WASHINGTON MUTUAL, INC., as successor in interest to H.F. Ahmanson & Co. and Subsidiaries, Plaintiff-Appellant, v. UNITED STATES of America, Defendant-Appellee, 856 F.3d 711

OPINION ENGLAND, District Judge: Plaintiff-Appellant Washington Mutual, Inc. (“Appellant”), as successor in interest to H.F. Ahmanson & Co., and Ahman-son’s wholly owned subsidiary Home Savings of America (“Home”), appeals from a judgment entered in favor of Defendant-Appellee United States of America (“Government”) after a bench trial in this tax refund action. Appellant argued in the district court that it was entitled to refunds attributable to losses and deductions it should have been afforded for certain intangible assets acquired during the savings and loan crisis of the 1970s and 1980s.

The district court, however, determined that the valuation model relied upon by Appellant’s expert was fundamentally flawed. As such, the district court held that Appellant failed to meet its burden to establish the value for the intangible assets, as well as its burden to establish a cost basis in those assets—a necessary requisite to allowing amortization deductions for those assets.

Further, the district court determined that Appellant failed to show that it abandoned the Missouri Branching Right when it closed its Missouri deposit-taking branches and, therefore, that it was not entitled to an abandonment loss deduction. As a result, the district court dismissed the case. We have jurisdiction under 28 U.S.C. § 1291, and we affirm. I A The parties’ dispute evolved out of transactions originating from the savings and loan crisis.

During the 1970s and 1980s, savings and loan associations, or “thrifts,” saw their profitability dissipate when the Federal Reserve chose to remedy rising inflation by allowing interest rates to skyrocket. See United States v. Winstar Corp., 518 U.S. 889, 844-45, 116 S.Ct. 2432, 135 L.Ed.2d 964 (1996). Thrifts were consequently forced to pay depositors higher interest rates, while the thrifts’ income streams, which derived from long-term mortgage loans with low, fixed rates, remained stagnant. Id. at 845, 116 S.Ct. 2432.

The high interest rates also decimated the housing market, further drying up the thrifts’ revenue streams and forcing the entire industry towards insolvency. See H.R. Rep. No. 101-54(1), at 296 (1989). In the event that a thrift’s liabilities exceeded its assets, the Federal Savings and Loan Insurance Corporation (“FSLIC”), as thrift regulator and insurer of thrift deposits, was required to initiate a takeover and liquidate the thrift. See Winstar, 518 U.S. at 844-47, 116 S.Ct. 2432.

FSLIC lacked the funds necessary to liquidate all of the thrifts that were failing at the time, however, and the Federal Home Loan Bank Board (“Bank Board”) instead chose to encourage healthy thrifts to agree to such takeovers through what were referred to as “supervisory mergers.” Id. at 847, 116 S.Ct. 2432. In order to make these supervisory mergers attractive to healthy thrifts, the FSLIC had to offer non-cash incentives, two of which—both exempting limitations otherwise imposed on the operations of savings and loan associations—are especially relevant here. Id. at 848, 116 S.Ct. 2432; see also Wash. Mut.

Inc. v. United States, 636 F.3d 1207, 1209 (9th Cir. 2011) (“WAMU F). First, thrifts were historically prohibited from opening branches outside of their home states. WAMU I, 636 F.3d at 1213. Accordingly, the FSLIC offered an incentive to healthy associations hoping to expand nationally by allowing those thrifts an opportunity to operate in a new state if the first branch in that state was acquired through a supervisory merger. Id. This incentive is referred to by the parties as the “Branching Right.” Id. at 1209, 1213.

Second, thrifts were limited by minimum regulatory capital requirements, which mandated that each thrift maintain minimum capital of at least three percent of its liabilities. See Winstar, 518 U.S. at 845-46, 116 S.Ct. 2432. This presented an obstacle to taking over a failing thrift since, by definition, the failing thrift’s liabilities already exceeded its assets. See id. at 850, 116 S.Ct. 2432. To counter this, regulators permitted healthy thrifts agreeing to a supervisory merger to apply the “purchase method” of accounting. Id. at 848, 116 S.Ct. 2432.

Under this method, an acquiring thrift was permitted to designate those excess liabilities as “supervisory goodwill,” which, in turn, could be counted toward the supervisory thrift’s minimum regulatory capital requirement. Id. at 848^49, 116 S.Ct. 2432. Thrifts were also permitted to amortize that supervisory goodwill over a period of 40 years. Id. at 851, 116 S.Ct. 2432. These incentives, which focused on Regulatory Accounting Principles, are referred to as the “RAP Right.” WAMU I, 636 F.3d at 1209, 1213.

Home was a “healthy” thrift and, in 1981, agreed to a supervisory merger by which it would take over three failing thrifts, two in Missouri and one in Florida. Id. at 1211-12. Through a series of transactions Home assumed the liabilities of the failing thrifts in exchange for a “generous incentive package.” Id. at 1219. Under this package, Home received, among other things, cash and indemnities as to covered assets, and was allowed to structure the transaction as a tax free “G” reorganization, giving it significant tax benefits. Id.

Home also received Branching Rights for Missouri and Florida, permitting it to open branches in .those states, as well as the RAP Right and its associated benefits. Id. at 1213,1219. B The^ current litigation arose after Home sold its Missouri branch offices in 1992 and 1993 and was later acquired by Appellant in 1998. Id. at 1209, 1214. Appellant filed amended tax returns on behalf of Home in 2005, requesting refunds for tax years 1990, 1992, and 1993. Id. at 1214.

