her family home seeking information 26 LAPADAT V. BONDI about potential political dissidents. Id. That conduct left the Fisher petitioner “ill,” id. at 960, but it hardly resembles (and pales in comparison to) the persecution suffered by the Lapadats. The BIA and the dissent also invoked Gormley v. Ashcroft, but that case is inapposite. 364 F.3d 1172, 1177– 78 (9th Cir. 2004). The Gormley petitioner experienced a few robberies committed by “anonymous thieves” who the government would willingly punish if they could. Id. at 1177–81. He also stated that white men like himself struggled to find jobs in post-apartheid South Africa. Id. at 1175; see also Hussain v. Rosen, 985 F.3d 634, 646–47 (9th Cir. 2021) (dismissing civil unrest and violence that harmed petitioner because it was “the result of a general attack on the town” and did not involve “any individualized physical attacks or threats”). None of “these factual scenarios are close to the sustained, repeated, specific” threats and violence to which Ion was subjected. Flores Molina, 37 F.4th at 637. There is no evidence in the record that targeted shootings and attempted kidnappings are “all too common” “general conditions” in Romania. Cf. Gormley, 364 F.3d at 1177 (concluding that robberies “are an all too common byproduct of civil unrest and economic turmoil” in South Africa). The dissent also relies on Gormley (and a host of other cases) for the proposition that “not all incidents qualify as persecution,” such as discrimination or harassment or other general conditions like criminal activity. Dissent at 41–42. We agree, but those are far from the circumstances presented here. Additionally, none of those cases hold that whether an incident is a “general condition” hinges on whether the harmful conduct is motivated by the petitioner’s status. See id. (finding that two robberies did not rise to the level of  LAPADAT V. BONDI 27 persecution because “robberies of this sort are an all too common byproduct of civil unrest and economic turmoil”). In sum, by properly applying our caselaw to the entirety of the record, any reasonable factfinder would be compelled to find that Ion’s past experiences “rose to the level of persecution.” Baghdasaryan v. Holder, 592 F.3d 1018, 1023 (9th Cir. 2010). On remand, the BIA must address the remaining elements of Ion’s claim for asylum based on past persecution. 12 Assuming that Ion satisfies those elements, a “rebuttable presumption of a well-founded fear arises,” and the government has the burden to prove “that there has been a fundamental change in circumstances such that the applicant no longer has a well-founded fear.” Tawadrus v. Ashcroft, 364 F.3d 1099, 1103 (9th Cir. 2004) (citation and quotation marks omitted). B. Because the BIA must still address whether Ion’s persecutory treatment was on account of his Roma ethnicity and whether Romanian officials were unable or unwilling to protect him, we also consider Ion’s challenge to the BIA’s determination that he failed to demonstrate a well-founded fear of future persecution. See Flores Molina, 37 F.4th at 637. “An asylum applicant demonstrates a well-founded 12 Specifically, the BIA must determine whether Ion’s persecution was on account of one or more of the five protected grounds, and that the persecution was committed either by the government or by forces that the government was unable or unwilling to control. See Chand v. I.N.S., 222 F.3d 1066, 1073 (9th Cir. 2000). As discussed, the “record evidence strongly suggests that” Ion was persecuted on account of his Roma ethnicity, often by government officials and police, and that those police were unwilling to do anything to protect him or his family. Flores Molina, 37 F.4th at 637 n.5. But “that is a determination the agency must make in the first instance.” Id. 28 LAPADAT V. BONDI fear of future persecution in either of two ways”: (1) by showing a “pattern or practice of persecution” for similarly situated people or (2) by demonstrating that he “is a member of a ‘disfavored group’” and is “likely to be targeted as a member of that group.” Sael, 386 F.3d at 925 (internal quotations and citations omitted). Ion chose the latter route. Under our disfavored-group precedent, we have held that the relationship between disfavored-group status and individual risk is “correlational”: “the more serious and widespread the threat of persecution to the group, the less individualized the threat of persecution needs to be.” Mgoian, 184 F.3d at 1035 n.4. And we require the BIA to “evaluate all relevant evidence in the record to determine whether [a petitioner] carried her burden” of proving a fear of future persecution. Davila v. Barr, 968 F.3d 1136, 1143 (9th Cir. 2020). i. This record compels a finding that the Roma are a disfavored group in Romania. 13 A “disfavored group” is “a group of individuals in a certain country or part of a country, all of whom share a common, protected characteristic, many of whom are mistreated, and a substantial number of whom are persecuted.” Wakkary v. Holder, 558 F.3d 1049, 1052 (9th Cir. 2009). In Sael, we held that Indonesia’s Chinese minority were a “disfavored group” because of Indonesia’s centuries-long history of anti-Chinese persecution, which included near genocidal pogroms, bloody riots, and burned homes. 386 F.3d at 925–26, 929. In reversing the BIA’s well-founded fear analysis, we determined that Indonesia’s anti-Chinese history outweighed the economic success that some ethnic Chinese had recently enjoyed in the country, as well as the Indonesian government’s official policies 13 The dissent does not dispute this finding.  LAPADAT V. BONDI 29 promoting ethnic tolerance. Id. at 929. So too in Tampubolon v. Holder, where we held that “Christian Indonesians” are a disfavored group. 610 F.3d 1056, 1062 (9th Cir. 2010). Like the Sael court, the Tampubolon court considered the Indonesian government’s failure to adequately protect the rights of its religious minorities, the country’s history of violently persecuting Christian residents, and its official and private acts of discrimination. Id. at 1060–62. By contrast, in Halim v. Holder, we held that the petitioner had not carried his burden of demonstrating that ethnic Chinese were still a disfavored group in Indonesia. 