United States v. Akoto

United States Court of Appeals
For the First Circuit

No. 21-1804

UNITED STATES,

Appellee,

v.

EMMANUEL AKOTO, a/k/a Kofi,

Defendant, Appellant.

APPEAL FROM THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF NEW HAMPSHIRE

[Hon. Steven J. McAuliffe, U.S. District Judge]

Before

Barron, Chief Judge,
Selya and Lynch, Circuit Judges.

Sara E. Silva, with whom Hogan Lovells US LLP was on brief,
for appellant.
Hannah Cook, Attorney, Tax Division, Department of Justice,
with whom David A. Hubbert, Deputy Assistant Attorney General,
S. Robert Lyons, Chief, Criminal Appeals & Tax Enforcement Policy
Section, Katie Bagley, Attorney, Tax Division, Joseph B. Syverson,
Attorney, Tax Division, and Jane E. Young, United States Attorney,
were on brief, for appellee.

February 23, 2023
 LYNCH, Circuit Judge. A New Hampshire federal jury

convicted Emmanuel Akoto of one count of conspiracy to commit wire

fraud, three counts of substantive wire fraud, and two counts of

aggravated identity theft. These charges were based on evidence

of Akoto's participation in an international scheme, involving

individuals in the United States, Nigeria, and Ghana, that used

stolen identities to file fraudulent federal income tax returns

with the Internal Revenue Service ("IRS"). At sentencing, the

district court determined that Akoto and his coconspirators had

filed at least 310 fraudulent tax returns seeking $1,326,633 in

refunds, $551,601 of which the IRS paid out. Based in part on

this loss amount, the district court sentenced Akoto to 70 months'

imprisonment, which represented a downward variance from his

Guidelines range.

On appeal, Akoto makes three arguments. First, he

contends that his conviction on one of the aggravated identity

theft counts should be vacated because his trial counsel's failure

to raise a statute of limitations defense to this count amounted

to ineffective assistance of counsel. Second, he argues that his

convictions on the three substantive wire fraud counts should be

vacated because the district court's jury instructions

constructively amended the indictment. Third, Akoto asserts that

his sentence should be vacated because the district court erred in

- 2 -
calculating the loss amount attributable to his conduct. We

affirm.

I.

A.

We recount the background facts in the light most

favorable to the jury's verdict, consistent with record support.

See United States v. Tkhilaishvili, 926 F.3d 1, 8 (1st Cir. 2019).

Between 2011 and 2013, Akoto and his coconspirators used

stolen identities to file fraudulent federal income tax returns.

The scheme worked as follows.

Akoto and his coconspirators first purchased stolen

identity information from Hieu Minh Ngo, a Vietnamese hacker.

Between 2007 and 2013, Ngo ran an illicit business selling personal

identifying information ("PII") over the internet. This

information came as "fullz" (or "fulls") packages -- short for

"full information" -- that typically included information like an

individual's name, Social Security number, date of birth, address,

driver's license number, and bank account numbers. Much of this

information constituted "means of identification" ("MOI"), as

defined at 18 U.S.C. § 1028(d)(7). Ngo maintained an inventory of

more than 176,000 fullz and sold fullz to at least 1,300

individuals around the world. Ngo often resold the same fullz to

different individuals.

- 3 -
 Akoto purchased between 900 and 1,000 fullz from Ngo.

Ngo sent Akoto these fullz in email attachments. Akoto routinely

requested newly hacked information that would work for the tax

fraud scheme. He asked Ngo for "fresh ones" and "the newest info

that you have," specified the timeframes he was looking for, and

sought information that "would pass" -- i.e., could be successfully

used in the scheme.

Ngo was eventually apprehended by American law

enforcement and agreed to cooperate. He allowed Secret Service

Special Agent Matthew O'Neill to take over his email accounts.

Agent O'Neill used Ngo's email to communicate with Ngo's customers,

including Akoto, for investigatory purposes.