According to Appellant, the Internal Revenue Service (“IRS”) had not credited Home for its RAP Right amortization deductions during those years, nor had it allowed an abandonment loss deduction in 1993 for the Missouri Branching Right. Id. Based on these denials, Appellant filed suit on Home’s behalf. Id. In its first review of this ease, the district court ruled in favor of the Government at

summary judgment, deciding that Home did not have a cost basis in either the RAP Right or the Branching Rights. Id. at 1216. As such, the district court held that Appellant was not entitled to amortization and loss deduction-related refunds. Id. Appellant appealed, and we reversed, holding that “Home Savings had a cost basis in the RAP rights and the branching rights equal to some part of the excess of the three acquired thrifts’ liabilities over the value of their assets.” Id. at 1209.

The panel remanded with instructions to the district court “to determine the cost basis and conduct further proceedings in accordance to [that] opinion.”./!! at 1221. C On remand, the district court heard evidence over the course of an eight-day bench trial and ultimately held that Appellant had failed to carry its burden of establishing a reliable cost basis for the Missouri Branching Right. Accordingly, the court ruled that Appellant had not established its right to a tax refund. Appellant similarly failed to convince the district court that Home had permanently abandoned its right to operate in Missouri.

As a consequence, the court ruled that it was not entitled to take an abandonment loss for the 1993 tax year. The court dismissed Appellant’s tax refund claims with prejudice and entered judgment for the Government. In reaching its decision, the district court initially reasoned that, based on our prior remand, Appellant was required to do two things to establish the cost basis for each right. First, it needed to establish the “Purchase Price” for the failed thrifts, which could be determined by subtracting the value of the three failed thrifts’ assets from their liabilities. See WAMU I, 636 F.3d at 1219.

Second, Appellant was required to show what portion of the Purchase Price should be allocated among the various rights. Importantly, however, because the Purchase Price was less than the total fair market value of Home’s incentive package, it was not enough for Appellant to determine the fair market value of a single asset and assign to it a proportionate amount of the Purchase Price. Instead, Appellant had to establish the fair market value of each individual asset to reach a total fair market value for the entire incentive package.

Appellant could then use this total fair market value to determine each asset’s proportionate value, and apply that value pro rata to the Purchase Price to establish the cost basis of each asset. The district court explained Appellant’s burden with a helpful illustration: Assume that a thrift paid $300 for an incentive package it received in a supervisory merger. Assume, as well, that the incentive package is comprised of three assets: A, B, and C. Now assume that the fair market value of A is $175, the fair market value of B is $125, and the fair market value of C is $100.

Therefore, the total fair market value of the combined assets constituting the incentive package is $400. This information is not sufficient to allow for the correct allocation of the purchase price among A, B, and C. For instance, one cannot simply allot $175 of the purchase price to A because that would result in too much of the purchase price being allotted to A.

In other words, A would be allotted one-hundred percent of its fair market value, while B and C would be left with some percentage less than one-hundred percent because only $125 of the purchase price would remain to allocate to B and C, whose fair market value is $125 and $100, respectively, for a total of $225. Wash. Mut., Inc. v. United States, 996 F.Supp.2d 1095, 1105 (W.D. Wash. 2014) (‘WAMU IF).

Thus, if Appellant failed to establish the fair market value for any of the individual assets, it would be impossible to determine the total fair market value of the incentive package and, thereby, the pro-rata share of the Purchase Price— and the cost basis—for each individual asset. Id. Considering the evidence adduced at trial, the district court concluded that Appellant “failed to establish, to a reasonable certainty, the fair market value for the Missouri Branching Right” and, therefore, could not establish a cost basis in that Right or in the RAP Right. Id. at 1106.

The district court found that Appellant failed to meet its evidentiary burden in large part due to shortcomings in the testimony of Appellant’s valuation expert, Roger Grabowski. The district court explained the “Grabowski Model” as follows: Mr. Grabowski used an income approach to determine the fair market value of the Missouri Branching Right to a hypothetical buyer. More specifically, Mr. Grabowski used a discounted cash flow model known as the “excess earnings” approach.

Excess earnings represent the cash flows that the hypothetical buyer would have expected the Branching Right to generate beginning in December 1981, net of charges for the use of contributory assets, and discounted to present value. In order to determine the excess earnings (ie. the cash flow) attributable to the Missouri Branching Right under the excess earnings approach, Mr. Grabowski employed a five-step analysis. First, he projected the net operating income generated by the Branching Right based on: (

a) the projected rate of overall statewide thrift deposit market growth in Missouri; (

b) the projected market share that the hypothetical buyer could be expected to capture in Missouri; (

c) the projected spread on loans funded by the new deposits; and (

d) the projected operating expenses for the hypothetical buyer. Second, he deducted income taxes from the net operating income to arrive at the projected net income. Third, Mr. Gra-bowski projected charges for the use of contributory assets (commonly referred to as “capital charges”) and deducted those charges from the net income to arrive at the projected cash flow attributable to the Missouri Branching Right. Fourth, Mr. Grabowski used a 22% discount rate to determine the present value of the projected cash flow. Lastly, Mr.

Grabowski deducted estimated transition costs and assemblage value from the present value of the cash flow. This resulted in Mr. Grabowski finding a fair market value of $28.8 million for the Missouri Branching Right. Id. at 1107 (citations omitted). The Government responded to the Gra-bowski Model, not by offering its own valuation, but by convincing the trial judge that the Model was fundamentally flawed and unreliable as a basis for determining the value of the Missouri Branching Right. For example, the district court determined that Mr.

Grabowski’s assumptions regarding Missouri deposit market growth were unreliable because his projections regarding statewide deposit growth failed to account for the effects of disintermediation and improperly included “interest credited” balances as a source of “new” deposit growth, thereby inflating his calculations. Id. at 1111. The court also discredited Mr. Grabow-ski’s assumption that Home’s ability to capture market share in Missouri could be adequately predicted by looking to its pri- or expansion into Northern California.

