FOR PUBLICATION
UNITED STATES COURT OF APPEALS
FOR THE NINTH CIRCUIT
UNITED STATES OF AMERICA, No. 16-50364
Plaintiff-Appellee,
D.C. No.
v. 8:08-cr-00280-
JVS-2
ARMANDO REYES VERA, AKA
Mando, AKA Armando Vera,
Defendant-Appellant.
UNITED STATES OF AMERICA, No. 16-50366
Plaintiff-Appellee,
D.C. No.
v. 8:08-cr-00280-
JVS-1
SALVADOR REYES VERA,
Defendant-Appellant.
OPINION
Appeal from the United States District Court
for the Central District of California
James V. Selna, District Judge, Presiding
Argued and Submitted May 17, 2018
Pasadena, California
Filed June 25, 2018
2 UNITED STATES V. VERA
Before: Kim McLane Wardlaw, Jacqueline H. Nguyen,
and John B. Owens, Circuit Judges.
Opinion by Judge Owens
SUMMARY *
Criminal Law
The panel vacated two defendants’ sentences imposed on
remand for resentencing in a drug-trafficking conspiracy
case, and again remanded for resentencing.
In the prior appeal, this court affirmed the convictions
but remanded the sentences because unreliable evidence had
been presented to the jury.
The panel held that the district court, on remand,
committed reversible error by relying heavily upon co-
conspirator plea agreements to determine the drug quantities
attributable to the defendants on the ground that the plea
agreements were reliable statements against interest under
Fed. R. Evid. 804(b)(3). The panel held that district courts
may not rely solely on Rule 804(b)(3) to use non-self-
inculpatory statements in a co-conspirator’s plea agreement
to determine a defendant’s drug-quantity liability.
The panel recognized that co-defendant plea agreements
could have some probative value at sentencing if otherwise
*
This summary constitutes no part of the opinion of the court. It
has been prepared by court staff for the convenience of the reader.
UNITED STATES V. VERA 3
supported by sufficient indicia of reliability. On this record,
the panel concluded there were not sufficient indicia of
reliability to support the plea agreements’ probable accuracy
as to drug quantity, and that the factual bases in the plea
agreements were not corroborated by other information that
made their reliability apparent.
COUNSEL
Thomas Paul Sleisenger (argued), Los Angeles, California,
for Defendant-Appellant Armando Reyes Vera.
Gretchen Fusilier (argued), Carlsbad, California, for
Defendant-Appellant Salvador Reyes Vera.
Bram M. Alden (argued), Assistant United States Attorney;
Lawrence S. Middleton, Chief, Criminal Division; United
States Attorney’s Office, Los Angeles, California; for
Plaintiff-Appellee.
OPINION
OWENS, Circuit Judge:
Defendants (and brothers) Salvador and Armando Vera
appeal their sentences for drug-trafficking conspiracy. We
previously affirmed their convictions but remanded their
sentences because unreliable evidence had been presented to
the jury. See United States v. Vera, 770 F.3d 1232, 1253 (9th
Cir. 2014) (Vera I). On remand, the district court used
evidence of questionable value in determining the drug
quantities attributable to each defendant, so we again vacate
and remand for resentencing.
4 UNITED STATES V. VERA
I. FACTUAL BACKGROUND AND
PROCEDURAL HISTORY
A. Vera I
As our previous opinion detailed the crimes of the Vera
brothers, we will get to the point. In October 2008, the
Veras, along with thirteen co-conspirators, were charged
with conspiracy to distribute narcotics in violation of
21 U.S.C. §§ 841(a)(1), 846. Their thirteen co-conspirators
eventually pled guilty, but the Veras proceeded to trial under
a superseding indictment. The jury was tasked with deciding
if the Veras were guilty of (1) conspiring to distribute
controlled substances and (2) using a minor in drug
operations. Vera I, 770 F.3d at 1235. And if the answer was
yes for either charge, then the jury would determine the
quantity of drugs attributable to the defendants. Id. at 1237.