After receiving the stolen identities, Akoto and his

coconspirators "washed" each identity by submitting a tax return

to the IRS using that information but deliberately using the wrong

date of birth. The IRS typically responded with a rejection letter

stating either that (1) the date of birth was incorrect or (2) the

date of birth was incorrect and a return for that individual had

already been filed. If the former, the coconspirators knew they

could file a potentially successful fraudulent tax return using

that person's name because the person had not yet filed a tax

return for that year. The purpose of "washing" was to avoid

purchasing prepaid debit cards (the next step of the scheme) in

- 4 -
the names of individuals for whom a fraudulent return could not be

successfully filed.

After a name had been successfully "washed," Akoto and

his coconspirators purchased a prepaid debit card in that person's

name, filed a fraudulent tax return (this time with the correct

date of birth), and routed the refund to the prepaid debit card.

The fraudulent returns were often filed by conspirators in Nigeria

and Ghana. If the IRS did not detect the fraud and issued a refund

to the prepaid debit card, Akoto or a coconspirator withdrew the

refund in cash from an ATM.1 Some of this money was sent to the

overseas conspirators by depositing it in different accounts, with

the conspirator who withdrew the cash also keeping some.

B.

On November 29, 2017, a federal grand jury returned an

indictment charging Akoto and codefendant Jeffrey Quaye with six

counts: one count of conspiracy to commit wire fraud (Count One),

see 18 U.S.C. §§ 1343, 1349; three counts of substantive wire fraud

and aiding and abetting wire fraud (Counts Two, Three, and Four),

see id. §§ 2, 1343; and two counts of aggravated identity theft

(Counts Five and Six), see id. § 1028A(a)(1).

Quaye entered into a plea agreement with the government

before trial and testified against Akoto at trial. The government

1 Some refunds were routed directly to bank accounts
rather than to prepaid debit cards.

- 5 -
also presented testimony from two other cooperating witnesses:

Ngo, the hacker and fullz seller, and Abiola Adeyemo, who was

present for a conversation between Akoto and Quaye regarding the

scheme and pleaded guilty to participation in a related tax fraud

scheme. The government further called as witnesses four government

agents and six victims whose PII had been used in the scheme.

Akoto's central defense at trial was that the government

could not tie him to the scheme because he did not control the

kwa2kg@yahoo.com email account that was used to communicate with

Ngo, receive fullz, share fullz with coconspirators, and share

other information such as "washed" names. The government

introduced evidence that emails sent from the account were sent

from an IP address tied to 2047 Paducah Lane, Grand Prairie, Texas,

the address where Akoto was living. Quaye also testified that the

email account belonged to Akoto. And the government introduced

other documentary evidence that the account belonged to Akoto,

including messages in the account referring to his business,

regarding airplane tickets for him and his wife, and addressing

him by his nicknames. The government further established that the

kwa2kg@yahoo.com account was used to correspond with two other

email accounts (kwa22kg@yahoo.com and nana2kg@gmail.com) to

further the conspiracy -- i.e., by sharing fullz and washed names.

Following a three-day trial, the jury convicted Akoto on

all counts.

- 6 -
 At sentencing, the district court adopted the findings

and recommendations set forth in the presentence investigation

report ("PSR"). It accepted that Akoto and his coconspirators had

filed at least 310 fraudulent tax returns seeking $1,326,633 in

refunds, of which the IRS actually paid out $551,601. Based on

this intended loss amount of $1,326,633 and several other

enhancements, the court calculated a total offense level of 25,

resulting in a Guidelines range of 57 to 71 months plus a mandatory

consecutive 24-month sentence for the aggravated identity theft

charges. The court varied downward and sentenced Akoto to 70

months' imprisonment: 46 months on the conspiracy count and on

each substantive wire fraud count, to be served concurrently, and

24 months on each aggravated identity theft count, to be served

concurrently to each other and consecutively to the 46 months.2

II.

A.

As to Akoto's first argument on appeal, based on alleged

ineffective assistance of counsel, such arguments are rarely

entertained on direct appeal if not previously raised below, and

2 The district court also imposed two years of supervised
release, $551,601 in restitution, and a $600 assessment.

- 7 -
we decline to do so here. See United States v. Miller, 911 F.3d

638, 641-42, 646 (1st Cir. 2018).