The court explained that Home’s intrastate expansion was not a reliable predictor of a hypothetical interstate expansion into the Missouri deposit market because:

(1) Home’s prior expansion occurred in the 1970s when the economic landscape for thrifts was entirely different;

(2) Home’s Northern California expansion was facilitated primarily by acquiring existing thrifts, while the Grabowski Model assumed Missouri growth would be by way of organic expansion; and (3) the Grabow-ski Model contradicted Home’s own internal market-share projections made at the time of the supervisory merger. The court further found that evidence presented at trial did not support the Gra-bowski Model’s underlying assumption that a hypothetical buyer could originate a sufficiently high volume of new loans to offset the costs of all projected new deposits in the market at that time.

According to the district court, this assumption was flawed because it failed to account for the fact that interest rates had risen dramatically and that the thrift industry was loan driven—that is, if loan demand was high, thrifts would seek to attract additional deposits; but if loan demand was weak, earnings would drop if there were an inordinate amount of new deposits. The Grabowski Model, on the other hand, was driven by deposits. Mr. Grabowski assumed that the hypothetical buyer would want to attract as many new deposits as possible, and would worry about originating mortgage income after the fact.

But underlying this assumption was the notion that the buyer would be able to place all newly generated deposits into income-generating mortgages. According to the district court, however, the likelihood of such an outcome was minimal at best, as the evidence produced at trial indicated that the then-existing high interest rates were not conducive to increasing loan volumes. The district court also rejected as “unrealistic” the Grabowski Model’s assumption that a hypothetical buyer could count on a net interest spread of 2.5% between its income-generating loans and its outgoing deposit payments. Id. at 1114.

According to the court, Mr. Grabowski assumed that Appellant would not have to pay the going market rate for new deposits, but would rather be able to attract new, low-interest rate deposits by marketing its secure reputation, just as it did during the Northern California expansion. But since the court had already rejected Appellant’s reliance on intrastate California market growth as a predictor of success in its interstate venture, the district court was quick to reject this assumption. With respect to Mr.

Grabowski’s final assumption, the court faulted the Grabow-ski Model for attempting to identify a license value for the Missouri Branching Right by comparing a “without” scenario to a “with” scenario. Id. at 1115. Under a “without” or start-up scenario, the hypothetical buyer would move into the new Missouri market without assemblage—i.e., branches, personnel, equipment, etc. Under the “with” or baseline scenario, however, the same buyer would enter the same market, but would do so with existing operational branches. Using this framework, Mr.

Grabowski attempted to value the Missouri license by comparing how long it would take a hypothetical buyer in the “without” scenario to reach the same point as a buyer in the “with” scenario—and projected that a “without” buyer would be able to catch up to the “with” buyer within four years. To reach this conclusion, however, the district court noted that the hypothetical “without” buyer in the Model would have to achieve new deposit growth of between 90% and 165% over, that period.

The court observed that such a conclusion “simply def[ied] explanation” because “[a]ssuming such extravagant growth during a time when depositors were fleeing the savings and loan market defle[d] credibility.” Id. The court further faulted the Grabowski Model’s license value projection because it failed to consider that the value of the Missouri Branching Right license might actually decline.

Since other thrifts were similarly obtaining branching rights, the district court determined that the Model should have addressed the possibility that additional thrifts could be granted the right to enter the Missouri market and, thereby, dilute the value of Home’s right. But rather than take this into account, Mr. Grabowski simply projected a perpetually increasing value. Ultimately, the court was more persuaded by the Government’s expert, Dr. Steven Mann. According to Dr.

Mann, the premiums paid for intrastate thrift mergers around the date of the supervisory merger in this case were similar to those paid by Home for the right to expand into the new Missouri and Florida markets. Accordingly, the Government’s theory was that if, as Appellant argued, Home paid a premium for the failing thrifts in 1981 because it wanted the Branching Rights, then a similar premium would not be present for intrastate mergers, which would, by default, not include Branching Rights.

But since the purchase premiums were similar for both types of mergers, the Government argued that the Grabowski Model improperly attributed too large a premium to the Branching Right, and that the purchase premiums must be driven by some other intangible asset associated with the mergers. Appellant argued in response that the Government’s comparisons were flawed because the interstate acquisitions involved assisted takeovers of failing thrifts that were qualitatively different than the intrastate mergers.

According to Appellant, intrastate purchasers benefitted from “marketing and operational efficiencies,” which Mr. Grabowski referred to as “synergy,” along with “trade name value and market positioning,” both of which did not exist in interstate mergers. Id. at 1116. As a result, Mr. Grabowski argued that, while the purchase premiums may be similar for both types of mergers, the premiums in each were a result of different assets—the intrastate mergers due to “synergies” and the interstate mergers due to Branching Rights—and that Dr. Mann’s analysis therefore did not undermine the Grabow-ski Model’s projections.

The district court, however, rejected this explanation. The court found that it did not adequately explain why the premiums were similar and that it was contradicted by other parts of Dr. Mann’s testimony, which indicated that acquiring premiums paid by both large and small thrifts were similar despite the fact that the identified “synergies” would not have been as beneficial to the bigger entities. The district court found Dr. Mann’s testimony undermined Mr. Grabowski’s testimony and the trustworthiness of his model.

And given the multiple weaknesses it identified in Appellant’s evidence, the district court ultimately concluded that Appellant failed to carry its burden of establishing either a cost basis in the Missouri Branching Right or its corresponding entitlement to the tax refunds it sought. The court then went on to consider Appellant’s abandonment claim, holding that Home had not abandoned the Missouri Branching Right in 1993 and was therefore not entitled to an abandonment loss deduction for that tax year.