After a five-day trial, the jury found the Veras guilty and
returned a special verdict holding them responsible for at
least: 100 grams of heroin, 500 grams of cocaine, and
280 grams of cocaine base. Id. With those findings in hand,
the district court sentenced Salvador to 360 months’ and
Armando to 210 months’ imprisonment. 1 Id.
On appeal, we affirmed their convictions but reversed
their sentences due to the testimony of the government’s
“key witness,” FBI Special Agent Lavis. Id. at 1236. Lavis
opined about the government’s “primary evidence” against
the Veras: over seventy wiretapped phone calls that were
played or read before the jury. Id. Aside from one proven
sale of heroin by Armando, “Lavis’ opinions interpreting the
wiretapped calls were the only evidence of specific
1
For consistency with Vera I, we refer to the defendants by their
first names, Salvador and Armando.
UNITED STATES V. VERA 5
quantities at trial.” Id. at 1243. This opinion testimony, we
concluded, “did not rest on reliable methods.” Id. at 1247.
In some instances, Lavis opined on the meaning of
ambiguous terms based on false assumptions. Id. at 1247–
48. At one point, Lavis construed a call with neither direct
nor encoded words as a drug transaction. Id. at 1248. We
recognized that this “opinion plainly rested on nothing more
than speculation.” Id.
And because “[t]he defendants’ lengthy sentences” and
statutory mandatory minimums “depended on [the jury’s]
drug quantity findings,” id. at 1235, 1249, we concluded that
resentencing was required 2—either with a new sentencing
jury, or by the district court under the default sentencing
provisions in 21 U.S.C. § 841(b)(1)(C), id. at 1253. The
government elected to proceed before the district court. 3
B. Resentencing
Rather than remedy Lavis’ improper methodology or call
the Veras’ co-conspirators to testify about their dealings with
the defendants, the government instead relied heavily upon
the co-conspirators’ plea agreements to establish the
quantities of drugs attributable to the Vera brothers. In
effect, the government swapped out the wiretapped calls for
2
We also vacated the sentences because the district court failed to
require that the government lay an adequately specific foundation for
Lavis’ testimony or to instruct the jury that Lavis was testifying as both
an expert and lay witness. Vera I, 770 F.3d at 1243–44.
3
Proceeding without a sentencing jury meant that the Veras faced
(1) no mandatory minimum sentence, and (2) a smaller statutory
maximum. See 21 U.S.C. § 841(b)(1)(C). But for practical purposes,
there was no difference to the Vera brothers—Salvador still faced a
sixty-year statutory maximum due to a prior conviction, and Armando a
forty-year statutory maximum. See 21 U.S.C. § 861(b).
6 UNITED STATES V. VERA
the plea agreements. The district court recognized as much
in its sentencing order. It explained that it found the
government’s sentencing memoranda “more credible” than
the presentence investigation reports and Armando’s
sentencing memorandum because it was the “least
dependent on interpretation of [the] recordings.” It also
recognized that the plea agreements were the government’s
“single most significant data source.”
These plea agreements, which the government drafted,
frequently pointed fingers at the Veras. Of the twelve plea
agreements relied upon by the government, ten named
Armando as a co-conspirator. Five of those also named
Salvador. The factual bases of some of the plea agreements
referenced the Veras more specifically. One provided that a
co-conspirator “conspired to distribute between five and
20 grams of cocaine base with Armando,” and another stated
that a co-conspirator “obtained cocaine base” from him.
These and several other plea agreements also catalogued
wiretapped calls between both Vera brothers and their co-
conspirators.
Over the defendants’ objections, the district court found
these plea agreements reliable. Citing Federal Rule of
Evidence 804(b)(3), (the statement-against-interest
exception to the hearsay rule), the district court concluded
that the co-conspirators’ “admissions in [the] plea
agreement[s] clearly amount[ed] to declarations against
interest.” “Moreover,” the district court explained, “the
Government provide[d] specific corroboration for a number
of the plea agreements.”
After applying some additional sentencing
enhancements, the court re-sentenced Salvador to
324 months’ and Armando to 168 months’ imprisonment.
UNITED STATES V. VERA 7
This timely appeal followed.