590 F.3d 971, 979 (9th Cir. 2009). Distinguishing Sael, the Halim court renewed its focus on “the government’s perspective,” noting that “the materials before the court indicate that the Indonesian government does not condone discrimination against ethnic Chinese.” Id. After all, “the Indonesian ‘government has taken concrete steps to suppress ethnic and religious violence and to encourage reconciliation between opposing groups.’” Id. (quoting Lolong v. Gonzales, 484 F.3d 1173, 1181 (9th Cir. 2007)). Here, any reasonable factfinder would be compelled to conclude that the Roma are a disfavored group in Romania. As in Sael and Tampubolon, the record evidence documents centuries of anti-Roma persecution and violence throughout Romania and Europe. Often derided as swindlers, thieves, beggars, and scum, European Roma have been viciously persecuted for centuries. In the 1930s and 1940s, Nazi racial ideology put Romani, Jewish, and Black people at the bottom of the racial scale, and the Nazi extermination of the Romani was so thorough that it led to the extinction of the Bohemian Romani language in 1970. Those pernicious sentiments have not abated with time. Thomas 30 LAPADAT V. BONDI Hammarberg, Council of Europe Commissioner for Human Rights, notes that “today’s rhetoric against the Roma is very similar to the one used by Nazis and fascists before the mass killings started in the thirties and forties.” Put simply, the Roma have endured centuries of exclusion and deprivation at the hands of their European compatriots. Romania’s persecution and mistreatment of its Roma population is particularly severe. Although Romania has one of the largest Roma populations in Europe, antigypsyism pervades private and public life throughout the country. According to the Romania 2018 Human Rights Report (the “2018 Report”), “endemic official corruption and police violence against the Roma” is a major problem, and Romani groups complain that police harassment, brutality, and beatings are routine. As was the case for Ion, officers assault Romani people who use public spaces for improper purposes. One example included officers using “excessive force against two Romani teenagers caught fishing in a public park.” Indeed, the 2018 Report and the record abound with examples of police and constables mistreating and abusing Romani people. This type of mistreatment is well- tolerated by Romania’s government. As discussed in the 2018 Report, the European Court of Human Rights ruled that the Romanian justice system has failed to deliver just outcomes in cases of police brutality against Romani individuals. Likewise, the Romani Center for Social Intervention and Studies found that, in forty-three cases of police brutality against Roma people over the previous twelve years, there were no convictions at the national level. The reason? Prosecutors refused to take the cases to court. Unlike in Halim, the record here also reflects that Romania’s local and national government takes a permissive view of antigypsyism. In 2010, for example, a member of  LAPADAT V. BONDI 31 Romania’s ruling Liberal Democratic Party sponsored legislation that would rename the Roma “Tigan,” which derives from the Greek word “untouchable” and is analogous to “n****r” in the United States. He claimed that “those Gypsy fellows in Transylvania who wear big hats” supported the move. Critics pushed back, describing Roma outrage towards the legislation and noting that such a name-change likely violates international law, given that it deprives the Roma of their right to self-determination. The legislation was quickly withdrawn, but the damage was done. Likewise, right-wing political representatives have proposed that Roma women be sterilized, endorsing plans to offer money to any Romani woman willing to take them up on the offer. Additionally, and despite prohibitions on educational and physical segregation, several non-governmental organizations and news outlets report that Roma are still physically separated from their non-Roma peers. This includes government-sponsored walls sealing off “Romanian gypsy ghetto[s]”—a modern simulacrum of the Josefov in Prague or the Judengasse in Frankfurt. Romanian authorities also frequently and arbitrarily deny government- sponsored pensions to Romani Holocaust survivors, misclassifying them as “resettled” and “not deported” to concentration camps during the war. As in Tampubolon, Romania’s Roma “have also suffered private discrimination and marginalization by the general populace.” 610 F.3d at 1061. The 2018 Report describes broad “societal discrimination against Roma.” Roma are consistently denied access to, or refused service in, many public spaces. They lack employment opportunities, and they are frequently denied adequate health care and educational opportunities. News outlets report that Romanian Roma also face mass evictions. Indeed, one 32 LAPADAT V. BONDI article details an incident in which a municipal government gave 400 Roma two days’ notice of eviction before removing them from their homes and placing them in a “garbage-dump” encampment on the outskirts of the city. Likewise, Romani people’s “lack of identity documents” leaves many of them unable to participate in elections, receive social benefits, access health insurance, or participate in the labor market. Other mistreatment is quieter but no less pervasive, and it is just as pernicious. The terms “gypsy and cheater have been so interchangeable historically that the word has entered the English language as a verb: he gypped me.” Prominent Romanian and European thinkers have written articles stating that “‘Gypsies’ are culturally inclined towards theft[,] and [they] use their minority status to ‘blackmail’ the majority.” Romanian academics have even claimed that Romani music increases aggressiveness and limits self-control, suggesting a “correlation” between preference for Romani music and “low cognitive skills.” Taken together, it is unsurprising that Romani people have a higher unemployment rate and a lower life expectancy than their non-Roma peers in Romania. Given this robust record of Roma mistreatment, the BIA conceded that “Roma individuals continue to experience discrimination and harassment” in Romania. But the BIA summarily discounted those concerns because the European Union’s Framework “attempt[s] to address issues experienced by Roma individuals” in Europe. By doing so, the BIA appears to adopt the view that we analyze whether a group is disfavored from the perspective of supranational organizations and not national governments. That approach distorts our precedent. “One factor critical to both a showing of ‘disfavor’ as well as individual targeting is the government’s perspective.” Halim, 590 F.3d at 979. The  LAPADAT V. BONDI 33 European Union is not a “government” under Halim. Id. And its policies are irrelevant, absent evidence that the country to which a petitioner faces removal has adopted and implemented them. After all, a “refugee” is one “who is outside any country of such person’s nationality” or country of habitual residence and is “unable or unwilling to return to . . . that country because of persecution or a well-founded fear of persecution.” 