Counts Five and Six of the indictment charged Akoto with

committing aggravated identity theft in violation of 18 U.S.C.

§ 1028A by "knowingly possess[ing] and us[ing], without lawful

authority, means of identification of other people, that is, the

names and Social Security Numbers of two individuals for the

purpose of filing false and fraudulent United States Income Tax

Returns claiming tax refunds . . . during and in relation to a

. . . conspiracy to commit wire fraud." The indictment alleged

that these offenses "continu[ed] to a date uncertain, but at least

as late as March 13, 2013." The indictment also specified that

the false tax returns associated with these counts were filed on

November 20, 2012 (Count Five) and January 31, 2013 (Count Six).

Akoto argues that Count Five was untimely "from the face

of the indictment" because the indictment was not returned until

November 29, 2017, more than five years after the false tax return

associated with Count Five was filed on November 20, 2012. See 18

U.S.C. § 3282(a) (setting a five-year statute of limitations period

for most noncapital offenses). He claims that his trial counsel's

failure to raise this statute of limitations defense amounted to

ineffective assistance of counsel in violation of the Sixth

Amendment.

- 8 -
 The government responds that (1) it is premature to

address Akoto's ineffective assistance claim in the present

appeal, not least because further factual development is needed,

and (2) should this court entertain Akoto's ineffective assistance

claim, it fails on the merits for multiple reasons. The government

argues that Count Five was timely for three reasons: First, the

charging language in the indictment expressly alleged that the

offense continued "at least as late as March 13, 2013," and the

inclusion of the earlier date on which the tax return was filed

should not override this charging language. Second, Count Five

was timely as a continuing offense: aggravated identity theft

requires the commission of a predicate felony, the statute of

limitations does not begin to run until the predicate felony is

completed, and the predicate felony here -- the conspiracy to

commit wire fraud -- lasted well into the statute of limitations

period. Third, Count Five was timely as proven at trial because

the refund associated with the false tax return was not issued

until December 5, 2012, within the statute of limitations period,

and because Akoto continued to "possess" the stolen identity

information in his email account within this period.

At the very least, the government argues, the uncertain

viability of a statute of limitations defense is such that trial

counsel's failure to raise it does not constitute ineffective

assistance and Akoto has also not met his burden to show prejudice.

- 9 -
The government makes the point that Akoto faced a mandatory two-

year sentence based on the other aggravated identity theft count

(Count Six), which Akoto does not challenge as untimely, so trial

counsel's choice to pursue an alternative defense theory which

could have resulted in acquittal on all counts was reasonable.

We reach only the threshold question and decline to

review the merits of Akoto's ineffective assistance claim on direct

appeal. This court has "held with a regularity bordering on the

monotonous that fact-specific claims of ineffective assistance of

counsel cannot make their debut on direct review of criminal

convictions, but, rather, must originally be presented to, and

acted upon by, the trial court." Tkhilaishvili, 926 F.3d at 20

(quoting United States v. Mala, 7 F.3d 1058, 1063 (1st Cir. 1993)).

The trial court, "by reason of [its] familiarity with the case, is

usually in the best position to assess both the quality of the

legal representation afforded to the defendant . . . and the impact

of any shortfall in that representation" in the first instance.

Id. (quoting Mala, 7 F.3d at 1063). "Thus, a criminal defendant

who wishes to pursue a claim of ineffective assistance not advanced

in the trial court is ordinarily required to defer that claim to

collateral proceedings." Miller, 911 F.3d at 642.

We have recognized a potential exception to this rule

"where the critical facts are not genuinely in dispute and the

record is sufficiently developed to allow reasoned consideration"

- 10 -
of the claim on direct appeal. Id. (quoting United States v.

