United States Court of Appeals For the Eighth Circuit ___________________________

No. 24-1827 ___________________________

United States of America

Plaintiff - Appellee

v.

Randolph Jay Forrest

Defendant - Appellant ____________

Appeal from United States District Court for the Northern District of Iowa - Eastern ____________

Submitted: April 15, 2025 Filed: September 2, 2025 ____________

Before ERICKSON, ARNOLD, and STRAS, Circuit Judges. ____________

ERICKSON, Circuit Judge.

Randolph Jay Forrest pled guilty to one count of wire fraud, in violation of 18 U.S.C. § 1343, admitting he and his co-conspirators rolled back odometers on used cars. The district court1 ordered Forrest to pay $140,178.56 in restitution. Forrest

1 The Honorable C.J. Williams, Chief Judge, United States District Court for the Northern District of Iowa. appeals, asserting the district court clearly erred when it adopted the government’s calculation, which Forrest contends was based on a novel, untested, and unproved methodology. Because the district court did not abuse its discretion, we affirm.

I. BACKGROUND

From 2012 to 2021, Forrest worked at a used car dealership in Iowa. He and the dealership’s owners rolled back odometers on dozens of vehicles, altered the vehicles’ titles, and resold them without disclosing the true mileage. After the scheme was uncovered, Forrest was indicted on five counts of odometer tampering, four counts of mail fraud, and four counts of wire fraud.

Forrest pled guilty to a single count of wire fraud pursuant to a plea agreement in which he agreed to pay full restitution to all victims, including statutory victims and persons directly and proximately harmed as a result of relevant conduct as determined by the court at sentencing. See 18 U.S.C. § 3663A. In a signed stipulation of facts, Forrest acknowledged that he personally requested odometer recalibrations on at least 27 vehicles. Seventeen of the vehicles were identified in a chart in the PSIR, two were rolled back by around 80,000 miles, others in the 10,000 to 30,000 range, for an average of about 47,000 each. With no agreement reached on the amount of loss or restitution, the district court was presented with several options at sentencing.

One methodology was proffered by Forrest’s expert, Randall McCathren. McCathren opined that because the effect of an odometer rollback is highly dependent on the vehicle’s age, condition, and initial mileage, the appropriate approach was to assume an average condition of the vehicle and determine an average value by relying on sources such as the Kelley Blue Book and Hearst’s Black Book. Using that method, McCathren found the fraud amount caused by the odometer rollback averaged about $810 per vehicle for a total loss amount of $38,070.

-2- Another methodology was suggested by the pretrial services and probation office in the Presentence Investigation Report (“PSIR”). This calculation applied a 40% loss valuation approved by the Fifth Circuit in United States v. Whitlow, 979 F.2d 1008 (5th Cir. 1992). Under this approach, the loss in this case amounted to $76,690.

The government proposed a third option, which was to calculate the loss by totaling the price each victim paid for their altered vehicle. The sum of the purchase prices of the 47 vehicles was $180,299. Another proposal was to impose restitution in the amount of $116,334. This figure represented the co-conspirators’ estimated profit, which was arrived at by totaling the difference between the price the dealership paid for a car and the price a victim paid to purchase the car. The other option presented to the district court was from Howard Nusbaum, an expert retained by the government. Nusbaum disagreed with McCathren’s calculation because it failed to account for the effect the branded title had on the value of the vehicles in question. Nusbaum assessed the diminished value of a vehicle by calculating the value as misrepresented against the actual remaining service life of the vehicle if it had been truthfully represented.

Nusbaum’s approach assumed the average driver expected to get approximately 150,000 miles of use from a vehicle. If a vehicle had already exceeded that number, Nusbaum adjusted that expectation upward because a purchaser would not buy a car that he or she believed had no useful life left. Nusbaum compared the estimate of a vehicle’s anticipated life with its true mileage. In his calculations, a vehicle with a significant rollback such that an informed buyer should have expected immediate failure retained no value. For vehicles with a useful life, Nusbaum calculated a percentage. For instance, if a purchaser received a car with only 40 percent of its represented useful life, Nusbaum estimated the loss at 60 percent of the vehicle’s purchase price. Using this method, he estimated the total loss at $140,178.55.