The court properly determined that to establish abandonment, Appellant was required to show both that “Home intended to abandon the Missouri Branching Right” and that it “performed an overt act of abandonment.” Id. at 1117. Appellant offered evidence that, in the early 1990s, Home entered into three agreements to either sell or exchange the Missouri branches, and that each agreement contained a covenant not to compete, prohibiting Home from soliciting deposits for a limited period (two or three years) in the designated geographic area.

The district court, however, focused on the fact that the non-compete clauses contained exceptions. For example, one of the agreements allowed Home to purchase and operate branches within Missouri if Home merged with, purchased, or was purchased by another thrift already operating in the non-compete area. The other two agreements permitted Home to continue operating its existing branches, and one of those agreements also permitted Home to open new loan offices in the area.

Home’s Senior Vice President, Verne Kline, further testified that the non-compete clauses were intended to provide Home with flexibility “should an opportunity arise or if there [was] a change of a decision.” In fact, Mr. Kline indicated that one of his goals in negotiating the sale of the Missouri branches was to retain “the most flexibility that [Home] could get.” As Mr. Kline explained, Home was trying to anticipate and avoid future obstacles in the event “[Home] were to go out and acquire another national firm that perhaps had branches in these [non-compete] areas.” Despite the above exceptions and Mr.

Kline’s téstimony, Appellant argued that the totality of the evidence demonstrated that Home intended to abandon the Missouri Branching Right. Appellant pointed out that it had notified stock analysts, shareholders, and the Office of Thrift Supervision that Home was closing its Missouri branches, and that these facts evidenced that it had abandoned the Missouri Right and the Missouri thrift market. The district court disagreed. It noted that the express language of the covenants not to compete undermined Appellant’s argument, and Mr. Kline’s testimony cast further doubt on its position as well.

The court was not persuaded by the notices Home provided to interested parties, as that evidence failed to demonstrate that Home was permanently abandoning its right to re-enter the Missouri market. In fact, the court found that the evidence actually supported the conclusion that “Home recognized the value in leaving the Missouri Branching Right intact,” id. at 1119, as it would remain useful to Home’s nationwide thrift business, rather than just its Missouri operations and, thus, had value to Home’s potential future growth.

The district court concluded that Home had taken action to “safeguard,” as opposed to “permanently disavow,” its Branching Right and that no deduction was warranted. Id. at 1120. The district court then dismissed Appellant’s claims. D On appeal, Appellant contends that the district court improperly concluded that no cost basis could be assigned to the relevant intangible assets because: (1) it committed reversible legal error when it declined to estimate a cost basis for the Missouri Branching Right; (2) its error was due, at least in part, to its misunderstanding of the applicable burden of proof;

(3) Appellant offered sufficient evidence from which the court could have, under the proper standard, calculated the value of the Missouri Branching Right, even if it disagreed with Mr. Grabowski’s ultimate valuation; and (4) in any event, the district court’s disagreement was itself based on a clearly erroneous criticism of the inputs Mr. Gra-bowski used for his mid-case scenario.

Appellant also contends that the district court erred by denying the abandonment loss deduction for the Missouri Branching Right because: (1) it misapplied the test for determining whether an abandonment loss was warranted and improperly required Appellant to show that it “permanently discarded” the Missouri Branching Right, ignoring the fact that the abandonment test is disjunctive and permits a loss if a taxpayer simply discontinues its business; (2) it improperly held that the Missouri Branching Right was useful to Home’s nationwide thrift business as opposed to only its Missouri operations; and (3) it attached too much significance to the covenants not to compete.

II We apply de novo review to questions of law, such as the question whether the district court applied the correct burden of proof. See Husain v. Olympic Airways, 316 F.3d 829, 835 (9th Cir. 2002); Taisho Marine & Fire Ins. Co. v. M/V Sea-Land Endurance, 815 F.2d 1270, 1274 (9th Cir. 1987). Whether that burden of proof has been met, however, is reviewed for clear error. See Potts, Davis & Co. v. Comm’r, 431 F.2d 1222, 1227 (9th Cir. 1970). Whether the district court applied the correct legal standard to evaluate an abandonment loss claim is subject to de novo review. See A.J. Indus., Inc. v.

United States, 503 F.2d 660, 662 (9th Cir. 1974). “Where the trial has been by a judge without a jury, the judge’s findings must stand unless clearly erroneous.” Comm’r v. Duberstein, 363 U.S. 278, 291, 80 S.Ct. 1190, 4 L.Ed.2d 1218 (1960) (internal quotation marks and citation omitted). “Clear error review is deferential to the district court, requiring a definite and firm conviction that a mistake has been made.” Husain, 316 F.3d at 835 (internal quotation marks and citation omitted).

Ill A The district court permissibly concluded that Appellant did not meet its burden of establishing a cost basis for its intangible assets.

More specifically, the district court (1) held Appellant to the correct burden; (2) did not make any clearly erroneous factual findings; (3) permissibly determined that the cumulative fundamental flaws underlying the Grabowski Model rendered it incapable of producing a reliable value for the Missouri Branching Right; and (4) was thus not required to sua sponte assign a value to that Right. 1 Because this is a refund case, Appellant’s burden is “to prove not only that the Commissioner erred in his determination of tax liability but also to establish the correct amount of the refund due.” Fed-Mart Corp. v.

United States, 572 F.2d 235, 238 (9th Cir. 1978) (quoting Crosby v. United States, 496 F.2d 1384, 1390 (5th Cir. 1974)). “Thus, if insufficient evidence is adduced upon which to determine the amount of the refund due, the Commissioner’s determination of the amount of tax liability is regarded as correct.” Id. (quoting Crosby, 496 F.2d at 1390).