Natanel, 938 F.2d 302, 309 (1st Cir. 1991)). That exception is

not applicable here. It is by no means clear, for reasons

articulated by the government, that a statute of limitations

defense was obviously available to Akoto in this case. Further

factual development, which is not present in the record before us,

is necessary to appraise counsel's performance.3 See, e.g.,

Tkhilaishvili, 926 F.3d at 20 (declining to review an ineffective

assistance claim on direct appeal for similar reasons); Miller,

911 F.3d at 646 (same); United States v. Leahy, 473 F.3d 401, 410

(1st Cir. 2007) (same).

Akoto contends that our recent decision in Miller not to

take up an ineffective assistance of counsel claim on direct appeal

"turned on" the fact that counsel in that case was at least aware

of a potential statute of limitations issue -- and that the absence

of evidence of a similar awareness in this case is grounds to

decide Akoto's claim on direct appeal. Nothing in Miller supports

that proposition. Miller noted by way of background that counsel

was aware of a potential statute of limitations issue, see 911

F.3d at 641, but that fact did not guide our analysis. Miller's

holding rested instead on the general rule that we will not resolve

3 Akoto points to certain portions of the trial transcript
he claims "affirmatively show[]" that counsel was unaware of the
statute of limitations defense. Not so.

- 11 -
an ineffective assistance claim on direct review if the record is

not sufficiently developed to allow for thoughtful consideration

of the claim. See id. at 646 ("We are left to guess at trial

counsel's thought processes . . . . When all is said and done, we

know little more than that trial counsel chose not to file a motion

to dismiss."); see also Leahy, 473 F.3d at 410 (declining to credit

the defendant's assertion that his trial counsel was simply unaware

of a key case in the absence of developed factual support for that

assertion).

We thus affirm Akoto's conviction on Count Five, without

prejudice to his ability to raise his claim of ineffective

assistance of counsel in a collateral proceeding should he wish to

do so. See 28 U.S.C. § 2255.

B.

Akoto next argues that the district court constructively

amended the three substantive wire fraud counts charged in the

indictment (Counts Two though Four) in its jury instructions.

Akoto did not object to the jury instructions in the district

court, so our review is for plain error. See United States v.

Brandao, 539 F.3d 44, 57 (1st Cir. 2008). Under that standard,

Akoto must show "(1) that an error occurred (2) which was clear or

obvious and which not only (3) affected [Akoto's] substantial

rights, but also (4) seriously impaired the fairness, integrity,

or public reputation of judicial proceedings." United States v.

- 12 -
Valdés-Ayala, 900 F.3d 20, 36 (1st Cir. 2018) (quoting United

States v. George, 841 F.3d 55, 64 (1st Cir. 2016)). We conclude

that, reading the challenged instructions in context, no error

occurred.

Counts Two through Four alleged, in relevant part, that

Akoto and Quaye "devised and intended to devise and aided and

abetted each other in devising a scheme and artifice to defraud"

and, in furtherance of that scheme, "transmitted and caused to be

transmitted by means of wire communications, as more particularly

described below, in interstate and foreign commerce, certain

writings." The three counts were specifically charged as follows:

Count Description of Wire Location of Date of Wire
Number Wire
Two AKOTO e-mail sent to Sent to February 14,
monitored e-mail District of 2013
account requesting New Hampshire
200 MOI
Three AKOTO e-mail sent to Sent to February 16,
monitored e-mail District of 2013
account requesting New Hampshire
MOI "that wld pass"
Four AKOTO e-mail sent to Sent to February 19,
monitored account District of 2013
requesting "new info New Hampshire
that is almost 9 or 6
months old."

In the jury instructions, the district court defined the

term "interstate wire communication," in relevant part, as

follows:

An "interstate wire communication" includes a
telephone communication from one state to
another, as well as an email transmission or

- 13 -
 other internet communication. It also
includes the electronic filing of a tax return
with the Internal Revenue Service from one
state to another.

The district court also instructed the jury on aider and abettor

liability as to Counts Two through Four.