8 U.S.C. § 1101(a)(42) (emphasis added). Moreover, under the removal statute, we remove noncitizens to “countries,” not to regional collectives of states. See 8 U.S.C. § 1231(b)(1) et seq.; 8 U.S.C. § 1158(c)(1)(A) (providing that a noncitizen granted asylum shall not be removed or returned to his “country of nationality” or, absent a nationality, his country of “last habitual residence”); 8 C.F.R. § 241.15(a) (granting “discretion” to remove a noncitizen to “any country described in section 241(b)” of 8 U.S.C. § 1231(b)). Indeed, our entire focus for a fear-of-future-persecution analysis centers on a petitioner’s fear of being removed to his country of origin or habitual residence. That is why we focus on “government” policies and practices. See, e.g., Tampubolon, 610 F.3d at 1061 (considering whether the “Indonesian government discriminates against Christians”); Sael, 386 F.3d at 929 (weighing whether “the Indonesian government[’s]” official policy of “ethnic tolerance” outweighs evidence of “[o]fficial discrimination” by the government); Castro-Martinez v. Holder, 674 F.3d 1073, 1078 (9th Cir. 2011) (considering “the Mexican government’s efforts to prevent violence and discrimination against homosexuals”); Kotasz v. I.N.S., 31 F.3d 847, 854 n.12 (9th Cir. 1994) (“[T]he Hungarian government ha[s] instituted a variety of programs specifically designed to 34 LAPADAT V. BONDI improve the gypsies’ economic situation.”). It would be peculiar if we permitted the policy goals of a supranational coalition of states to obviate or otherwise negate the conduct of a single state. Absent evidence to the contrary, the former has no bearing on the latter. Regardless, even if the BIA were permitted to impute the EU’s efforts to combat antigypsyism onto the Romanian government, the record still compels a finding that the Roma are a disfavored group in Romania. Although official attempts to counteract discrimination are relevant, “[w]hen a minority group’s ‘disfavored’ status is rooted in centuries of persecution, year-to-year fluctuations cannot reasonably be viewed as disposing of an applicant’s claim.” Sael, 386 F.3d at 929. Here, the EU does not report on the success of its Framework in individual countries or member states. Nor does it report on how extensively any member state has adopted its proposed measures to combat antigypsyism. At best, it provides a table summarizing thirty-nine measures that member states can “report,” each of which promote Roma integration. Of those thirty-nine measures, Romania “reported” ten of them: six of which related to “education” (including one measure labeled “Other”); one related to “employment” (the measure labeled “Other”); one related to “healthcare;” and two related to housing (again, reporting the measure labeled “Other”). There is little that we can glean from Romania’s “reporting” of these measures. Instead, we are left with the EU’s assessment of the situation in Europe overall: (1) “no real improvements can be seen on the ground” in combatting discrimination; (2) there has been some attention given to changing educational segregation; (3) “Roma participation in the labour market remains very weak”; and (4) “[t]ackling the health inequalities endured by Roma remains an ongoing  LAPADAT V. BONDI 35 challenge.” Or, as the EU puts it: “Roma exclusion and discrimination has existed for centuries,” “[s]tructural changes need time,” and “real impact may not be seen for at least a generation.” This evidence compels the finding that the Roma are a disfavored group in Romania—a reality that the Roma have certainly grappled with for centuries. ii. Given the errors in its past-persecution and disfavored- group analyses, it comes as no surprise that the BIA likely erred when it determined that Ion lacked a “sufficient individualized risk of future persecution to make his fear of return to Romania objectively reasonable.” “Evidence of both individual and group targeting are relevant to demonstrate the likelihood that a particular individual will be persecuted.” Tampubolon, 610 F.3d at 1062 (citing Wakkary, 558 F.3d at 1062–63). When, as here, a petitioner presents “egregious” evidence of “group persecution,” he can provide “less evidence of individualized persecution” to meet the objective prong of a well-founded fear showing. Kotasz, 31 F.3d at 853; see also Wakkary, 558 F.3d at 1063. We leave it to the agency, however, to determine whether Ion meets this requirement. 14 14 The BIA denied Ion’s claim for withholding of removal because it determined that he had not met the requirements for asylum. Because the BIA erred in its denial of asylum, we remand Ion’s withholding of removal claim for further consideration. As we noted in Flores Molina, “if the BIA determines that [Ion] experienced past persecution on account of a protected ground,” then it “must credit [Ion] with a rebuttable presumption of eligibility for withholding of removal.” 37 F.4th at 638 (citing 8 C.F.R. § 1208.16(b)(1)(i); Ahmed, 504 F.3d at 1199). If the BIA determines that Ion is not entitled to a presumption of eligibility for withholding of removal, it must consider “all probative evidence related to [Ion’s] fear of future persecution.” Id. at 638–39. 36 LAPADAT V. BONDI IV. The record compels our conclusion that Ion Lapadat’s past experiences rise to the level of persecution, and that the BIA erred when it determined that Ion failed to demonstrate his membership in a disfavored group. Accordingly, we grant Ion’s petition and remand for further proceedings, consistent with this opinion. GRANTED and REMANDED. Smith, N. Randy, Circuit Judge, dissenting. My colleagues begin their opinion by giving lip service “to the reasoned and expert judgment of our colleagues in the [BIA], who we trust to carefully review the record and apply the law.” Maj. Op. at 5. Then, in order to reach the result they seek, they make no effort to apply the correct standard of review to the judgment of these experts they profess to trust. Instead, they substitute their judgment for that of the IJ and the BIA in order to reach their desired result, overturning the holdings of these “trusted” decision makers. The record demonstrates that the BIA and the IJ did not make any reversible mistakes. Instead, my colleagues make the following mistakes in determining whether Lapadat suffered past persecution. 1 1 I agree with the majority that Lapadat failed to challenge his CAT claim both before the BIA and here. Thus, the claim is forfeited.  LAPADAT V. BONDI 37 1. The majority first incorrectly limits its review to the BIA’s decision. In determining which underlying decision to review, my colleagues state that the “the BIA conduct[ed] its own review of the evidence and law, rather than adopting the IJ’s decision.” Maj. Op. at 12. Thus, they limited their review to the BIA’s decision. See id. at 12–13. In so doing, they allowed themselves to ignore the record and the reasoning behind the BIA’s decision. This narrow review was error. When one reads and reviews this decision, the BIA explicitly stated that it reviewed the IJ’s decision under the “clearly erroneous” standard of review. Although the BIA did not “adopt” the IJ’s decision, it expressly agreed with the IJ and cited to her opinion. In these situations, we “look to the IJ’s oral decision as a guide to what lay behind the BIA’s conclusion.” Tekle v. Mukasey, 533 F.3d 1044, 1051 (9th Cir. 2008). “In so doing, we review here the reasons explicitly identified by the BIA, and then examine the reasoning articulated in the IJ’s oral decision in support of those reasons.” Id.; see also Garcia-Martinez