Appellant contends that, in concluding it had not met its burden, the district court improperly-held Appellant to an unwarranted standard and required Appellant to establish the market value of the Missouri Branching Right under a heightened level of “certainty” and “precision,” instead of simply requiring only a reasonable estimation of its value. We disagree. Contrary to Appellant’s argument, the district court did not require an unprecedented level of precision.

The court made clear that the burden was on Appellant to prove its refund entitlement and that the case law at times used the word “exact.” See, e.g., Compton v. United States, 334 F.2d 212, 216 (4th Cir. 1964) (“To put it another way, the ultimate question in a suit for refund is not whether the Government was wrong, but whether the plaintiff can establish that taxes were in fact overpaid. The plaintiff, to prevail, must establish the exact amount which she is entitled to recover.”).

But the district court tempered its use of that word by employing the phrase “reasonable certainty” to make clear that it was not holding Appellant to a more stringent standard of “exactness,” “certainty,” or “precision.” See, e.g., Trigon Ins. Co. v. United States, 234 F.Supp.2d 581, 586 n.7 (E.D. Va. 2002) (“The ‘exact amount’ text is not to be taken literally, but instead is subject to the general proviso that claims for refunds, like those for damages, must be supported by evidence proving the claim amount with reasonable specificity.”).

In fact, the district court agreed with Appellant that, assuming the existence of a sufficient evidentiary record, an estimate would suffice for valuation purposes. Accordingly, when the isolated phrases on which Appellant relies are taken in context, they are unremarkable and do not detract from the conclusion that the court understood the appropriate standard. Appellant’s argument based on the district court’s use of the phrase “heavy burden” similarly misses the mark. The district court used that phrase only while discussing Capital Blue Cross v.

Commissioner, 431 F.3d 117 (3d Cir. 2005), a case relied upon by Appellant for the proposition that where “the Government refuses to submit expert valuation testimony regarding particular adjustments to a proffered valuation, a court will essentially be forced to start from the taxpayer’s valuation.” While the district court did observe that the taxpayer bears a “heavy burden” to show that an intangible asset is capable of separate valuation, see Capital Blue Cross, 431 F.3d at 129-30, the district court also noted that this “heavy burden” was not a new burden being imposed on Appellant.

Rather, the district court was merely explaining the difficulties that may confront some taxpayers who are simply unable to show that their assets are capable of individual valuation. Indeed, Capital Blue Cross elaborated that “[o]nce it is established that the assets have a reasonably ascertainable value, the court is obligated to seek the correct value of the contracts not, upon catching the taxpayer in an error, to deny any deduction automatically.” Id. at 130.

Thus, the “heavy burden” referred to .by the district court here only went to the potential difficulty in proving that intangible assets are capable of separate valuation. In this context, the district court’s use of the phrase “heavy burden” did not alter the burden of proof. It merely explained that in some instances that burden is difficult to carry. 2 Having determined that the district court correctly understood and applied the appropriate standards, we turn to Appellant’s contentions that the district court’s criticisms of the Grabowski Model were clearly erroneous.

According to Appellant, the district court’s findings in this regard were flawed for two primary reasons: (1) the court’s approach to the valuation was overly pessimistic; and (2) the court overlooked the fact that Mr. Grabowski factored risk into his model by using a 22% discount rate and, therefore, improperly faulted Mr. Grabowski for failing to adequately address the risks inherent to the thrift industry in 1981.

Appellant further contends that the district court misunderstood a number of complex and technical issues that arose during trial, which resulted in the court’s clearly erroneous rejection of the Grabow-ski Model in its entirety.

More specifically, Appellant asks us to find that the district court: (1) erred by rejecting the Grabowski Model’s reliance on Home’s intrastate Northern California expansion; (2) misconstrued the Grabowski Model’s projected growth of the Missouri deposit market; (3) incorrectly held that “interest credited” should not be considered a source of “new” funds for lending; and (4) failed to recognize that its criticism of the Grabowski Model’s net interest spread was actually an attack on Home’s business judgment and the basic premise that a willing buyer would have thought that the projected spread was feasible.

However, because the district court’s findings were based on a permissible view of the evidence, none of these arguments are persuasive. As to the first argument, Appellant takes the position that the district court’s pessimistic perspective is incompatible with the perspective of a “willing buyer.” But Appellant points to no authority standing for the proposition that a willing buyer cannot also realistically assess a market in the manner undertaken by the district court. Moreover, the district court’s conclusion was amply supported by the record.

It is undisputed that Home engaged in the supervisory mergers against the backdrop of the savings and loan crisis. Despite this, Mr. Grabowski projected that Home would see healthy growth and that it would be able to convert all of its projected Missouri deposits into new loans. Not only did Mr. Grabowski’s projections seem to contravene the economic realities at the time, they were also contrary to Home’s own 1981 market projections. Additionally, the Government offered evidence that projecting high loan volumes in 1981 was ill advised given the high interest rates at the time.

Not only was it difficult for buyers to qualify for loans, specifically adjustable rate mortgages, but Home’s management was not sure whether such loans would have any widespread appeal. Given the profuse evidence supporting a more pessimistic market outlook, the district court was justified in rejecting Mr. Grabowski’s optimistic assumptions. Appellant further contends that the district court ignored the Grabowski Model’s use of a 22% discount rate and, therefore, incorrectly concluded that Mr.

Grabowski failed to adequately address risk in his Model and improperly required Appellant to account for the same risks twice. This argument is flawed, however, because it suggests that the 22% discount rate was intended to counter-balance the overly optimistic projections identified by the district court. In fact, Mr. Grabowski used the same discount rate for each of his scenarios, regardless of whether he used conservative or optimistic industry-wide assumptions for those scenarios.