Akoto argues that the instruction that an interstate

wire communication includes "the electronic filing of a tax

return," given in conjunction with the aiding and abetting

instruction, "permitted the jury to convict [Akoto on Counts Two

through Four] upon a finding that he aided and abetted any co-

conspirator in electronically filing any tax return despite the

fact that those counts were expressly premised on specific emails."

(Emphasis omitted.) He contends that this amounted to an

impermissible constructive amendment to the indictment.4

"A constructive amendment occurs when the charging terms

of an indictment are altered, either literally or in effect, by

prosecution or court after the grand jury has last passed upon

them." Brandao, 539 F.3d at 57 (quoting United States v. Pierre,

484 F.3d 75, 81 (1st Cir. 2007)). "The prohibition on constructive

amendment exists to preserve the defendant's Fifth Amendment right

to indictment by grand jury, to prevent re-prosecution for the

same offense in violation of the Sixth Amendment, and to protect

4 In his reply brief, Akoto disclaims any separate
challenge based on the aiding and abetting instruction.

- 14 -
the defendant's Sixth Amendment right to be informed of the charges

against him." Id. In assessing whether a district court's jury

instructions constructively amended the indictment, we evaluate

the challenged instructions not in isolation but in the context of

the entire charge. See United States v. McBride, 962 F.3d 25, 33

(1st Cir. 2020); United States v. Lopez-Cotto, 884 F.3d 1, 9-11

(1st Cir. 2018).

Taken in context, the district court's definition of

"interstate wire communication" did not constructively amend the

indictment and did not amount to error, let alone plain error.

The district court began by instructing the jurors that

they "should not single out any one instruction, but instead apply

these instructions as a whole to the evidence in the case."

Turning to the substantive instructions, the district court told

the jury that it would explain "the elements of the substantive

crime of wire fraud" before explaining conspiracy, "because [the

conspiracy] instructions will be better understood if the

substantive counts have been explained first." The district court

then explained that Counts Two through Four "allege that [Akoto]

. . . devised and intended to devise, and aided and abetted another

in devising, a scheme and artifice to defraud, and for obtaining

money and property by means of materially false and fraudulent

pretenses, representations, and promises" and "[i]n furtherance

of, and for the purpose of executing such schemes and artifice to

- 15 -
defraud . . . transmitted and caused to be transmitted by means of

wire communications, as more particularly described below, in

interstate and foreign commerce, certain writings." The district

court then read the description of the three specific emails

charged for Counts Two through Four verbatim from the indictment.

The district court then provided definitions. The

district court defined terms such as "scheme," "defraud," and aider

and abettor liability. The district court also provided the

definition of "interstate wire communication" discussed above.

The district court went on to explain Count One,

conspiracy to commit wire fraud. The court referred the jury back

to its earlier discussion of wire fraud before turning to the

conspiracy-specific elements of the count: "As I instructed you

earlier, section 1343 of Title 18 makes it unlawful to commit wire

fraud and provides that . . . ." The court provided definitions

for terms like "conspiracy" and "overt act," but did not repeat

the substantive wire fraud definitions it had previously given,

including the definition of "interstate wire communication."

The district court defined the substantive elements of

wire fraud only once and had the jury use that definition for both

the substantive wire fraud counts and the conspiracy to commit

wire fraud count. The court sensibly noted that the filing of a

tax return could be a wire communication, because Count One alleged

that such filings were overt acts in furtherance of the conspiracy.

- 16 -
This definition did not shift the theory of the case as to Counts

Two through Four. Indeed, the jury instructions on those counts

described the specific email which served as the basis for each

count.

The challenged jury instruction did not constructively

amend the indictment and was not error, let alone plain error.

See McBride, 962 F.3d at 33 (rejecting constructive amendment

challenge after reviewing jury instructions as a whole);

Lopez-Cotto, 884 F.3d at 10-11 (same).

C.