-3- The district court rejected the approach used by the Fifth Circuit in Whitlow because it did not account for the branded title affixed to a vehicle involved in a rollback. The court rejected McCathren’s calculation for this same reason—the Kelley Blue Book and Hearst’s Black Book values assume the title and mileage on the vehicle are valid. The district court noted that in many ways the loss should be the entire purchase price because the victims would not have bought the car in the first place had they known the vehicles had rolled back mileage. But, because the buyer got something of value, the district court found some consideration to value must be factored into the analysis. Given there were at least 47 vehicles identified in the scheme, the court found the collection of specific data and underlying information from witnesses to calculate an exact loss amount for each vehicle would be “cost prohibitive and logistically unrealistic.”

The district court set forth several reasons for its decision to rely on Nusbaum’s calculations, including: (1) he was an “eminently qualified” expert in vehicles, specifically in rollback schemes and the impact on the market; (2) the assumptions underlying his calculations were reasonable; and (3) his methodology was reasonable. The court expressly rejected Forrest’s assertions that the purchase prices listed in the government’s exhibit, and relied on by Nusbaum, were unreliable such that the actual price paid by the victims was unknown or the dealership might have made improvements to the vehicles that would increase their value. In rejecting these arguments, the court noted the business model utilized by the co-conspirators was to obtain vehicles with high mileage, roll back the odometers, and flip them as quickly as possible, with cleaning being the only likely change to the vehicle prior to the sale. The court further found the purchase prices listed by the government likely understated the actual amount of money paid to the co-conspirators because part of the scheme was to underreport the purchase price and take money under the table so the buyers could reduce the amount of taxes they had to pay for the car.

On appeal, Forrest contends the district court clearly erred when it adopted Nusbaum’s methodology that lacked sufficient individualization and did not provide “any credit for actual, measurable value that most purchasers enjoyed and retain[ed], -4- despite their vehicles having inaccurate odometer clusters.” Forrest asserts some, if not many, of the vehicles are still currently on the road and even non-working vehicles have value for trade, parts, salvage, materials, etc. Forrest claims the district court awarded 32 of the 47 victims a clear windfall and the remaining purchasers will be reimbursed at a high rate of 36% of their combined total purchase price.

II. DISCUSSION

We review restitution awards for abuse of discretion and factual findings supporting the amount for clear error. United States v. Nelson, 106 F.4th 719, 724 (8th Cir. 2024). No clear error exists if the determination is plausible in light of the entire record. Id.

It is the government’s burden to prove by a preponderance of the evidence the victims’ losses. United States v. Ruff, 420 F.3d 772, 775 (8th Cir. 2005). In fraud cases, “courts must be wary of over-inclusiveness” when determining actual loss. United States v. Hansmeier, 988 F.3d 428, 440 (8th Cir. 2021). “Restitution ‘must be based on the amount of loss actually caused by the defendant’s offense,’ and the burden is on the government to ‘prove that the restitution awarded does not exceed the actual, provable loss caused by the offense.’” Id. (quoting United States v. Fonseca, 790 F.3d 852, 854 (8th Cir. 2015)). On the other hand, the defendant bears the burden of showing financial resources and needs. Ruff, 420 F.3d at 775. The burden as to all other matters “shall be upon the party designated by the court as justice requires.” Id. at 776 (quoting 18 U.S.C. § 3664(e)).

While Forrest contends the district court’s restitution award was insufficiently individualized, a district court is obligated only to reasonably estimate the loss “when the amount lost through fraud is difficult to estimate.” United States v. Garbacz, 33 F.4th 459, 473 (8th Cir. 2022) (quoting United States v. Adejumo, 848 F.3d 868, 870 (8th Cir. 2017)). This principle does not allow a court to overstate a victim’s loss simply because determining the loss is complicated due to the complex

-5- nature of the scheme. In every case, restitution awards are “limited to the victim’s provable actual loss.” Id.

Here, Forrest stipulated to pay restitution for the full amount of the victims’ losses, including offenses dismissed as a result of the plea agreement and any uncharged offenses that were part of this scheme. Due to the number of vehicles involved, the court found that obtaining all the records and documentation regarding a vehicle’s current location and condition, repair bills incurred by the victims, and subsequent sales made a precise loss calculation infeasible. Thus, each party was given an opportunity to present evidence and arguments on the amount of provable loss. Cf. United States v. Sutton, 520 F.3d 1259, 1264 (10th Cir. 2008) (collecting cases and observing “there is simply more than one permissible way to measure loss in criminal odometer tampering cases”).