The 22% risk premium was therefore directed at his view of the thrift industry as a whole and was not adjusted at any point to address the overly optimistic assumptions identified by the district court. The district court’s remaining factual findings were also not clearly erroneous. Appellant- claims that the district court improperly rejected the Grabowski Model’s reliance on Home’s previous intrastate expansion because the court erroneously concluded that disintermediation would have reduced both the Missouri deposit market as a whole and the hypothetical buyer’s proportional share of that shrinking market.

But while disintermediation may not directly affect a hypothetical buyer’s ability to capture deposit market share, that was not the only reason the district court rejected Appellant’s attempt to use Home’s Northern California expansion as a predictor of its anticipated results in Missouri. To the contrary, the court determined that predicting future success of an interstate merger at a time when the thrift industry was insolvent by reference to Home’s prior ability to capture intrastate market share in a growing market during a relatively stable period was unreliable.

The court also noted that this comparison was especially problematic because Home’s growth into Northern California was based on acquiring existing institutions, which it did not have in Missouri, and because Home’s own projections with regard to Missouri market share were much more conservative. These findings were not in error. The district court’s conclusion that Mr. Grabowski overstated potential statewide deposit growth was also permissible. As the district court noted, Mr.

Grabowski included “interest credited” as a source of “new deposit growth.” And while it is true that interest posted to the accounts of existing customers can still be used for lending, the district court did not err when it determined that “interest credited” did not constitute “new” growth and, therefore, should not have been included as such in the Model. Unlike true new deposit dollars that are essentially open for capture, “interest credited” dollars typically stay within their existing institution.

Thus, these dollars are not a categorical source of new funds flowing into the thrift market; rather, a hypothetical buyer would have to convince a depositor to leave its current establishment in favor of the newcomer. But to do so, the buyer would have to offer a higher rate of return to the depositor than the existing institution—a scenario that was highly unlikely during the savings and loan crisis. Accordingly, the district court correctly faulted the Model for including these funds and artificially skewing Missouri’s deposit market growth. . Last, the district court did not clearly err when it rejected Mr.

Grabowski’s projected 2.5% net interest spread, which he claimed a hypothetical buyer could expect to achieve in Missouri. Appellant takes the position that the district court should have deferred to that projection because it mirrored Home’s actual forecasts at the time of the underlying transactions. Nothing in the record or the case law, however, supports the conclusion that the court was bound by Home’s estimates. ' In sum, none of the district court’s challenged factual'findings were clearly erroneous, and we decline to reverse on this ground.

See Husain, 316 F.Sd at 835. 3 Finally, the district court did not dearly err when it determined that the cumulative flaws underlying the Grabowski Model rendered it incapable of producing a reliable value for the Branching Right. Appellant pleads to the contrary, arguing first that no such finding was ever made and, second, that any such finding was unsupported. Again, we disagree.

When read against the backdrop of the extensive record in this case, including counsels’ post-trial arguments, the district court’s decision makes clear that it accepted the Government’s primary contention that the Grabowski Model was “too flawed to form a reliable basis for valuing the Missouri Branching Right.” WAMU II, 996 F.Supp.2d at 1109. The trier of fact agreed when it ruled that the evidence was too unreliable to support even an estimation of the value of the Rights.

And although the court did not expressly state that no reliable evidence would permit an “informed estimation” of the intangible assets, that conclusion was implicit in its findings that Mr. Grabowski’s assumptions were “unreliable” and “unrealistic.” While it may have been helpful if the trial court had made a more exacting statement to that effect, it is clear to us that the court did in fact reach that conclusion.

Even more, the court made clear that it was familiar with the relevant authorities, including those standing for the proposition that, assuming sufficient evidence had been produced, an informed estimate was all that was required. The district court also went to great lengths to explain that it “[was] not required to, and indeed [could not], derive the cost basis from unreliable evidence.” WAMU II, 996 F.Supp.2d at 1103 (citing Norgaard v. Comm’r, 939 F.2d 874, 879 (9th Cir. 1991)); see also id. (“[Cjourts decline to apply Cohan[ v.

Commissioner, 39 F.2d 540 (2d Cir. 1930),] in cases where there is no doubt that the taxpayer incurred some deductible expense, but the taxpayer failed to present evidence sufficient to allow the court to make an accurate finding on the amount of the deduction.”). In sum, the court’s extensive discussion of the evidentiary prerequisites to making an informed estimate undermines the theory that the district court misunderstood the standard or applied it erroneously.

The very fact that the court did not engage in an estimation exercise—when it knew that an estimation should be made if sufficient evidence was presented—further supports the conclusion that the court had already determined that the evidence before it was inadequate. See Norgaard, 939 F.2d at 879-80. We see no error in the district court’s understanding of this evidentiary standard. We also agree with the district court’s application of that standard and its ultimate determination that estimation based on the evidence in the record was effectively impossible.

Contrary to Appellant’s arguments, the district court’s criticisms of Appellant’s evidence were not limited to inputs but were, in fact, much more fundamental. The district court did not just disagree with Mr. Grabowski’s quantitative inputs; it disagreed with the qualitative assumptions underlying the Model itself. While interest rates or market share inputs could potentially be modified in isolation, it is much less clear how the court could have unilaterally modified the Model to address Mr.

Grabowski’s reliance on: (1) an unrealistically “rosy” view of the future of the thrift industry in 1981 (e.g., by failing to properly consider disintermediation and relying on statistics that included “interest credited” balances); (2) an unrealistic estimate of Home’s ability to capture market share; (3) a flawed income-based approach to evaluating whether a hypothetical buyer in a loan-driven industry would be able to use new deposits to generate new loans; (4) an unrealistic hypothetical net interest spread; (5) an improbable growth rate during a time when associations were struggling to attract and retain depositors; and (6) a potential decrease in the Missouri Branching Right’s license value.