II. DISCUSSION
A. Jurisdiction and Standard of Review
We have jurisdiction under 18 U.S.C. § 3742(a) and
review for abuse of discretion the district court’s evaluation
of the reliability of evidence at sentencing. United States v.
Hernandez-Guerrero, 633 F.3d 933, 935 (9th Cir. 2011).
B. The District Court Abused its Discretion in
Relying on the Co-Defendants’ Plea
Agreements to Determine the Veras’ Drug
Quantities
At sentencing, the Confrontation Clause does not apply,
United States v. Littlesun, 444 F.3d 1196, 1197 (9th Cir.
2006), and district courts have wide latitude when deciding
upon which information to rely, United States v. Showalter,
569 F.3d 1150, 1159 (9th Cir. 2009). But that information
must still have “sufficient indicia of reliability to support its
probable accuracy.” U.S.S.G. § 6A1.3(a). Here, the district
court relied heavily upon the co-conspirator plea agreements
to determine the drug quantities attributable to the Veras,
concluding that the plea agreements were reliable statements
against interest under Federal Rule of Evidence 804(b)(3).
This was reversible error.
The text of Rule 804(b)(3) explains why a purely
inculpatory statement is deemed reliable: “a reasonable
person in the declarant’s position would have made [it] only
if the person believed it to be true because, when made, it . . .
had so great a tendency . . . to expose the declarant to civil
or criminal liability.” Fed. R. Evid. 804(b)(3). Yet a factual
basis in a negotiated plea agreement, pointing the finger at
8 UNITED STATES V. VERA
others, is no such thing. This principle was recognized in
Williamson v. United States, 512 U.S. 594 (1994), where the
Supreme Court held that a statement must be “genuinely
self-inculpatory” to qualify under Rule 804(b)(3), and
vacated a defendant’s conviction because it was supported
by his accomplice’s confession which, in certain parts, “did
little to subject” the accomplice “to criminal liability.” Id. at
604–05. And as our court has recognized, the Supreme
Court’s “time-honored teaching” that “a codefendant’s
confession inculpating the accused is inherently unreliable”
is “equally applicable in the sentencing as in the conviction
context.” United States v. Pimentel-Lopez, 859 F.3d 1134,
1144 (9th Cir. 2016) (quoting Lee v. Illinois, 476 U.S. 530,
546 (1986)).
A defendant signing a plea agreement may adopt facts
that the government wants to hear in exchange for some
benefit, usually a lesser sentence. In pointing their fingers at
the Vera brothers, the co-conspirators were acknowledging
neither their own guilt nor conduct that would necessarily
enhance their own sentences. Rather, these statements
merely helped the government’s prosecution of the Veras.
See Williamson, 512 U.S. at 600 (“The fact that a statement
is self-inculpatory does make it more reliable; but the fact
that a statement is collateral to a self-inculpatory statement
says nothing at all about the collateral statement’s
reliability.”). And while the factual basis in a plea agreement
binds the party who signed it, that factual basis carries far
less weight against a co-defendant. See Lee, 476 U.S. at 541
(“Due to his strong motivation to implicate the defendant
and to exonerate himself, a codefendant’s statements about
what the defendant said or did are less credible than ordinary
hearsay evidence.” (internal quotation marks omitted)). The
Advisory Committee Notes to Rule 804(b)(3) stress that “a
statement admitting guilt and implicating another person,
UNITED STATES V. VERA 9
made while in custody, may well be motivated by a desire to
curry favor with the authorities and hence fail to qualify as
against interest.” Fed. R. Evid. 804(b)(3) advisory
committee’s note to 1972 proposed rules. Neither the district
court nor the government cited any authority suggesting that
a factual basis in a plea agreement pointing the finger at
someone else qualifies as Rule 804(b)(3) material, and there
is ample case law, in addition to Williamson, suggesting
otherwise.
For example, in United States v. Magana-Olvera,
917 F.2d 401 (9th Cir. 1990), we reviewed a similar issue
and held that statements of an in-custody co-conspirator
implicating the defendant did not qualify as a statement
against interest because “they were made in an attempt to
curry favor from the federal authorities.” Id. at 409; see also
United States v. Monaco, 735 F.2d 1173, 1177 (9th Cir.