v. Sessions, 886 F.3d 1291, 1293 (9th Cir. 2018) (noting that when “the BIA agrees with the IJ’s reasoning, we review both decisions”); see also Bhattari v. Lynch, 835 F.3d 1037, 1042 (9th Cir. 2016) (reviewing both opinions when the BIA “agrees with and incorporates specific findings of the IJ while adding its own reasoning”). Moreover, even if “the BIA announced it was conducting de novo review,” our review is not limited to the BIA decision “when the BIA incorporates parts of the IJ’s reasoning,” including “by giving examples from it.” Szonyi v. Barr, 942 F.3d 874, 897 (9th Cir. 2019); see also Fon v. Garland, 34 F.4th 810, 815 (9th Cir. 2022) (explaining that we may look to the IJ’s decision because “[t]he BIA’s lack of analysis, 38 LAPADAT V. BONDI along with the citation to the IJ’s opinion, suggests that the BIA gave significant weight to the IJ’s findings” (internal quotation marks and citation omitted)). To justify their mistake regarding this review, my colleagues assert that they do not need to review the IJ’s decision, because the BIA did not “expressly adopt” the decision. Maj. Op. 12 n.4. However, this justification does “not hold water.” As the reviewing court, we look to whether the BIA adopted or incorporated an IJ’s decision in order to determine what portions of the IJ’s decision we may review. To suggest that we cannot look at the IJ’s “decision as a guide to what lay behind the BIA’s conclusion,” Tekle, 533 F.3d at 1051, would ignore the Supreme Court’s directive that “a reviewing court must ‘uphold’ even ‘a decision of less than ideal clarity if the agency’s path may reasonably be discerned,’” Garland v. Ming Dai, 593 U.S. 357, 369 (2021) (citation omitted). Notably, “reviewing courts remain bound by traditional administrative law principles, including the rule that judges generally must assess the lawfulness of an agency’s action in light of the explanations the agency offered for it rather than any ex post rationales a court can devise.” Id. 2. The majority next improperly substitutes its view for that of the IJ and the BIA in determining past persecution. While my colleagues cite to the proper standard of review for reviewing factual findings, Maj. Op. at 12–13, they fail to apply the standard they cite. To reiterate the proper standard of review: an appellate court reviews the IJ’s factual findings for substantial evidence. Kalulu v. Garland, 94 F.4th 1095, 1099 (9th Cir. 2024). “Substantial evidence is more than a mere scintilla and is such relevant evidence as  LAPADAT V. BONDI 39 a reasonable mind might accept as adequate to support a conclusion.” Rivera v. Mukasey, 508 F.3d 1271, 1274 (9th Cir. 2007). The substantial evidence standard of review is “deferential,” Parada v. Sessions, 902 F.3d 901, 909 (9th Cir. 2018), and a “stricter” standard of review than “clearly erroneous,” see Dickinson v. Zurko, 527 U.S. 150, 156 (1999). “The BIA’s determination that [Lapadat] was not eligible for asylum must be upheld if ‘supported by reasonable, substantial, and probative evidence on the record considered as a whole’ . . . and [that determination] can be reversed only if the evidence presented by [Lapadat] was such that a reasonable factfinder would have to conclude that the requisite fear of persecution existed.” INS v. Elias- Zacarias, 502 U.S. 478, 481 (1992). In other words, “the evidence [must] compel[] a contrary conclusion from that adopted by the BIA.” Parada, 902 F.3d at 909. Instead of applying this proper standard of review, my colleagues substitute their own judgment for that of the IJ and the BIA and make new determinations based on facts that were never argued by the Lapadats. Because these findings are dispositive to their decision, these mistakes are paramount. 3. The majority improperly reviews Lapadat’s claims. My colleagues make several errors in reviewing the BIA’s denial of Lapadat’s claim based on his failure to establish that the incidents he suffered rose to the level of persecution. First, my colleagues assess the harms suffered out of context by failing to tie them to persecution. Second, my colleagues conclude that the kidnapping incident rises to the level of persecution individually. Third, my colleagues conclude that the shooting incident rises to the level of 40 LAPADAT V. BONDI persecution individually. Finally, my colleagues conclude that cumulatively the incidents suffered rise to the level of persecution. A. Incidents must be tied to persecution. An applicant for asylum must have been subjected to past “persecution” or have “a well-founded fear of [future] persecution.” 8 U.S.C. § 1101(a)(42). In order to establish eligibility for asylum, an applicant “who seeks to demonstrate that []he was persecuted in the past must prove (1) that []he was the victim of ‘an incident, or incidents, that rise to the level of persecution’; (2) that the persecution was ‘on account of’ one of the protected grounds; and (3) that such persecution was ‘committed by the government or forces the government is either unable or unwilling to control.’” Parussimova v. Mukasey, 555 F.3d 734, 738 (9th Cir. 2009) (quoting Navas v. INS, 217 F.3d 646, 655–56 (9th Cir. 2000)). In this case, the BIA and the IJ concluded that Lapadat was not eligible for asylum, because he failed to meet his burden of showing the incidents rose to the level of persecution. My colleagues do not assess the validity of these findings under the substantial evidence standard of review but instead merely assert that the harms suffered, by themselves, compel a different conclusion. In so concluding, my colleagues apply too broad of a reading to the first prong. We do not consider every harm that has been subjected upon an applicant regardless of context. Instead, we only consider the “incidents” that could constitute “persecution.” See Navas, 217 F.3d at 655. We have “defined persecution as the infliction of suffering or harm upon those who differ (in race, religion or political opinion) in a way regarded as offensive.” Korablina  LAPADAT V. BONDI 41 v. INS, 158 F.3d 1038, 1043 (9th Cir. 1998) (emphasis added) (internal quotation marks and citation omitted). We have also recognized that persecution “is an extreme concept that means something considerably more than discrimination or harassment.” Sharma v. Garland, 9 F.4th 1052, 1060 (9th Cir. 2021). Thus, not all incidents “qualify as persecution, despite the fact that such conditions have caused the petitioners some harm.” Mihalev v. Ashcroft, 388 F.3d 722, 729 (9th Cir. 2004). Accordingly, before an IJ determines whether the “incident” can rise to the level of persecution, he or she must necessarily first determine whether “the infliction of suffering or harm” arose because the applicant differed from the persecutor. In other words, for an incident to rise to the level of persecution, some evidence must exist that shows the incident was not the result of civil unrest or criminal activity.2 See Gormley v. Ashcroft, 364 F.3d 1172, 1177 (9th Cir. 2004) (holding that the two 2 My