After receiving conflicting evidence, the district court noted the vehicles’ fair market value, now with a branded title, was diminished to the point that it would not be unreasonable to assess the loss at the purchase prices. It did not do so. The district court recognized that the vehicles, although marred with a branded title, may have some value, although not the value provided for in the calculations that were based on assumptions of valid title and mileage. While Forrest believes more weight should have been given to the remaining useful life of the vehicles, he ignores the notion that the loss calculation did not account for other losses that could have been attributed to his conduct, including towing bills for broken down cars and other unexpected repair bills due to the misrepresentation of the vehicle’s mileage. In some cases, the court noted the victim could have ended up losing more money than the purchase price.

The dissent suggests unexpected repairs are simply an accelerated expense, rather than an expense buyers might otherwise have avoided. However, Nusbaum testified the cars in Forrest’s scheme were probably purchased from individuals who held them until they became inoperable or prohibitively expensive to repair. The district court also concluded that a cleaning was likely the only change to the -6- vehicles prior to sale. As a result, the buyers in the scheme were not necessarily faced with routine repairs on an unexpected schedule. Rather, they faced prohibitively expensive repairs of the sort that would convince an ordinary person to give up on their vehicle—or, in the victims’ case, to refrain from purchasing it. That some victims could have spent more on repairs and related expenses than they paid to purchase their vehicle supported the court’s conclusion “that the restitution awarded [did] not exceed the actual, provable loss caused by the offense.” Hansmeier, 988 F.3d at 440 (citation omitted).

The district court’s findings with respect to the nature of Forrest’s scheme reinforce the point. At the urging of Forrest and his co-conspirators, buyers agreed to underrepresent the purchase price of their vehicle to the state in which they registered it. For example, the district court reviewed evidence that one buyer paid $4,000 cash for his vehicle and yet received a receipt stating the car had been purchased for $1,500. The same buyer submitted evidence that he later incurred $3,500 in repair costs. The district court found this sort of underreporting was part of the scheme and, as a result, the government’s figures likely understated the amount paid by the victims to Forrest and the dealership.

Forrest also contends the district court ignored the value of the vehicles for trade, parts, salvage, or other materials. His contention is unpersuasive for two reasons.

First, salvage value is a type of resale value, and the district court concluded loss estimates based on resale value were a poor fit for this case when it adopted Nusbaum’s estimate. Nusbaum testified his estimate turned on use rather than resale value because the victims in Forrest’s scheme were unique: they sought cheap, high mileage cars and did not plan to resell those cars. Because they planned to drive their cars as long as possible, they paid for and negotiated use value—without any expectation they might recoup a portion of their investment by reselling their vehicle. Forrest did not offer evidence to undermine the district court’s conclusion that use value, not resale value, was the best measure of his victims’ losses. -7- Second, the salvage value or the value of component car parts is not self- evident. Forrest had the burden to demonstrate his theory of value fit the facts of his case. Although the government must prove the victims’ losses by a preponderance of the evidence, the court may allocate the burden of proof as to all other matters. Ruff, 420 F.3d at 775–76. The district court offered Forrest and the government an opportunity to litigate the appropriate amount of loss. Forrest took that opportunity, but he did not present evidence of salvage and parts value. He did not attempt to rebut the government’s theory that use value was the best measure of his victims’ losses, either. As the proponent of a rival theory of value, Forrest bore the burden to demonstrate his method was correct and support his demonstration with evidence. See United States v. Arrington, 97 F.4th 593, 596 (8th Cir. 2024) (noting that seven circuits place the burden of proving an offset on the defendant and permitting the court to assign the burden to the defendant). Without supporting evidence, the district court had neither a basis nor an obligation to speculate about the impact of salvage or parts value. We cannot conclude the district court clearly erred based on Forrest’s speculation. See id. (concluding the defendant “presented sufficient evidence to establish” a claimed offset); United States v. Karie, 976 F.3d 800, 806 (8th Cir. 2020) (defendant presented no evidence to support his offset theory).