Akoto's final contention on appeal is that the district

court erred in calculating the loss amount attributable to his

conduct during sentencing. Akoto makes two arguments. First, he

asserts that there is not sufficient evidence connecting him to

the 310 fraudulent tax returns on which the district court based

its loss calculations. He argues that the fact that PII used in

these fraudulent returns was found in the conspiracy's email

accounts is insufficient, because other fraudsters may have had

access to this same PII.5 Second, Akoto contends that the district

court should have used the same loss amount for him as for Quaye,

which would have resulted in a 12-level (rather than 14-level)

5 Akoto points to the facts that Ngo sold PII to at least
1,300 individuals around the world, that Ngo often resold the same
fullz to multiple individuals, and that at least 13,673 fraudulent
tax returns were filed using information from the fullz.

- 17 -
increase in his total offense level and potentially a shorter

sentence.

"'In a fraud case resulting in financial loss, the

defendant's guideline sentencing range is determined in part' by

the amount of loss." United States v. Flete-Garcia, 925 F.3d 17,

28 (1st Cir. 2019) (quoting United States v. Naphaeng, 906 F.3d

173, 179 (1st Cir. 2018)). The loss amount "is the greater of

actual loss or intended loss," with intended loss being "the

pecuniary harm that the defendant purposely sought to inflict."

USSG § 2B1.1 cmt. n.3(A). "Since intended loss normally subsumes

actual loss, intended loss is often the greater of the two."

Flete-Garcia, 925 F.3d at 28.

District courts have "considerable discretion in

determining what evidence should be regarded as reliable in making

findings as to the amount of loss." Id. And a district court's

loss calculations "need not be precise: the sentencing court need

only make a reasonable estimate of the range of loss." Id.; accord

USSG § 2B1.1 cmt. n.3(C). For example, a district court can

estimate loss by looking to factors like "[t]he approximate number

of victims multiplied by the average loss to each victim" and "the

scope and duration of the offense." USSG § 2B1.1 cmt. n.3(C).

The government bears the burden of proving the loss

amount by a preponderance of the evidence. Flete-Garcia, 925 F.3d

- 18 -
at 28. We review the district court's factual findings as to the

loss amount for clear error. Id. at 26, 32-33.

Here, the district court accepted the government's

submission that Akoto and his coconspirators had filed at least

310 fraudulent tax returns seeking $1,326,633 in refunds, $551,601

of which was paid out. The government reached these figures

through a multistep process. First, the IRS examined emails in

three email accounts that the trial evidence established were used

in the conspiracy (kwa2kg@yahoo.com, kwa22kg@yahoo.com, and

nana2kg@gmail.com).6 The IRS flagged any PII found in these

emails, and then determined whether this PII had been used in tax

returns. If so, the IRS took further steps to identify whether

these tax returns were in fact fraudulent, such as by comparing

the wage information reported on the return with wage information

reported to the IRS by employers. This process identified the 310

fraudulent returns that utilized PII found in the email accounts

used in the conspiracy.7 The government provided Akoto with a list

6 There was evidence that the second two accounts were
also controlled by Akoto. For example, Quaye testified that
nana2kg@gmail.com was one of Akoto's accounts (in addition to
kwa2kg@yahoo.com), and that Akoto identified himself by placing
"2kg" in the usernames of his accounts. The government did not
seek to directly establish Akoto's control over the second two
accounts, presumably because the evidence showed that he
controlled the first account and that, at the least, the second
two accounts were used by coconspirators to further the conspiracy.
7 The IRS initially identified at least 490 fraudulent tax
returns that utilized PII found in these email accounts, amounting
to $2,363,349 in requested refunds and $665,728 in paid refunds.

- 19 -
of these 310 returns, including the tax return ID number, the names

and Social Security numbers used, the tax year, the bank

information used, the refund amount requested, and the refund

amount paid. The total refund amount fraudulently requested was

$1,326,633, and the total refund amount paid was $551,601. Akoto

did not challenge any of these specific fraudulent returns as not

being associated with the conspiracy.

Based on these figures, the district court determined

that the intended loss, and thus the amount of loss, was

$1,326,633. Because this loss amount was over $550,000, it

resulted in a 14-level increase in Akoto's total offense level.

See USSG § 2B1.1(b)(1).