colleagues agree that not all incidents qualify as persecution. Maj. Op. at 26. Yet, they assert that context does not matter, disregarding Gormley’s clear holding. They attempt to distinguish Gormley by asserting that “targeted shootings and attempted kidnappings” are not “all too-common general conditions.” Maj. Op. at 26. First, there is no evidence in this record that Lapadat’s shooting was targeted. Lapadat was admittedly trespassing on private property when he was shot. Moreover, this court has rejected social groups of persons who “are subject to kidnappings and extortion,” because asylum cannot be based on “fear of harm resulting from general conditions of violence and civil unrest affecting the home country’s populace as a whole.” Alanniz v. Barr, 924 F.3d 1061, 1064–65 (9th Cir. 2019). In this case, country condition evidence noted that “[i]n 2016 the Directorate for Investigating Organized Crime and Terrorism (DIICOT) uncovered a human trafficking ring that had forced its kidnapped victims, including children, into beggary, slavery, and other forms of forced labor.” Thus, substantial evidence would support the agency’s conclusion that the attempted kidnapping was a criminal act, not persecution. 42 LAPADAT V. BONDI criminal attacks petitioner suffered did not rise to level of persecution because “robberies of this sort are an all too common byproduct of civil unrest and economic turmoil”); see also Matter of M-E-V-G-, 26 I. & N. Dec. 227, 235 (BIA 2014) (explaining that “asylum and refugee laws do not protect people from general conditions of strife, such as crime and other societal afflictions”); Ming Ming Wijono v. Gonzales, 439 F.3d 868, 872 (8th Cir. 2006) (“Low-level intimidation and harassment alone do not rise to the level of persecution, nor does harm arising from general conditions such as anarchy, civil war, or mob violence ordinarily support a claim of persecution.” (citation omitted)); Lie v. Ashcroft, 396 F.3d 530, 532 (3d Cir. 2005) (superseded in unrelated part by 8 U.S.C. § 1158(b)(1)(B)(I)) (holding that “substantial evidence supports the BIA’s conclusion that these robberies were not motivated by religion or ethnicity, and that, at all events, such robberies were not sufficiently severe so as to rise to the level of persecution”). To be sure, a criminal act by its very nature cannot constitute persecution, because it bears no nexus to a protected ground. 3 See Zetino v. Holder, 622 F.3d 1007, 1016 (9th Cir. 3 My colleagues claim that assessing harm under the lens of persecution conflates the two prongs of the past persecution analysis. Maj. Op. at 19 n.7. Not so. Our precedent is clear: each prong is tied to persecution. See Parussimova, 555 F.3d at 738. “[W]e do not look to the level of harm experienced by the petitioner,” but rather we look at “whether the treatment the victim received rises to the level of persecution.” Kaur v. Wilkinson, 986 F.3d 1216, 1226 (9th Cir. 2021) (cleaned up). “In other words, it is the conduct of the persecutor . . . that matters for purposes of determining what constitutes persecution.” Id. Here, my colleagues do not consider the conduct of the persecutor, because they do not care who caused the harm. Under their theory, it is merely the level of harm that is assessed in the first prong without any context. Assessing harm in the  LAPADAT V. BONDI 43 2010) (“An [applicant’s] desire to be free from harassment by criminals motivated by theft or random violence by gang members bears no nexus to a protected ground.”). Thus, in order for an IJ to properly assess whether an applicant suffered from past persecution, the IJ cannot ignore the cause of or reason for the harm when determining whether the incidents rose to the level of persecution. See Beskovic v. Gonzales, 467 F.3d 223, 226 (2d Cir. 2006). In assessing whether physical mistreatment amounts to persecution, the Second Circuit explained that “persecution is the infliction of suffering or harm upon those who differ on the basis of a protected statutory ground, while harassment is words, conduct, or action usually repeated or persistent that, being directed at a specific person, annoys, alarms, or causes substantial emotional distress in that person and serves no legitimate purpose.” Id. at 225–26 (cleaned up). The court further explained that “the difference between harassment and persecution is necessarily one of degree, [and] the degree must be assessed with regard to the context in which the mistreatment occurs.” Id. at 226 (internal quotation marks and citation omitted). The Second Circuit cautioned the BIA to “be keenly sensitive to the fact that a ‘minor beating’ or, for that matter, any physical degradation designed to cause pain, humiliation, or other suffering, may rise to the level of persecution if it occurred in the context of an arrest or detention on the basis of a context of persecution does not merge the two prongs; a separate analysis is still required to determine whether the “protected ground represent[s] ‘one central reason’ for an asylum applicant’s persecution.” See Parussimova, 555 F.3d at 740. 44 LAPADAT V. BONDI protected ground.” 4 Id. Notably, asylum is limited to persons subjected to persecution; thus, allowing any incident of random violence to be included in an analysis would unnecessarily burden the agency in assessing claims for asylum. Moreover, it creates an unworkable standard. B. Attempting kidnapping incident was unrelated to a claim of persecution. Lapadat challenged the BIA’s exclusion of the attempted kidnapping incident, 5 claiming that the incident was not a criminal act but that his daughters were targeted on account of their Roma ethnicity. The BIA upheld the IJ’s findings noting that the incident did not rise to the level of persecution. The BIA recognized the IJ’s finding that the incident was caused by unknown individuals. As the IJ explained, the evidence showed “that the [attempted] kidnapping [was] . . . essentially a crime, and though [it was] abhorrent in nature, . . . [it did not] rise to the level of persecution under 9th Circuit case law.” 4 My colleagues attempt to minimize the language in Beskovic, arguing that the court does not “provide a general rule allowing courts to downgrade or disregard severe harm” because there is no connection to a protected ground. Maj. Op. at 21 n.8. Beskovic is clear: “the context in which the mistreatment occurs” is a necessary factor to consider in assessing whether physical mistreatment amounts to persecution. 467 F.3d at 226. Based on my colleagues interpretation, context does not matter. Thus, any person who credibly testifies that he or she was shot (accidentally or for some other reason) or subject to an attempted kidnapping (even if a purely criminal act) automatically meets prong one. 5 During the attempting kidnapping incident, Lapadat’s wife was threatened with death if she continued to yell and was cut with a knife on her back. Lapadat’s daughters were also threatened with rape.  