1984) (recognizing that “courts have closely scrutinized
statements made while the declarant is in custody and
offered against the accused, and have consistently held that
the circumstances render such statements unreliable”).
And we are not alone in coming to this common sense
conclusion. The Fifth Circuit explained the dynamic of the
co-conspirator in custody this way:
There were, in addition [to the potentially
coercive circumstances of custody], obvious
motives for falsification [—;] the very natural
desire to curry favor from the arresting
officers, the desire to alleviate culpability by
implicating others, the enmity often
generated in a conspiracy gone awry . . . all
might lead an arrestee-declarant to
misrepresent or to exaggerate the role of
others in the criminal enterprise.
10 UNITED STATES V. VERA
United States v. Sarmiento-Perez, 633 F.2d 1092, 1102 (5th
Cir. 1981). This reluctance to allow reliance on co-
conspirator admissions is widespread throughout the
circuits. See United States v. Johnson, 802 F.2d 1459, 1465
(D.C. Cir. 1986) (holding a post-arrest statement not
sufficiently against interest because it is “highly logical” for
an arrestee to trivialize his own involvement by implicating
the defendant as the kingpin in a drug operation); United
States v. Riley, 657 F.2d 1377, 1384 (8th Cir. 1981) (holding
a hearsay statement not sufficiently against interest because
the declarant was in police custody and warned of the
adverse consequences of conviction, and thus “may well
have believed that it was in her best interest to make a
statement implicating [the defendant] in order to ingratiate
herself with the authorities and divert attention to another”);
United States v. Bailey, 581 F.2d 341, 345 & n.4 (3rd Cir.
1978) (holding a hearsay statement not sufficiently against
interest because the declarant was in custody and aware of
the possibility of leniency if he confessed and implicated the
defendant); cf. United States v. Nagib, 56 F.3d 798, 805 (7th
Cir. 1995) (holding that a hearsay statement qualified under
Rule 804(b)(3) in part because there was “no record of any
plea agreement or downward departure for cooperation” that
could have called the declarant’s penal interest into
question).
Of course, “hearsay is admissible at sentencing, so long
as it is accompanied by some minimal indicia of reliability.”
Littlesun, 444 F.3d at 1200 (internal quotation marks
omitted). But here, the district court’s primary rationale for
relying upon the plea agreements—Rule 804(b)(3)—was
incorrect. At sentencing, district courts may not rely solely
on Rule 804(b)(3) to use non-self-inculpatory statements in
a co-conspirator’s plea agreement to determine a defendant’s
drug-quantity liability.
UNITED STATES V. VERA 11
At the same time, we recognize that co-defendant plea
agreements could have some probative value at sentencing
if otherwise supported by “sufficient indicia of reliability.”
U.S.S.G. § 6A1.3(a). In United States v. Berry, 258 F.3d 971
(9th Cir. 2001), we held that the district court properly relied
on co-defendant hearsay statements because multiple
statements corroborated each other and thus provided
“external consistency” that evidenced their reliability. Id. at
976–77; see also United States v. Valensia, 222 F.3d 1173,
1184 (9th Cir. 2000) (“[T]he hearsay statements at issue . . .
consist of three identical statements, given independently
under circumstances which limited the possibility for
collusion, that corroborate one another.”), vacated on other
grounds sub nom. Valensia v. United States, 532 U.S. 901
(2001). And “[w]hile we encourage and appreciate express
findings by a district court regarding the reliability of
hearsay statements introduced at sentencing,” we have not
reversed for failure “to articulate such findings” where the
statements’ reliability “is apparent from the record.” Berry,
258 F.3d at 976.
On this record, however, we do not find “sufficient
indicia of reliability to support [the plea agreements’]
probable accuracy” as to drug quantity. U.S.S.G.
§ 6A1.3(a). It was the question of drug quantity, not
conspiracy, that was before the district court. And thus it
was quantity, not conspiracy, that required corroboration.