As to Akoto's first challenge, we conclude that the

district court did not clearly err in determining that the 310

fraudulent returns were tied to Akoto and his coconspirators by a

preponderance of the evidence.

Each of the 310 returns at issue utilized PII found in

the conspiracy's email accounts. And the methodology the

government used to identify these 310 returns was virtually

identical to that which we approved in Flete-Garcia.8 In that

However, after further consideration the government opted for a
more conservative loss amount figure based on the 310 tax returns.
8 We focus on Flete-Garcia's treatment of the $5 million
in intended loss resulting from tax returns that the conspirators
in that case filed but which were rejected by the IRS. See 925
F.3d at 31-33. The government was able to substantiate the $7.7

- 20 -
case, as here, the government took PII possessed by the

conspirators, identified tax returns that utilized this

information, determined whether the returns were in fact

fraudulent, then prepared summary charts cataloging the losses.

See id. at 31-32. In affirming the district court's loss

calculation, we emphasized the fact that "the record shows with

conspicuous clarity that the IRS used the PII" possessed by the

coconspirators "to identify the suspect tax returns." Id. at 32.

The fact that unrelated fraudsters may also have

possessed some of this PII does not render the loss amount clearly

erroneous so long as there is sufficient evidence tying Akoto to

the loss. Indeed, Akoto concedes that the government did not have

to "eliminate the possibility" that an unrelated culprit filed

some of the fraudulent returns at issue. Flete-Garcia is again

instructive. There, we held that the district court did not

clearly err in accepting a $5 million intended loss figure even

though one of the defendant's coconspirators had "admitted to doing

some 'freelancing,'" including by selling lists of PII to others,

and certain tax returns in the government's summary charts could

million actual loss figure in Flete-Garcia by reviewing the
conspirators' bank accounts for evidence of tax-refund checks.
Id. at 29. Such an approach to actual loss was not viable in this
case because Akoto and his coconspirators structured their scheme
to conceal receipt of the proceeds (by routing the refunds to
prepaid debit cards in others' names and then making withdrawals
in cash).

- 21 -
be read to have been filed by individuals outside of the

defendant's conspiracy. Id. (noting Flete-Garcia's argument that

the addresses and W-2 forms associated with some returns pointed

toward other culprits). We reasoned that "the [district] court

reasonably credited the government's explanation that [the

freelancer's] separate activities were not included in the loss

calculation" and that, viewing the record as a whole, there was no

clear error. Id.; see id. at 32-33. The mere possibility that

others possessed the PII at issue was not dispositive.

So too here. The fact that the conspiracy possessed the

PII used in the fraudulent returns is itself strong evidence of

culpability. See United States v. Clayton, 108 F.3d 1114, 1118-

19 (9th Cir. 1997) (holding that the fact that the defendant

possessed 29 stolen phone ID numbers and had illegally cloned at

least two of them "support[ed] the district court's inference that

he was responsible for the loss associated with the remaining

stolen numbers found in his possession"). And there was also

additional evidence beyond this fact of possession which tied Akoto

and his coconspirators to the fraudulent returns.

To begin, Akoto overstates the potential for overlap

between the PII found in the conspiracy's email accounts and PII

possessed by unrelated fraudsters. Ngo sold PII to individuals

around the world and often resold the same fullz. But Ngo

maintained a total inventory of more than 176,000 fullz; not all

- 22 -
of his clients necessarily possessed the fullz found in the

conspiracy's email accounts, which constituted a tiny fraction of

Ngo's total inventory. And Ngo sold fullz between 2007 and 2013,

whereas Akoto was charged based on conduct between 2011 and 2013.

All 310 of the fraudulent tax returns at issue were for tax years

2010 through 2012 (corresponding to calendar years 2011 to 2013),

the time period when Akoto and his coconspirators were actively

filing such returns.

Further, Akoto's practice was to specifically request

new information that would work for the scheme. He asked Ngo for

"fresh one[s] that no one has" and "the newest info that you have."