We recognize the loss amount calculated in this case is relatively close to the sum of the purchase amounts listed in the government’s exhibit; however, part of the scheme, as found by the district court, was to conceal the actual purchase price by accepting money on the side and underrepresenting the purchase price to allow the purchaser to pay less in taxes. Thus, the difference between the loss amount as found by the district court and the total amount paid by the buyers is likely greater. Moreover, in other circumstances, a sentencing court may need to consider some of the details missing in this case, such as an estimate that accounts for salvage value. See Karie, 976 F.3d at 805. But the record here is devoid of such evidence. Forrest has not provided a convincing reason demonstrating that the district court’s findings were clearly erroneous, or that the district court abused its discretion when it estimated the loss involved in this odometer tampering scheme. -8- III. CONCLUSION

For the foregoing reasons, we affirm the judgment of the district court.

ARNOLD, Circuit Judge, dissenting.

Because I cannot endorse the economic theory that underpins the district court’s restitution award, I respectfully dissent. Randolph Forrest participated in an odometer tampering scheme, and the district court purported to credit him with the value of the vehicles that he and his coconspirators sold in determining the restitution that he had to pay for his victims’ losses. But it calculated the after-credit losses by adopting the government’s model treating nearly seventy percent of those vehicles as worthless. It takes no special insight to realize that cannot be right. Automobiles have value; everyone knows that. See Vogt v. Progressive Cas. Ins. Co., 129 F.4th 1071, 1073 (8th Cir. 2025). The best the court can say for the model is that it might be less wrong than it seems. But there is no need to rationalize a valuation the record does not support, no matter how difficult it may be to estimate what Forrest’s victims lost. The district court had options to simplify the problem and reach a defensible estimate, either by shifting the burden of proving the credit to Forrest or by denying restitution if calculating the amount lost turned out to be too complex. See 18 U.S.C. §§ 3663A(c)(3), 3664(e). I would vacate the restitution award and remand to let it consider those options.

As the court and the parties do, I assume that the restitution award should have equaled the net losses suffered by Forrest’s victims after accounting for the value of the vehicles that they received. Cf. Robers v. United States, 572 U.S. 639, 640–45 (2014). On that assumption, the district court had to make a reasonable estimate of the victims’ net losses, see United States v. Adejumo, 848 F.3d 868, 870 (8th Cir. 2017), and the government bore the burden to “prove that the restitution awarded [did] not exceed the actual, provable loss caused by the offense.” See United States v. Hansmeier, 988 F.3d 428, 440 (8th Cir. 2021). -9- The government fell well short of carrying that burden. Its unique restitution model, never before used in litigation, produced results that were unreasonable on their face. Here is how it worked. First, the government’s expert assumed that victims expected their vehicles to last about 150,000 miles, or occasionally a little longer. Then, he subtracted each vehicle’s advertised mileage from its expected mileage to estimate how many miles the victim who purchased it thought it would have left. He also subtracted an estimate of each vehicle’s actual mileage from its expected mileage to estimate how many miles it actually had left. The ratio between actual remaining mileage and anticipated remaining mileage was the fulcrum of the model. It represented the percentage of the mileage a victim thought he was purchasing that he purportedly received. To determine the value of a vehicle, the expert multiplied this percentage by an estimate of the vehicle’s purchase price. If the value was negative, he set it to zero. All the expert had to do to transform his calculation of value into an estimate of a victim’s loss was subtract it from the estimate of the purchase price.

When the expert performed these steps, the model told him that nearly seventy percent of the vehicles had zero value, a clear sign that the model made no sense. Even vehicles in poor condition have some value as salvage. See Vogt, 129 F.4th at 1073. The model just assumed that they did not. All value had to come from remaining mileage, or the model did not count it.

To make matters worse, the model was doing no better in valuing mileage. It said that none of the zero-value vehicles could run another mile, which was demonstrably untrue because the victims were driving away in them. The mistake behind this error was the more or less arbitrary assumption that vehicles would only run for about 150,000 miles in their lifetimes. If a vehicle had already accrued more than the expected lifetime mileage when a victim drove off in it, the model assumed it could not run at all and treated it as worthless, regardless of what the evidence showed.

-10- Even the expert who designed the model did not think that was right. At least not really. The expert departed from the 150,000-mile assumption when Forrest or his coconspirators advertised that a vehicle had close to or more than 150,000 miles. He reasoned that the vehicle was “better than normal” and might “go to 170,000 miles” or more. In other words, the probability that a vehicle would run for a given number of miles depended on the number of miles it had already run. If a vehicle exceeded its expected lifetime mileage before a victim bought it and drove away in it, the expectation needed to change. Considering its track record, the vehicle might keep running, as some or all of the vehicles did. The model, however, largely failed to account for this additional mileage because the expert only adjusted vehicles’ expected lifetime mileages when their advertised mileages bumped up against them. When vehicles’ advertised mileages were lower but their actual mileages surpassed their expected lifetime mileages, the expert ignored his own reasoning and assumed that the vehicles would not work. Setting aside salvage value, that is precisely the error that made the model treat so many vehicles as worthless.