Given the above, Appellant’s argument that the court was required to sua sponte estimate some value for the Rights is foreclosed. On these facts, such a proposition would essentially do away with the taxpayer’s burden. Instead, the cases make clear that a district court may only be obligated to value an asset when it determines that the asset “may be valued separately” or has “a reasonably ascertainable value,” or “sufficient evidence was introduced to allow the [district] court to reach a reasonable conclusion” as to value. Capital Blue Cross, 431 F.3d at 129-30; R.M. Smith, Inc. v.

Comm’r, 591 F.2d 248, 251 (3d Cir. 1979). Since the district court permissibly determined that the Missouri Branching Right did not have a “reasonably ascertainable value,” it was not required to undertake such an exercise here. The district court applied the proper legal standards and did not clearly err in criticizing the Grabowski Model or in determining that the evidence was insufficient to reliably value the Missouri Branching Right. The court properly determined that Appellant failed to meet its burden and, thus, appropriately dismissed Appellant’s claims on this basis.

B Even assuming the Missouri Branching Right could be valued, Appellant nonetheless failed to show reversible error as to the denial of its abandonment deduction. Internal Revenue Code § 165(

a) provides for an abandonment deduction for “any loss sustained during the taxable year and not compensated for by insurance or otherwise.” 26 U.S.C. § 165(a). An abandonment loss may be appropriate if “[the] business or transaction is discontinued or where such property is permanently discarded from use therein.” Treas. Reg. § 1.165-2(a). A deduction is proper if there is “(1) an intention on the part of the owner to abandon the asset, and (2) an affirmative act of abandonment.” A.J. Indus., Inc. v. United States, 503 F.2d 660, 670 (9th Cir. 1974).

Appellant contends that the district court misapplied the test for determining whether an abandonment loss was warranted and, thus, committed legal error. First, Appellant argues that the district court improperly required Appellant to show that it had “permanently discarded” the Missouri Branching Right and ignored the disjunctive nature of the abandonment test, which also permits a loss if the taxpayer discontinues its business.

Second, Appellant contends that the court improperly defined the business in which the Missouri Branching Right was used as Home’s nationwide thrift business, rather than as a more limited Missouri-based operation. In addition, Appellant argues that the district court clearly erred in its factual analysis by attaching too much significance to the covenants not to compete that were included in the transfer agreements for the sale of the Missouri branches.

Despite the fact that the district court never explicitly stated in its written order that the disjunctive test requires inquiry into whether “such business ... is discontinued or ... such property is permanently discarded from use therein,” it nonetheless applied the correct legal standard. Treas. Reg. § 1.165-2(a) (emphasis added). The district court expressly acknowledged that, “[according to [Appellant], Home abandoned the Missouri Branching Right by closing its Missouri deposit-taking branches.” WAMU II, 996 F.Supp.2d at 1118.

More specifically, the district court identified Appellant’s argument to be that “Home [ ] abandoned] the [Missouri Branching] Right by abandoning the economically inefficient business in which it used the Right (ie., Missouri deposit taking), disposing of the assets it used in that business the Missouri branches), and ceasing to use the Right in that business.” Id. The district court then explained the Government’s competing position.

According to the district court, the Government claimed that the closing of the Missouri deposit-taking branches was insufficient to prove abandonment in light of (1) the covenants not to compete that reserved for Home the right to re-enter the market, and (2) the testimony of Home’s executives indicating that Home had not foreclosed the option of using the Missouri Branching Right in the future. Id. Based on both parties’ arguments and the evidence before it, the district court expressly determined that Home had not shown that it intended to discontinue the business in which the Branching Right .was used.

In doing so, the district court repeatedly addressed Home’s sale of the Missouri branches and its decision to withdraw from the Missouri market at the time. Appellant’s problem is not that the district court misunderstood the legal standard, but that the court disagreed with Appellant as to the factual importance of the limitations on the covenants not to compete and the contemporaneous evidence that Home intended to keep its future options open. Accordingly, the district court applied the correct standard. Appellant simply disputes the result.

To that end, Appellant primarily takes issue with the district court’s conclusion that Home’s relevant “business” be viewed at the national level. But contrary to Appellant’s arguments, the district court’s determination on that issue was permissible. The district court properly determined that the Missouri Branching Right remained potentially useful to Home’s ongoing national business to the extent Home could either decide to reenter the Missouri market or attempt to entice a merger with or a sale to another thrift with similar inclinations to expand outside of its home state.

The record clearly permits this finding and the court’s broad view of Home’s business was not clearly erroneous. We further conclude that the district court attached the proper significance to the covenants not to compete and to the testimony of Home’s executives that Home had not permanently given up on the idea of operating deposit institutions in Missouri.

The district court permissibly reasoned that the Missouri Branching Right retained value upon disposition of the existing branches because the Right—and the attendant ability to re-enter the Missouri market—held value beyond just the ability to purchase or operate branches. Stated another way, Home did not show that it intended to discontinue its business—-or abandon the Branching Right— when it expressly reserved the right to reenter the Missouri market.

Rather, it is clear that Home recognized that the Branching Right itself remained potentially valuable to its nationwide thrift business as an asset that might be attractive to a suitor in a merger or acquisition. The district court’s

interpretation of the evidence before it was appropriate, and Appellant has failed to point to any error warranting reversal. Costs are awarded to the Appellee. AFFIRMED. . The district court also concluded that, given this failure, it was not required to reach the parties’ remaining disputed issues (e.g., the extent of the failing thrifts' liabilities, assets, and the values for each). . Disintermediation refers to the withdrawal of deposit accounts from the thrift industry.