As corroboration for the plea agreements, the
government has offered other co-defendants’ plea
agreements and the intercepted calls admitted at trial, as well
as the criminal complaint, investigative reports, and DEA
laboratory reports that the government submitted with its
sentencing memoranda. While these sources may have
corroborated the plea agreements as to the Vera brothers’
12 UNITED STATES V. VERA
participation in the drug-trafficking conspiracy, they were
not corroborative as to the drug quantities attributable to
them.
Turning first to the plea agreements themselves, we
reject the government’s contention that they corroborate
each other. The twelve plea agreements in the record—all
drafted by the government—collectively listed over forty
narcotics transactions. Each transaction included the
specific type and quantity of drugs purchased or sold. Of all
these transactions, we found only four that were specifically
referenced in the factual bases of more than one plea
agreement. None of these four specifically referenced
involvement by either Vera brother. The plea agreements
therefore do not corroborate each other as to the drug
quantities attributable to the Veras.
Nor do the majority of the wiretapped calls. In our
review of the record, we found only five transactions listed
in the plea agreements that the calls corroborated as to
quantity. And while the government argues that encoded
words like “rock” or “pants” are corroborative, unless those
terms are preceded by quantity designations, we decline to
find they corroborate anything more than the Veras’
participation in the drug-trafficking conspiracy.
The government also offers the criminal complaint,
investigative reports, and DEA laboratory reports as
corroboration. Assuming without deciding that a criminal
complaint can be used against a defendant at sentencing, 4 we
4
In a 28(j) letter, the government provided two out-of-circuit cases
suggesting that district courts may rely on complaint affidavits at
sentencing. See United States v. Clark, 538 F.3d 803, 814 (7th Cir. 2008)
(holding that the sentencing court could rely on the complaint to
UNITED STATES V. VERA 13
found that only two transactions in the plea agreements were
indeed corroborated by the complaint as to quantity.
Likewise, though we recognize that investigative reports
may be relied upon at sentencing, see United States v. Mara,
523 F.3d 1036, 1039 (9th Cir. 2008), we found that only two
transactions in the plea agreements were corroborated by the
reports. And finally, while the DEA laboratory reports may
indicate the reliability of the methamphetamine quantities
listed in one co-conspirator’s plea agreement, neither the
plea agreement nor the DEA laboratory reports directly ties
the Veras to these methamphetamine sales, and we decline
to make this inference without factual support in the record.
Reversal may be necessary when a district court bases a
sentence on sources whose reliability is not apparent from
the record. See Berry, 258 F.3d at 976. And “remand is
required” when a “sentencing judge considers unreliable
information” that was “demonstrably made the basis for the
sentence.” United States v. Huckins, 53 F.3d 276, 280 (9th
Cir. 1995) (internal quotation marks omitted). Here, the
factual bases in the plea agreements were neither inherently
reliable as statements against interest nor corroborated by
other information that made their reliability apparent. They
were also demonstrably made the bases of the Veras’
sentences. Accordingly, we vacate the sentences and
remand for resentencing.
At oral argument, the government argued for the first
time that there was ample evidence to support the sentences
without considering the plea agreements, so any reliance on
corroborate testimony as to the drug quantity attributable to the
defendant); cf. United States v. Jones, 453 F.3d 777, 780 (6th Cir. 2006)
(“An affidavit of complaint is a type of record that a district court can
properly rely on in determining the nature of predicate offenses . . . .”).
14 UNITED STATES V. VERA
the plea agreements was harmless. And while that may
ultimately be true (and this appears to be a much stronger
argument than the one presented in the appellate briefs), we
think it best for the district court to consider this argument
on remand: it previously considered the plea agreements to
be the “single most significant data source,” for its drug-
quantity calculations, and it is better positioned to reassess
the reliable evidence and make factual findings in the first
instance. We also decline to review the other sentencing
enhancements at this time, as we remand this case to the
district court to resentence on an open record. 5
VACATED AND REMANDED.
5
Because the errors here would require remand under either a
preponderance or a clear-and-convincing evidentiary standard, we do not
take up the government’s invitation to reconsider the validity of United
States v. Staten, 466 F.3d 708 (9th Cir. 2006).