The model’s inconsistency is all the more indefensible because the victims themselves offered evidence that the vehicles they purchased ran and had value. Most of the victims who provided evidence stated that their vehicles were drivable, either without problems or with repairs. The rest did not include vehicle troubles among the losses they suffered. And all but one of the victims who claimed restitution of specific loss amounts asked for less than the full purchase prices of their vehicles, which means that they did not believe that their vehicles had zero value. Yet the model treated even these victims’ vehicles as worthless merely because they exceeded expected lifetime mileages that the government’s expert failed to adjust properly. As a result, the victims in this group recovered the full purchase prices, costing Forrest over $6,000 in restitution that the victims never requested. In fact, if the district court had taken every vehicle that the model deemed worthless and valued it at the lowest price any of these victims implicitly admitted their vehicles were worth, it would have shaved about $20,000 off Forrest’s restitution obligation.

-11- In short, as one of the government investigators himself conceded, “all the victims in this case [got] something of value”; the government’s model just pretended that a large majority of them did not. That is not a reasonable way of estimating the victims’ net losses.

The court suggests that the victims might nevertheless have suffered unexpected repair costs that counterbalanced some of the value they got from their vehicles. Maybe so, but not much of it. Vehicles need repairs as they accumulate mileage. This was no secret to the victims; only the mileage of their vehicles was. That means that an unexpected repair would generally be a repair that a victim expected to make later but instead had to make earlier because his vehicle had more mileage than he thought. Even if the repair were costly or clustered with other repairs, the loss to the victim would still only be the cost of accelerating the expense and not the full cost of the repair. Think of it like losing some interest on the expended amount rather than losing the amount itself, which the victim would have to expend on the repair later anyway.

The court may be right that prospective repair costs would have deterred some victims from purchasing their vehicles if they had known the vehicles’ true mileages, but that is beside the point. The vehicles had value, and the victims retained them, yet the district court compensated the victims as if they had not. That was a windfall regardless of whether odometer tampering induced the victims to purchase the vehicles. See United States v. Allen, 529 F.3d 390, 395–97 (7th Cir. 2008). If victims could obtain this sort of bonus recovery simply because they entered transactions due to fraud, actual loss would no longer be a limit on restitution awards in most fraud cases. Few victims would deal at a price inflated by fraud unless they were unable to obtain a better price elsewhere. Just consider the investor who thought he was buying a business but only got a business license (and a van), see United States v. Matsumaru, 244 F.3d 1092, 1109 (9th Cir. 2001), the tribe that thought it was paying for mold testing by an expert but received services from someone less qualified, see Allen, 529 F.3d at 391–92, 395–97, the man who thought he was lending on the security of a car and tools but could only recover the tools because -12- the borrower secretly pledged the car to other lenders, see United States v. Williams, 292 F.3d 681, 684, 689 (10th Cir. 2002), and the military base that thought it was buying goods and services but only received some of them, see United States v. Huff, 609 F.3d 1240, 1244, 1248–49 (11th Cir. 2010)—none of whom could previously recover the sort of windfall that the district court allowed here.

It does not matter that gross repair costs and similar expenses might have exceeded purchase prices. The high costs of operating a vehicle, assuming they were indeed high, did not negate the vehicle’s value. That would be a little like saying a printer is worthless because the cost of replacing its ink cartridges exceeds its purchase price. It does not account for the benefits the product provides.

And it is far from clear that gross repair and similar costs outweighed the vehicles’ benefits, in use and in salvage, just because those costs might have been greater than the vehicles’ purchase prices. One reason is that a rational vehicle owner would not pay to repair and operate a vehicle if the benefits of operating the repaired vehicle were less than the costs of doing so. Another is that gross costs include costs that the vehicle owner expected, like the costs of future repairs before adjusting for their acceleration. If these costs exceeded the vehicle’s benefits, the owner would not have bought the vehicle. All that matters is the residuum of unexpected costs, and those are unknown.