LAPADAT V. BONDI 45 Although the facts surrounding a particular incident maybe appalling, persecution may not be found unless there is sufficient evidence in the record that would compel the conclusion that the incident was anything but a criminal act. See Bromfield v. Mukasey, 543 F.3d 1071, 1076–77 (9th Cir. 2008) (“If the perpetrator is motivated by his victim’s protected status . . . he is engaging in persecution, not random violence.”). Here, Lapadat’s wife and daughter testified that they did not know who the assailants were. In fact, Lapadat’s wife did not discount the possibility that the assailants themselves could have been “gypsies.” During the attack and subsequent threats, the assailants did not use ethnic slurs or make any indication that they were targeting the Lapadats based upon their Roma ethnicity or any other protected ground. Thus, this purely criminal act cannot be considered in Lapadat’s persecution claim. See Gormley, 364 F.3d at 1177. My colleagues ignore this record evidence and only consider the criminal acts out of context to conclude that the incident rose to the level of persecution. 6 But the record is clear there was no testimony that established the incident 6 Despite my colleagues’ claims that context does not matter, they nevertheless include it in their analysis. They claim that Lapadat was shot because he was a gypsy (ignoring the IJ’s findings that Lapadat was shot because he was admittedly trespassing). Maj. Op. at 25. They also attempt to sidestep the IJ’s other findings that the shooting was an isolated incident and the attempted kidnapping was a criminal act by arguing that the shooting and attempted kidnapping “did not occur in isolation” but were ongoing acts of persecution, because “the Lapadats, like many Romani people, had been mistreated by the police and Romanian public for years because of their ethnicity.” See Maj. Op. at 24. These conclusions improperly disregard the IJ’s findings of fact, and undermine my colleagues assertion that the harm suffered should be assessed without context. 46 LAPADAT V. BONDI was related to a claim of persecution. Accordingly, this incident cannot rise to the level of persecution. C. Shooting/trespassing incident was an isolated incident. Lapadat argues that the BIA erred because it did not consider the shooting incident. 7 Although the BIA did not specifically address this incident in its opinion, it did cite to the IJ’s decision, Lapadat’s testimony before the IJ, and Lapadat’s asylum application, all which included this shooting incident. Thus, we are required to presume the BIA considered this evidence. Cf. Larita-Martinez v. INS, 220 F.3d 1092, 1095 (9th Cir. 2000) (“We embrace the view of our sister circuits and hold that an alien attempting to establish that the [BIA] violated his right to due process by failing to consider relevant evidence must overcome the presumption that it did review the evidence.”). Moreover, in circumstances such as these, we look to the IJ’s decision to see what lies behind the BIA’s conclusion. See Tekle, 533 F.3d at 1051. Here, the IJ found that the shooting incident was isolated. Lapadat admits he was shot while he was trespassing. Although the individual (who shot Lapadat) also shouted a racial epithet, substantial evidence supports the IJ’s conclusion that the incident was isolated. Notably, the incident occurred over eight years before Lapadat fled to the 7 For the first time on appeal to us, Lapadat characterized the shooting incident as an “attempted murder.” Before the IJ and the BIA, Lapadat testified that the man “shot toward” or “shot at” him. Lapadat’s new characterization has a much different connotation from that to which he testified and which we are now reviewing. Thus, we should not consider these newly raised assertions on appeal, which he uses merely to enhance his claim here.  LAPADAT V. BONDI 47 United States, and there were no other incidents with this individual. See Sharma, 9 F.4th at 1061 (explaining that “sporadic incidents, unaccompanied by an ongoing pattern of harm” are less likely to rise to the level of persecution). Although my colleagues and Lapadat attempt to enhance this claim on appeal, Lapadat’s actual argument about the incident is revealing. Lapadat asserts that the trespassing/shooting incident “cannot be excluded from the realm of possibility that the Romanian aggressor shot the Lead Petitioner knowing there would not likely be any consequences; and, the police would not care because his victim was a gypsy.” This characterization does not contest the fact that he was trespassing; moreover, it does not suggest an “attempted murder,” because Lapadat was Roma. Instead, this argument merely provides a different interpretation of the facts than those found by the IJ, for which we cannot substitute our judgment. Cf. Lianhua Jiang v. Holder, 754 F.3d 733, 740 (9th Cir. 2014), overruled on other grounds by Alam v. Garland, 11 F.4th 1133 (9th Cir. 2021) (“Taking the position suggested by [petitioner] would require that we supplant the IJ’s . . . determination with our own, as if we were conducting a de novo review.”). Although “significant physical violence” may compel a finding of past persecution, past persecution is only found when coupled with “other indicators of persecution.” Sharma, 9 F.4th at 1061. Thus, we must consider “whether the petitioner’s harm was an isolated incident or, conversely, part of an ongoing pattern of serious maltreatment.” Id. In this case, the IJ found that the incident was exactly that: an isolated event caused by Lapadat’s trespassing. Notably, this shooting was the first and only incident when Lapadat was physically harmed and required medical attention. Accordingly, the record supports the IJ’s finding that the 48 LAPADAT V. BONDI shooting was caused by Lapadat’s trespass and was isolated event. Thus, we have no basis for reversal. Dong v. Garland, 50 F.4th 1291, 1300 (9th Cir. 2022); see also Sharma, 9 F.4th at 1061 (noting that “serious maltreatment that is sustained and recurring is more likely to compel the conclusion of past persecution”). Yet, ignoring this standard of review, my colleagues instead substitute their judgment for that of the agency. 8 My colleagues do not explain why the IJ’s depiction of this incident compels a different conclusion. Maj. Op. at 17. Instead, they presume that, just because Lapadat was physically harmed, the incident rises to the level of persecution. However, this presumption ignores our standard of review. Although, as an IJ, one may have concluded that an isolated shooting incident rises to the level of persecution, Maj. Op. at 24 n.11, that is not our decision to make on appeal, and the record does not compel it, see Aden v. Wilkinson, 989 F.3d 1073, 1082 (9th Cir. 2021) (holding that even if “a reasonable factfinder could conclude” the harm rose to the level of persecution, the record “did not compel” it). As the IJ concluded, the act of shooting Lapadat for trespassing was “offensive”; however, Lapadat did not present sufficient evidence that would compel a conclusion that the shooting was more than an isolated act of violence. 