During the savings and loan crisis, this outflow resulted from the desire of depositors to get higher interest rates than thrifts were allowed to offer on accounts under then-existing regulations. . As part of his scenario analysis, Mr. Gra-bowski ran his Model using a variety of assumptions that were probability weighted. He ultimately picked the "mid-case scenario" for his final analysis and testimony in the district court. . To be clear, this is precisely what occurred in the district court.

The district court determined that Appellant's evidence was insufficient to support a valuation of the Missouri Branching Right and, as such, Appellant could not meet its burden to establish a cost basis in any of the rights. And because Appellant had the burden of proof in this action, Appellant consequently could not prove that it was entitled to a refund. Despite the fact that the result to Appellant was ultimately a zero recovery, the district court did not hold that Appellant had a cost basis of zero in the Rights—it simply found a failure of proof.

To the extent this seems draconian, we note that the United States is not entirely immune from this type of rule. In a deficiency case, for example, if a taxpayer proves that the Government’s deficiency notice is not only wrong, but also that it is arbitrary and lacking in foundation, then the IRS is likewise required to prove the amount of the deficiency. If the Government fails to meet its respective burden in such a case, the deficiency noticed is reversed in its entirety, judgment is entered for the taxpayer, and there is zero recovery by the IRS. See, e.g., Morrissey v.

Comm'r, 243 F.3d 1145 (9th Cir. 2001); Estate of Simplot v. Comm’r, 249 F.3d 1191 (9th Cir. 2001). . In fact, there was evidence in the record to support the conclusion that, contrary to Appellant's current arguments, Home had no interest in the Missouri market and it purchased those branches only to acquire the Florida Branching Right. To the extent Appellant is addressing the value of the Missouri Branching Right, this evidence cuts against Appellant’s argument that the district court took too pessimistic a view of the evidence. .

It appears the district court did err in citing the wrong line of an exhibit when it was discussing Mr. Grabowski's projections for Missouri deposit growth. Rather than citing to the numbers corresponding to statewide growth, the court mistakenly cited to a line that corresponded to a hypothetical buyer's anticipated growth in deposits within the state. Despite this essentially clerical error, however, the district court made clear that it properly understood the Grabowski Model to reflect that, "in real dollar terms, Missouri's deposit market did not actually grow.” WAMU II, 996 F.Supp.2d at 1107.

Further, it recognized that Appellant's "contention that the Grabowski Model actually shows nominal deposit growth if it is adjusted for inflation” was "only true for the Model’s statewide deposit projections” and not for a hypothetical buyer’s individual deposit projections. Id. at 1111. The court was thus clear that the higher percentages to which it referred were tied to individual deposit growth as opposed to statewide growth. . This conclusion is not undermined by the fact that the district court cited to the incorrect percentages in one part of its discussion. See supra note 6.

The district court’s qualitative challenges to projected statewide deposit growth are valid, and this minor error does not warrant reversal. . The substantial flaws identified in this case also render it distinguishable from Capital Blue Cross, 431 F.3d 117. The Capital Blue Cross case depends on the assumptions that (1) the taxpayer has shown that an asset has a reasonably ascertainable value and (2) the Government identified only minor flaws in the taxpayer's analysis. Id, at 130.

Here, however, the Government identified major, systemic flaws, and it did not concede that the value of the Rights may be determined separately and with reasonable precision. . We need not decide whether the ascertaina- ■ bility of an asset’s value is a question of law or of fact, thus affecting our standard of review. Compare Steen v. United States (In re Steen), 509 F.2d 1398, 1404 n.9 (9th Cir. 1975) ("While the amount of the fair market value of property is a question of fact, whether value is ascertainable is reviewable question of law.”), with Clodfelter v.

Comm’r, 426 F.2d 1391, 1395 (9th Cir. 1970) ("Whether property has ascertainable market value on a particular date and the amount of that value is a question of fact.”).

Under any standard of review, we agree with the district court that the evidence was insufficient to derive a value. .While we do believe it possible that some of the flaws identified by the district court could potentially have been addressed in reliance on other evidence in the record (e.g., the district court could perhaps have substituted a different net interest spread), the record does not reflect any obvious manner in which the court could have addressed, for example, Grabowski's overestimates regarding a hypothetical buyer’s ability to generate new loans or the value of the license.

These flaws go to defects in the Model itself, and it remains unclear to us how a court could address all of Model's flaws simultaneously and reach any principled result. . Nor was the district court constrained by the mandate from the first appeal to calculate a cost basis for each of the rights. That appeal did not address any arguments regarding the actual valuation itself. On remand, the court simply directed the district court to “proceed to determine the cost basis and conduct further proceedings in accordance to this opinion.” WAMUI, 636 F.3d at 1221.

The district court remained free to conduct the proceedings as it deemed necessary to independently calculate the value of the Rights. . Our conclusion is further supported by the fact that, as argued by Appellant, it was impossible for Home to "permanently discard" the Missouri Branching Right. Since the Right could not be discarded, the only relevant prong that the court could have considered was whether the business in which the Right applied had been discontinued. . To the extent Appellant’s argument turns on the district court's use of the word "permanently,” that argument is not well taken.

The district court may have used the phrases "permanently abandoned” and "permanently disavowed,” but it appears that use of the word "permanently” was not necessarily intended to refer to the issue of whether an asset was discarded. Rather, it seems “permanently” was used to differentiate between a "permanent” and "temporary” action. The district court essentially concluded that, because of the limitations on the covenants not to compete, Home's withdrawal from Missouri was, by definition, only "temporary” and, thus, did not support the finding that Home really intended to discontinue its business there. .

Appellant’s argument that the district court incorrectly applied a heightened burden of proof akin to a criminal standard to this issue is rejected. Passing use of the word “doubt” is insufficient to justify the inference that the district court confused the civil preponderance standard with the tougher criminal standard.