That points the way to a larger problem. Whatever the victims’ unexpected repair and similar costs might have been, the notion that they balanced out the values of the victims’ vehicles is speculative. So far as the record reveals, no victim had to pay for repairs he did not expect. The government’s expert expressly declined to model such repairs “because no evidence was presented to” him that they occurred. And assuming it is relevant, the possibility that vehicle repair costs exceeded vehicle purchase prices is, as the court acknowledges, merely something that “could have” happened. The record does not even contain a back-of-the-envelope estimate for how often unexpected repairs might be necessary and what they might cost given these uncertainties. For other unexpected costs, the story is much the same. That the expert -13- may have underestimated some victims’ losses because they underreported their purchase prices does little to fill the void in the record; the expert himself testified that he could not account for underreported prices because they were insufficiently substantiated. Without evidence for the balancing theory, the argument boils down to this: if the government’s model had measured what it did not measure, the result would compensate for the obvious errors in what it did measure. Everything would supposedly cancel out—at least well enough.

With respect, the court’s attempt to cast doubt on the salvage value of the victims’ vehicles in defense of this conclusion is also unsuccessful. The court dismisses the vehicles’ salvage value as the product of Forrest’s speculation. But the fact that the vehicles had such value is indisputable even though its magnitude was uncertain. And it makes no difference whether the salvage value was “a type of resale value” as the court says. The court’s paraphrase notwithstanding, the government’s expert did not conclude that the victims bought their vehicles “without any expectation they might recoup a portion of their investment by reselling” them. It would not matter if he did since the vehicles had salvage value whether or not the victims expected to extract it at the time of purchase. But the expert only questioned whether victims would resell their vehicles in typical transactions since he suspected that they would drive the vehicles until they died. At that point, he opined, they would dispose of the vehicles. Nothing in his testimony implied that they would dispose of them without recovering their salvage value or explained why they would take such an irrational course.

More fundamentally, the focus on what Forrest demonstrated about salvage value turns the burden of proof on its head. The government, not Forrest, had to show that the restitution award was no greater than the loss that Forrest’s offense caused. See Hansmeier, 988 F.3d at 440; Adejumo, 848 F.3d at 870. In that task it failed. The victims’ vehicles had some value, in salvage or in future use, yet the government neglected to credit Forrest with the value of nearly seventy percent of those vehicles. If losses not captured by its model offset that credit, the government never proved it. -14- The district court’s authority to shift the burden of calculating the credit to Forrest is no help to the government in the present posture of the case because the district court did not exercise that authority. Under 18 U.S.C. § 3664(e), a district court may “designate[]” the party that has the burden of proof on most restitution issues “as justice requires.” And our decision in United States v. Arrington establishes that the magnitude of a restitution credit for property given to a victim is one such issue, at least in some cases. 97 F.4th 593, 596 (8th Cir. 2024). But the district court did not designate Forrest as the party with the burden to prove the magnitude of the credit for the value of the victims’ vehicles. Its only comment on the burden of proof was a statement that the government had the burden of proving the victims’ losses.

That Forrest offered a method of estimating the victims’ losses to “rival” the government’s method did not shift the government’s burden. For the sake of argument, I accept the court’s conclusion that Forrest assumed “the burden to demonstrate his method was correct and support his demonstration with evidence.” But that burden has no bearing on this appeal. The district court did not adopt Forrest’s method. It adopted the government’s. And by the very same reasoning, the government assumed the burden to demonstrate that its method was correct and support its demonstration with evidence. That is the burden it failed to carry.

The difficulty of estimating the victims’ losses does not excuse the deficiencies of the government’s method. Though the government emphasizes the complexity of the estimation, the district court had better ways to handle the challenge than stretching what we mean by a reasonable estimate beyond its reasonable limits. One of them was making a proper designation to shift to Forrest the burden to prove any credit for the value of the victims’ vehicles. See id. If the district court found that “the number of identifiable victims [was] so large as to make restitution impracticable,” a second was to deny restitution. See 18 U.S.C. § 3663A(c)(3)(A). And the district court could have done the same if it found that “determining complex issues of fact related to the cause or amount” of the victims’ -15- losses “would complicate or prolong the sentencing process to a degree that the need to provide restitution to any victim [was] outweighed by the burden on the sentencing process.” See id. § 3663A(c)(3)(B); see also United States v. Martinez, 690 F.3d 1083, 1089 (8th Cir. 2012). Since the district court did not consider any of these options, I would vacate the restitution award and remand so it could do so. ______________________________

-16-