8 My colleagues argue that the BIA committed “legal error” because it did not mention the shooting incident. Maj. Op. at 17. However, the BIA did not commit legal error. Before the BIA (like here), Lapadat argued that cumulatively the harm suffered amounted to past persecution. With regard to this incident, Lapadat challenged the IJ’s finding that Lapadat was shot because he was trespassing. Moreover, as addressed above, the BIA cited to the IJ’s decision regarding why the incidents did not rise to the level of persecution, thus we may look at the IJ’s reasoning.  LAPADAT V. BONDI 49 Accordingly, this incident individually does not rise to the level of persecution. D. Cumulatively, the record does not compel a conclusion of harm rising to the level of persecution. Lapadat argues that the incidents of harm cumulatively rise to the level of past persecution. 9 Lapadat and his family testified that they experienced the following incidents of persecution: 1. The Lapadats lived in segregated areas where Roma live. 2. Romas were unwelcome and unable to live outside major Roma population centers because of entrenched segregation enforced by social convention. Lapadat claimed he has been denied entry into restaurants and stores. 3. Lapadat was excluded from formal employment market. 4. In 2005, Lapadat was unable to obtain prompt healthcare for an appendicitis. 9 Lapadat did not argue that the kidnapping or shooting incident would individually rise to the level of persecution. Rather, he only argued that the BIA erred in not finding the harms cumulatively rose to the level of persecution. Nevertheless, my colleagues ignore Lapadat’s argument and assess the shooting and attempted kidnapping incidents individually. See Maj. Op. at 16–18. This determination only emphasizes the error made by my colleagues. See Greenwood v. FAA, 28 F.3d 971, 977 (9th Cir. 1994) (holding that “[w]e will not manufacture arguments for an appellant”). 50 LAPADAT V. BONDI 5. In 2007, Lapadat was shot and called “gypsy” after he trespassed on private property. 6. In 2011, Lapadat was stopped by a policeman who temporarily detained, kicked, and slapped him. 7. In 2012, Lapadat’s daughters were victims of attempted kidnapping by unknown assailants.10 Lapadat’s wife was injured when she thwarted the attempted kidnapping. Thereafter, Lapadat’s daughters were threatened with rape if the Lapadats reported the attempting kidnapping incident to the police. 8. In 2015, Lapadat was hit in the chest when he attempted to access a pool. 9. In 2015, Lapadat had an incident with policeman, Chocolo, wherein he lightly slapped him and kicked him. My colleagues characterize the IJ as having rejected “a laundry list of abuse (like being denied healthcare, work, and access to public pools and restaurants), as well as criminal acts, police beatings, and threats.” Maj. Op. at 18. But, even considered cumulatively, the events do not compel a conclusion of past persecution. “Cumulative-effect review is essential where a single isolated incident may not rise to the level of persecution, but the cumulative effect of several incidents may constitute persecution.” Salguero Sosa v. 10 Lapadat’s wife testified that she did not know if the unknown assailants were “Romanian or gypsies,” because “[y]ou can’t tell bad guys now.”  LAPADAT V. BONDI 51 Garland, 55 F.4th 1213, 1218 (9th Cir. 2022) (internal quotation marks and citation omitted). The BIA properly considered the cumulative effect of all relevant harm when assessing Lapadat’s persecution claim. See id. at 1218–19. The BIA “agree[d] with the [IJ] that [Lapadat] did not demonstrate that he experienced past persecution as his past experiences do not together rise to the level of persecution.”11 As explained above, the attempted kidnapping incident was a criminal act that does not establish persecution in the cumulative harm analysis.12 See Gormley, 364 F.3d at 1177 (“Random, isolated criminal acts perpetrated by anonymous [persons] do not establish persecution.”). The remainder of the incidents claimed by Lapadat were isolated or amount 11 As my colleagues recognized, our case law has tension with regard to whether de novo or a substantial evidence standard of review applies to whether past harm rises to the level of persecution. Maj. Op. at 14–15. That said, I agree with my colleagues that we need not resolve that tension here, because the result is the same. Under either de novo review or substantial evidence review, we are not allowed “to substitute our view of the matter for that of the [BIA].” Gu v. Gonzales, 454 F.3d 1014, 1020 (9th Cir. 2006) (citation omitted). Thus, because the IJ found that the shooting incident was an isolated event caused by Lapadat’s trespassing, and the kidnapping incident was a criminal act, neither incident would not compel a conclusion that Lapadat suffered from persecution under either standard of review. Moreover, the cumulative harm also would not compel a different conclusion under either standard of review. 12 My colleagues also improperly reference the subjective harm the Lapadats continue to suffer. See Maj. Op. at 21–22. However, we have made it clear that “‘it is the conduct of the persecutor’ that is relevant to evaluating whether past treatment rises to the level of persecution—not ‘the level of harm’ or ‘subjective suffering’ the petitioner experienced.” Flores Molina v. Garland, 37 F.4th 626, 636 (9th Cir. 2022) (quoting Kaur, 986 F.3d at 1226). 52 LAPADAT V. BONDI only to discrimination or harassment. See Ghaly v. INS, 58 F.3d 1425, 1431 (9th Cir. 1995) (“Discrimination . . . as morally reprehensible as it may be, does not ordinarily amount to ‘persecution’ within the meaning of the Act.”). Although “[s]evere and sustained discrimination, or discrimination in combination with other harms . . . may compel a finding of past persecution,” Wakkary v. Holder, 558 F.3d 1049, 1059 (9th Cir. 2009), the IJ and the BIA determined that the harm suffered here is not sufficient to rise to the level of persecution. Lapadat’s exclusion from formal employment does not support a claim of persecution. Although “substantial economic deprivation that constitutes a threat to life or freedom can constitute persecution[,] . . . mere economic disadvantage alone, does not rise to the level of persecution.” Zehatye v. Gonzales, 453 F.3d 1182, 1186 (9th Cir. 2006) (internal quotation marks and citation omitted). Here, even though Lapadat could not enter the formal workforce in Romania, Lapadat testified that he earned a living working for other Roma citizens. Hence, he was not prevented from earning a living. Lapadat’s employment restrictions, while discriminatory, are not significant in comparison to our cases that have concluded that there was substantial economic deprivation. See, e.g., Chand v. INS, 222 F.3d 1066, 1074 (9th Cir. 2000) (citing cases). Moreover, Lapadat lived in France for two of these years; thus, Lapadat’s employment