Eric Nyandwi v. MERRICK B. GARLAND , Attorney General of the United States

In the
United States Court of Appeals
For the Seventh Circuit
____________________
No. 20‐3215
ERIC NYANDWI,
Petitioner,
v.
MERRICK B. G ARLAND ,
Attorney General of the United States,
Respondent.
____________________
On Petition for Review of an Order of the
Board of Immigration Appeals
No. A094‐731‐778
____________________
A RGUED SEPTEMBER 17, 2021 — D ECIDED OCTOBER 8, 2021
____________________
Before SYKES , Chief Judge, and FLAUM and K IRSCH , Circuit
Judges.
KIRSCH, Circuit Judge. Eric Nyandwi, facing removal from
the United States because of multiple felony convictions, ap‐
plied for a deferral of removal under the Convention Against
Torture. He claimed that he faced a substantial risk of torture
if returned to the country of Burundi, of which he is a citizen.
Both an immigration judge and the Board of Immigration

-- 1 of 11 --

2 No. 20‐3215
Appeals found no such substantial risk and so denied his ap‐
plication. In this petition for review, Nyandwi asks us to re‐
mand the case, arguing that the immigration judge and the
Board committed various legal errors when they denied his
application. We disagree, find no error, and therefore deny
the petition.
I
Nyandwi, a citizen of Burundi and a native of Tanzania,
came to the United States as a refugee on August 9, 2006 and
became a lawful permanent resident. After Nyandwi was
convicted of robbery in the second degree, receiving a stolen
firearm, and illegal possession of a controlled substance, the
Department of Homeland Security began removal proceed‐
ings against him under 8 U.S.C. § 1227(a)(2)(A)(iii), (B)(i).
Nyandwi conceded his removability as an aggravated felon
but filed an Application for Asylum and Withholding of Re‐
moval, seeking a deferral under the Convention Against Tor‐
ture (CAT). At a removal hearing before an immigration
judge (IJ), Nyandwi submitted evidence of country conditions
in Burundi and testified that he was an ethnic Twa whose par‐
ents fled Burundi in 1996 because of a civil war that resulted
in the death of Twas, including his family members. Nyandwi
told the IJ that he feared being removed to Burundi because
he was a returning Twa refugee, was unable to speak the na‐
tive language, had no proof of political allegiance to the gov‐
erning regime, and was unable to pay compulsory election
contributions.
The IJ denied Nyandwi’s claim for deferral under CAT,
concluding that “Respondent has not established a substan‐
tial risk that he will be targeted for torture if he returns to Bu‐
rundi” because he did not claim that anyone in Burundi had

-- 2 of 11 --

No. 20‐3215 3
a current intent to torture him specifically. Rather, his fears
relied on what the IJ called a hypothetical chain of events: that
he would be detained by authorities upon return to Burundi
because he was a refugee, viewed as an oppositionist to the
ruling regime because he was a Twa and financially incapable
of paying compulsory election contributions, and therefore
singled out for torture. The IJ faulted Nyandwi for failing to
show that he would be unable to make money or make elec‐
tion contributions in Burundi and for failing to present any
specific cases of persons in Nyandwi’s position as a returning
Twa refugee being tortured in Burundi.
Nyandwi appealed to the Board of Immigration Appeals
(BIA), claiming that the IJ had failed to use the correct legal
standard and had ignored relevant evidence. The BIA af‐
firmed, finding no error in the IJ’s “careful analysis of the
component parts of a holistic claim.” Nyandwi timely filed
this petition for review.
II
Because the BIA adopted and supplemented the IJ’s deci‐
sion, we review the IJ’s decision supplemented by the Board’s
reasoning. Herrera‐Garcia v. Barr, 918 F.3d 558, 561 (7th Cir.
2019). We review questions of law de novo. Lozano‐Zuniga v.
Lynch, 832 F.3d 822, 826 (7th Cir. 2016). We will remand for
further consideration when the agency commits legal errors
such as using an incorrect legal standard or overlooking key
evidence that could have resulted in a different conclusion.
Sirbu v. Holder, 718 F.3d 655, 656 (7th Cir. 2013); Lam v. Holder,
698 F.3d 529, 533–34 (7th Cir. 2012).
Nyandwi first asserts that the IJ committed legal error by
failing to consider in the aggregate three risk factors that