In November 2011, he asked for information "for 2011." In April

2012, he asked for "2011 and 2012, if possible." His goal was to

obtain information that "would pass" for purposes of the fraud.

Ngo testified that Akoto wanted to purchase "up-to-date," newer

information, and that newer fullz were less likely to have been

sold to many other purchasers. And Agent O'Neill testified that,

in the context of this investigation, he understood "fresh ones"

to mean "PII that ha[d] been recently stolen or acquired and not

sold to anyone else." Fullz that had already been used by other

fraudsters would not work for purposes of Akoto's fraud scheme, so

he specifically requested newer fullz that "would pass."

We add a final point. As in Flete-Garcia, the district

court here "was operating with a substantial cushion." 925 F.3d

- 23 -
at 32. The court found that the amount of loss was $1,326,633.

Any loss amount above $550,000 would have led to the same 14-level

increase in Akoto's offense level. See USSG § 2B1.1(b)(1). So to

demonstrate reversible error, Akoto must convince us that roughly

$775,000 -- or almost sixty percent of the loss amount attributed

to him by the district court -- was the product of clear error.

See Flete-Garcia, 925 F.3d at 32. He has not done so.9

9 Akoto directs our attention to United States v. Cabrera,
172 F.3d 1287 (11th Cir. 1999), a case in which the defendant was
prosecuted for knowingly possessing telephone cloning equipment.
Id. at 1289. At sentencing, the government offered a loss amount
based on the electronic serial number/mobile identification number
("ESN/MIN") combinations found in Cabrera's possession. Id. at
1290-91. The Eleventh Circuit rejected this approach. Reasoning
that "[m]ultiple unauthorized users often use the same ESN/MIN
combinations simultaneously" and "sellers provide the same ESN/MIN
combinations to multiple buyers," the Eleventh Circuit held that,
to attribute telephone cloning fraud loss to a defendant, the
government must "provide evidence specifically linking the amount
of fraud loss to the defendant's cloning activities." Id. at 1292;
see also id. 1293-94.
It is first worth noting that the Ninth Circuit reached
the opposite conclusion as Cabrera on similar facts. See Clayton,
108 F.3d at 1118-19. In addition, our case is distinct from
Cabrera in an important way: while in Cabrera, the government's
loss amount was based on ESN/MIN combinations that it found "on
Cabrera's handwritten list," "computer" and "computer disks," 172
F.3d at 1293, here, the government's loss amount is based
exclusively on PII that Akoto exchanged by email, often with
coconspirators. It is fair to presume, then, that Akoto intended
this PII to be used in the scheme, and therefore "intended" the
"loss" that would result from its successful use. See also
Flete-Garcia, 925 F.3d at 31 (approving intended loss amount based
on PII that defendant "gave" to coconspirator). Thus, even if we
were to agree with Akoto that the government did not sufficiently
tie him to the fraudulent returns to support its "actual loss"
calculation, the government has certainly done enough to support
its "intended loss" calculation, which is sufficient to establish
the "loss amount" for the purposes of sentencing. See USSG § 2B1.1

- 24 -
 As to Akoto's second argument, we conclude that the

district court did not clearly err by declining to adopt Quaye's

$364,758 loss amount as the loss amount attributable to Akoto.

The government calculated the loss attributable to each defendant

using the same multi-step methodology. The difference was that

the universe of emails associated with Akoto was simply broader.

For Quaye, the government limited its review to emails exchanged

between Quaye's email address and Akoto's email address, whereas

for Akoto, the government reviewed emails in his kwa2kg@yahoo.com

account and in the two other accounts he corresponded with to

advance the conspiracy (accounts with which Quaye did not

correspond). And the government persuasively contends that this

difference in scope reflects Akoto's deeper involvement in the

conspiracy -- as evidenced by, for example, his initial recruitment

of Quaye into the scheme and his more extensive connections with

coconspirators.

III.

For the foregoing reasons, the judgment of the district

court is affirmed.

cmt. n.3(A) ("[L]oss is the greater of actual loss or intended
loss.").

- 25 -