status during this time is not relevant to his persecution claim. The record does not show that Lapadat experienced escalating harm over the last several years, Flores Molina, 37 F.4th at 634, nor does it show that there was an “ongoing pattern of serious maltreatment,” Sharma, 9 F.4th at 1061. Since the isolated trespassing/shooting incident in 2007,  LAPADAT V. BONDI 53 Lapadat did not experience any incidents that were anything beyond mere discrimination and harassment. My colleagues know that, absent the inclusion of the shooting and kidnapping incidents, Lapadat only experienced discrimination or harassment, which is insufficient to establish persecution. Sharma, 9 F.4th at 1060. However, my colleagues avoid the standard of review and our precedent, instead looking at the incidents in a vacuum, ignoring the purpose of asylum claims. My colleagues’ reliance on Korablina, 158 F.3d 1038 (death threats because of her religion), Baballah v. Ashcroft, 367 F.3d 1067 (9th Cir. 2004) (threats, physical violence, and economic hardship based on ethnicity and religion), and Kaur, 986 F.3d 1216 (attempted rape based on political activities), to conclude these events compel a conclusion of past persecution is misplaced. In all these cases, the harms were not isolated and were linked to a protected ground. My colleagues’ reliance on these cases highlight why their broad focus on harm is unworkable. They suggested that the prongs “may bleed together,” but that they can still rely on these cases to assess prong one only. I agree the prongs “bleed together,” because they must: whether an incident rises to the level of persecution, cannot exist without some evidence of “persecution.” Nevertheless, the fact that the harm suffered in Korablina, Baballah, and Kaur was not isolated and was linked to a protected ground established that the incidents rose to the level of persecution. In sum, a proper review of the agency’s decision requires us to determine whether substantial evidence in record supported its findings. My colleagues do not review the agency’s findings in this deferential framework. If they had, 54 LAPADAT V. BONDI they would have held that substantial evidence in the record supports these factual findings. 4. Even assuming a disfavored group, the majority improperly relies on its previous fact finding to support a claim of future persecution. Absent past persecution, the BIA must nevertheless consider whether Lapadat has a well-founded fear of persecution. Lapadat may establish a well-founded fear of future persecution by “prov[ing] that []he is a member of a ‘disfavored group’ coupled with a showing that []he, in particular, is likely to be targeted as a member of that group.” Sael v. Ashcroft, 386 F.3d 922, 925 (9th Cir. 2004) (citation omitted). My colleagues conclude that Lapadat has such a fear, because he is a member of a disfavored group of Roma citizens and, based on his past harm, he established an individualized risk of future harm. Maj. Op. at 28–35. However, even assuming that Roma people are a disfavored group in Romania, the record does not compel the conclusion that Lapadat has “an individualized threat” of future persecution. Najmabadi v. Holder, 597 F.3d 983, 992 (9th Cir. 2010) (emphasis in original). The BIA concluded that Lapadat did not have an objectively reasonable fear of future harm, because his past harm was based on “isolated incidents caused by different individuals.” The BIA further concluded that there was insufficient evidence that any of those individuals were likely to target Lapadat in the future. These conclusions are supported by substantial evidence. My colleagues ignore these conclusions and instead claim that they will now “leave it to the agency to determine” whether Lapadat can meet the objective prong of a well-  LAPADAT V. BONDI 55 founded fear showing under a disfavored group analysis.13 Maj. Op. at 35. However, the agency has already made such an alternative finding; my colleagues just fail to address the agency’s conclusions in this regard.14 The BIA was clear that Lapadat did not meet this lower standard: We also agree with the Immigration Judge that, even if Roma constituted a disfavored group, the lead respondent has not established that he has a sufficient individualized risk of future persecution to make his fear of return to Romania objectively reasonable (IJ at 8; Respondent’s Br. at 12-13). Most of the harm experienced by the lead respondent occurred before he moved to France in 2013 (IJ at 8). After returning to Romania in 2015, the lead respondent experienced an isolated altercation at a swimming pool and another 13 My colleagues nevertheless attempt to bolster its opinion and Lapadats claim, arguing that Lapadat is able to show a well-founded fear of future persecution by dramatically reciting facts not established by the IJ. See Maj. Op. at 6–10, 27 n.12. Notably, my colleagues mischaracterize the evidence in the record by (1) asserting that Lapadat was shot because he was a “gypsy,” id. at 25, (2) implying that the attempted kidnapping was because of the Lapadats’ daughters ethnicity, id. at 7, and severely overstate Lapadat’s encounter with a policeman, id. at 9–10. Further, my colleagues ignore the fact that the Lapadats left and returned to Romania freely and without incident. Despite my colleague’s desire for the facts to be different, “[w]e are required to accept administrative findings of fact unless any reasonable adjudicator would be compelled to conclude to the contrary.” Dong, 50 F.4th at 1299–1300. 14 My colleagues avoid this issue, because the BIA must first determine past persecution before it can properly assess a well-founded fear of future persecution. See Maj. Op. at 35 n.14. 56 LAPADAT V. BONDI brief encounter with a police officer who lightly slapped his head kicked his buttocks (IJ at 8). Although evidence of past harm is relevant to whether an individual’s fear of future persecution is objectively reasonable, all of the harm described by the lead respondent were isolated incidents caused by different individuals. See Sael, 386 F.3d at 928–29. There is insufficient evidence in the record that any of these individuals are looking for the lead respondent or would be inclined to single out and target him for persecution in the future. In light of the foregoing, the lead respondent has not shown a sufficient individualized risk to establish a well-founded fear of persecution under a disfavored group analysis (IJ at 8). As a result, the lead respondent and his derivative beneficiaries are ineligible for asylum. Even though Lapadat need only “demonstrate a ‘comparatively low’ level of individualized risk in order to prove that []he has a well-founded fear of future persecution,” Sael, 386 F.3d at 927, he did not present any evidence that any individual (whom he encountered) was still looking for him or would target him for future persecution. Thus, there is insufficient evidence to compel the conclusion that he will be persecuted upon his return to Romania.