-- 3 of 11 --

4 No. 20‐3215
Nyandwi had raised: (1) Nyandwi is Twa; (2) Nyandwi is a
returning refugee; and (3) Nyandwi is likely to be targeted as
an oppositionist due to his inability to pay the ruling party’s
compulsory election contributions. According to Nyandwi,
these three reasons combine to render him particularly sus‐
ceptible to torture. Because the IJ never considered that com‐
bination explicitly, Nyandwi argues, the IJ committed legal
error. And, says Nyandwi, the Board did not cure this sup‐
posed error when it affirmed the IJ’s decision. Second,
Nyandwi alleges that the IJ erred in demanding corroborating
evidence of specific cases in which returning refugees have
been tortured and demanding corroborating evidence that
Nyandwi would be unable to earn money in Burundi. Be‐
cause the Burundi regime hides evidence of such specific
cases, making them unavailable, and because Nyandwi
lacked personal knowledge of Burundi and was in detention,
Nyandwi claims that it was impossible for him to provide this
information. Finally, Nyandwi argues the IJ and BIA commit‐
ted legal error by ignoring three main pieces of evidence: (1)
evidence that Burundi officials extort bribes disguised as
mandatory election contributions; (2) evidence of Nyandwi’s
characteristics that could make Nyandwi seem like an oppo‐
sitionist: his inability to speak the language, his identifiable
Twa ethnicity, and the history of familial death in the earlier
conflict; and (3) general evidence about the authoritarian
tendencies of the Burundi government and that similarly‐sit‐
uated people have been tortured.
A
We first consider whether the IJ erred in its analysis of the
risk factors put forth by Nyandwi. To determine whether the
IJ committed error, we must first establish the legal standard

-- 4 of 11 --

No. 20‐3215 5
the IJ was bound to follow. Both parties agree that agencies
should use the aggregate risk legal standard, which requires
that claims under CAT be considered in terms of the aggre‐
gate risk of torture and not only as separate, divisible claims.
See Matter of J‐R‐G‐P‐, 27 I. & N. Dec. 482, 484 (BIA 2018). Five
of our sister circuits have adopted this standard. See Kamara
v. Attorney General, 420 F.3d 202, 213–14 (3d Cir. 2005); Qui‐
jada‐Aguilar v. Lynch, 799 F.3d 1303, 1308 (9th Cir. 2015); Ro‐
driguez‐Arias v. Whitaker, 915 F.3d 968, 972–73 (4th Cir. 2019);
Marqus v. Barr, 968 F.3d 583, 589 (6th Cir. 2020); Abdi Omar v.
Barr, 962 F.3d 1061, 1065 (8th Cir. 2020). We see no reason to
disagree with the parties, the BIA, or other circuits, so we
adopt the aggregate risk approach for the determination of
substantial risk and hold that the agency may address risk fac‐
tors individually so long as it considers all sources of and rea‐
sons for risk cumulatively to determine whether there is a
substantial risk of torture.
Nyandwi urges us to follow the leads of the Third and
Fourth Circuits and take a quantitative or statistical approach
to calculating aggregate risk. Those circuits have adopted
such an approach to calculate the risk of torture an alien faces
from different sources. The Third Circuit has held that an al‐
ien facing removal “is entitled to CAT protection if he is able
to demonstrate that the cumulative probability of torture by
[multiple sources] exceeds 50%.” Kamara, 420 F.3d at 213–14.
To calculate the sum of the weighted probability of torture by
each potential source of torture, the court devised a formula
for the adding of probabilities for mutually exclusive events.1
1 Agencies were instructed to calculate the risk of torture from one source
by multiplying the percentage chance of the alien’s capture by that source
by the percentage chance of torture by that source. Kamara v. Attorney

-- 5 of 11 --

6 No. 20‐3215
Id. at 214. The Fourth Circuit has followed the Third Circuit’s
approach, faulting an IJ for failing to “consider the aggregated
risk caused by all three entities in unison by adding the prob‐
ability of torture from each entity and determining whether
that sum exceeded 50%.” Rodriguez‐Arias, 915 F.3d at 973. But,
as far as we can tell, those two circuits have not applied that
approach to calculate the risk of torture an alien faces due to
different reasons for risk. Nyandwi’s theory would be novel
even under the statistical approach of the Third and Fourth
Circuits.2
Either way, we decline to follow the Third and Fourth Cir‐
cuits’ quantitative or statistical approach to aggregate risk. We
do not think this is the appropriate methodology for
General, 420 F.3d 202, 214 (3d Cir. 2005). Then, the agencies were told to
add together the percentage risk from each source to calculate aggregate
risk. Id. The Third Circuit went on to suggest that agencies may need to
calculate risk based on non‐mutually exclusive events, for which the for‐
mula for calculating the overall risk of torture would be “(the weighted
probability of [petitioner] being tortured by [Source A], if he finds himself
only in [Source A] territories) + (the weighted probability of [petitioner]
being tortured by the [Source B], if he finds himself only in [Source B] con‐
trolled territories) + (the weighted probability of [petitioner] being tor‐
tured by either, or both, [Source A] and [Source B] if [petitioner] finds him‐
self in both [Source A] and [Source B] territories).” Id. at 214 n.10.
2 Although the Sixth Circuit has described the Third and Fourth Circuits’
quantitative approach as applying to “the cumulative probability of tor‐
ture by [all of the] entities, or for all reasons,” even that case’s facts con‐
cerned threats from different sources, not threats for different reasons, and
the Sixth Circuit did not apply the formulas described by the Third Circuit
in Kamara, instead posing the question as whether the cumulative risks of
torture exceed 50%. Marqus v. Barr, 968 F.3d 583, 589 (6th Cir. 2020)
(cleaned up).

-- 6 of 11 --

No. 20‐3215 7
determining substantial risk. We have already held that “sub‐
stantial risk” under CAT is a “non‐quantitative restatement”
of the “more likely than not” standard. Perez‐Montes v. Ses‐
sions, 880 F.3d 849, 850 (7th Cir. 2018). Adding percentages is
not a tenable method to determine substantial risk because
“[t]he data and statistical methodology that would enable a
percentage to be attached to a risk of torture simply do not
exist.” Rodriguez‐Molinero v. Lynch, 808 F.3d 1134, 1135 (7th
Cir. 2015); see also Perez‐Montes, 880 F.3d at 850 (“[A] statisti‐
cal requirement cannot be taken seriously and . . . the best an
agency or court can do is look for substantial risk”). Taking a
holistic, non‐quantitative approach to the aggregation of fac‐
tors accords with our requirements for agencies and judges in
other contexts. In reviewing Social Security claims, for exam‐
ple, we have held that “an ALJ is required to consider the ag‐
gregate effects of a claimant’s impairments” but have not
taken a quantitative approach, finding it sufficient where an
ALJ “stated that he had considered all of [the claimant’s]
symptoms together.” Getch v. Astrue, 539 F.3d 473, 483 (7th
Cir. 2008). Similarly, in reviewing whether probable cause ex‐
isted for warrants, we have not assigned a percentage thresh‐
old to the requirement of “fair probability.” See, e.g., United
States v. Orr, 969 F.3d 732, 736 (7th Cir. 2020) (quoting Illinois
v. Gates, 462 U.S. 213, 238 (1983)). And in determining proba‐
ble cause, factors must be analyzed in “totality,” but we have
never attached a statistical methodology for summing up the
various factors. See id.; see also Rodriguez‐Molinero, 808 F.3d at
1135.
A similar holistic methodology is appropriate for the ag‐
gregate risk standard for CAT claims. The agency need not
assign percentage values to risk and engage in the adding of
probabilities but is instead required to show that, in the end,

-- 7 of 11 --

8 No. 20‐3215
it looked at all the factors together. Therefore, we must ask in
this case if the IJ’s decision suffices to show that it had consid‐
ered all of Nyandwi’s risk factors cumulatively. The IJ con‐
cluded its decision by stating that “[h]aving considered all the
relevant evidence, the Court finds Respondent has not estab‐
lished a substantial risk that he will be targeted for torture if
he returns to Burundi.” Admin. R. at 51. Although we create
no hardline rule for what an IJ must say to satisfy the aggre‐
gate risk standard, what the IJ stated here is sufficient. The IJ
treated risk as one single risk, illustrating that he was consid‐
ering all the different factors as unified, aggregated together
under one cumulative substantial risk umbrella, not as sepa‐
rate, independent risks. We agree with the BIA:
The record reflects that in evaluating the respondent’s
claim, the Immigration Judge considered the relevant
factors such as evidence of past torture, ability to relo‐
cate within the country, evidence of grave human
rights violations or other relevant country conditions.
The Immigration Judge’s detailed consideration of var‐
ious risk factors reflects a careful analysis of the com‐
ponent parts of a holistic claim. We do not view this
analysis as a failure to consider aggregate risk.
Admin. R. at 4 (internal citations and quotation marks omit‐
ted). There was no legal error here warranting remand.
B
We next consider Nyandwi’s argument that the IJ erred by
requesting corroborating evidence that was not reasonably
available. We need not reach the merits of this argument be‐
cause Nyandwi has waived it by failing to raise it before the
BIA. See Hernandez‐Alvarez v. Barr, 982 F.3d 1088, 1094 (7th

-- 8 of 11 --

No. 20‐3215 9
Cir. 2020). To exhaust an argument, and thus avoid waiver, it
must be “actually argued” in the administrative proceedings.
Duarte‐Salagosa v. Holder, 775 F.3d 841, 846 (7th Cir. 2014). It is
not enough that the new argument bears some relation to the
evidentiary record. Id. Rather, an argument is actually argued
only when it puts the BIA “on notice” that the petitioner is
trying to challenge the IJ’s decision based on that argument.
Hamdan v. Mukasey, 528 F.3d 986, 991 (7th Cir. 2008).
Nyandwi alleges that the IJ erred in demanding corrobo‐
rating evidence of specific cases in which returning refugees
have been tortured and in demanding corroborating evidence
that Nyandwi would be unable to earn money in Burundi and
so be viewed as an oppositionist, but Nyandwi never argued
either point in his BIA brief. Nyandwi contends that he raised
the issue of corroborating evidence of specific cases when he
criticized the IJ for discounting a United Nations report be‐
cause the report did not include specific examples of torture.
But Nyandwi’s argument, which consisted of the single sen‐
tence, “such an expectation is unrealistic[,]” appeared in a sec‐
tion of Nyandwi’s BIA brief that argued that the IJ mischarac‐
terized and discounted evidence, not that the IJ unreasonably
requested additional evidence. See Admin. R. at 18. This is not
enough to put the BIA on notice that Nyandwi was challeng‐
ing the IJ’s decision for the reason he now asserts. Second,
Nyandwi argues that one sentence asserting that the IJ ig‐
nored a principle that country conditions alone can sustain an
applicant’s burden under the Immigration and Nationality
Act “necessarily challenges the IJ’s demand for corroborating
evidence to demonstrate that Mr. Nyandwi would be viewed
as an oppositionist.” Pet’r’s Reply Br. at 20. But this gives
away the whole game. If Nyandwi had actually argued before
the BIA his contention that the IJ demanded unavailable

-- 9 of 11 --

10 No. 20‐3215
corroborating evidence that because he would be unable to
earn money he would be treated as an oppositionist, he would
not now need to claim that he had necessarily—meaning im‐
pliedly—argued it. Nyandwi failed to make the argument to
the Board that he seeks to pursue now, and, thus, it is waived.3
C
Last is Nyandwi’s claim that the IJ ignored critical evi‐
dence. It did not. First, the IJ considered and correctly rejected
Nyandwi’s argument concerning extorted bribe payments
disguised as election contributions. In order for such evidence
to support Nyandwi’s risk of torture argument, two inferen‐
tial steps in a chain of hypotheticals are required: if Nyandwi
does not make the payments, then he may be deemed an op‐
positionist; and if he is deemed an oppositionist, he may be
personally singled out for torture. The IJ rightly rejected such
a string of hypotheticals. When evidence is only tenuously
connected to risk of torture via a string of hypotheticals, the
evidence cannot be key. Second, the IJ considered Nyandwi’s
characteristics that could make him seem like an opposition‐
ist. The IJ in its analysis noted both Nyandwi’s Twa ethnicity
and inability to speak the language, and the IJ recited
Nyandwi’s family history in the fact section of its decision.
Finally, the evidence that Burundi’s government is authoritar‐
ian and that people in Nyandwi’s situation face persecution
by the government constitutes evidence of generalized vio‐
lence, which we have consistently held does not show risk
that the petitioner will be targeted specifically. Barry v. Barr,
916 F.3d 666, 671 (7th Cir. 2019). It is not error for the IJ not to
3 Nor would Nyandwi have succeeded on the merits of his argument be‐
cause the IJ never requested such corroborating evidence from Nyandwi
during the proceedings.

-- 10 of 11 --

No. 20‐3215 11
consider evidence that is insufficient to support a claim that
Nyandwi is personally subject to a particularized risk of tor‐
ture. See id.
Nyandwi may face a number of challenges in returning to
Burundi, but at no point did the IJ err in determining that
Nyandwi failed to show that he faces a substantial risk of tor‐
ture.
The petition for review is D ENIED.

-- 11 of 11 --