OPINION
MOORE, Circuit Judge.In 1988 and 1991, Plaintiff-Appellee The Dow Chemical Company (“Dow”) purchased corporate-owned life insurance (“COLI”) policies on the lives of thousands of its employees. In the taxable years 1989 to 1991, Dow claimed deductions for interest incurred on loans used to pay the COLI premiums- and for fees related to the administration of the policies. The Internal Revenue Service (“IRS”) disallowed these deductions and assessed tax deficiencies and interest, which Dow paid under protest and attempted to recover by suing for a refund. Following a bench trial, the district court ruled that the IRS had improperly disallowed the deductions and ordered judgment in Dow’s favor.
Because the COLI plans were economic shams and the deductions therefore were properly disallowed, we REVERSE the district court’s judgment and remand for entry of judgment in favor of the United States.
I. BACKGROUND
A. Factual Background
The district court conducted a bench trial lasting over two months, heard testimony from twenty-six witnesses, and received 1,526 exhibits and numerous factual stipulations. The facts of the case and the parties’ arguments are discussed at length in the district court’s two thorough opinions. Dow Chem. Co. v. United States (Dow I), 250 F.Supp.2d 748 (E.D.Mich. 2003); Dow Chem. Co. v. United States (Dow II), 278 F.Supp.2d 844 (E.D.Mich.2003). Because the case turns not on factual disputes but on issues of law, we present here only a brief factual background.
In 1988, Dow purchased COLI policies on the lives of 4,051 employees from Great West Life Assurance Company (“Great West”). In 1991, Dow purchased COLI policies on the lives of 17,061 employees from Metropolitan Life Insurance Company (“MetLife”). Dow, which was both the owner and the beneficiary of these policies, paid the premiums through two primary funding mechanisms.
First, Dow borrowed money from the insurers, using the cash values of the policies as collateral.1 These policy loans *597were the principal source of funding during the first, second, third, eighth, and ninth years of the Great West plan and during the first three years of the MetLife plan. For example, Dow paid (i) Great West’s first-year premium of $40,582,000 using $38,866,000 in proceeds from the policy loan and $1,717,000 of its own cash2 and (ii) MetLife’s first-year premium of $170,510,000 using $158,756,000 in proceeds from the policy loan and $11,754,000 of its own cash.
Second, Dow made partial withdrawals from the unencumbered cash values of the policies (i.e., the value not already used as collateral for a policy loan).3 Partial withdrawals were the principal source of funding during the fourth through seventh and tenth through thirteenth years of the Great West plan and during the fourth through eighth years of the MetLife plan. For example, Dow paid (i) Great West’s fourth-year premium of $40,360,000 (plus $11,278,000 in interest accrued on the policy loans) using $45,149,000 in proceeds from the partial withdrawals and $6,489,000 of its own cash and (ii) Met-Life’s fourth-year premium of $169,770,000 (plus $38,132,000 in interest) using $195,779,000 in proceeds from partial withdrawals and $12,123,000 of its own cash.4
All told, from 1988 to 2000, Dow paid $377,062,000 in premiums and $131,986,000 in interest to Great West from the following sources: $201,317,000 in policy loans, $238,844,000 in partial withdrawals, $9,203,000 in policy dividends, and $59,679,000 (about 16% of the premiums) in cash. From 1991 to 1998, Dow paid $849,890,000 in premiums and $239,371,000 in interest to MetLife from the following sources: $509,574,000 in policy loans, $495,844,000 in partial withdrawals, and $83,844,000 (about 10% of the premiums) in cash.
In addition to the payment mechanisms described above, both COLI plans shared several other pertinent characteristics. First, both plans were projected to generate negative pre-deduction (i.e., without taking into account the benefit of income-tax deductions) cash flows for many years — eighteen and seventeen years for the Great West and MetLife plans, respectively — before eventually generating positive cash flows contingent on the infusion of large amounts of cash by Dow. The net present value (“NPV”) of these cash flows was either positive or negative, depending on the discount rate used in the analysis. Second, neither plan was projected to experience significant inside build-up (i.e., the accrual of interest on the value of the policies) in the short term, because any such value would either be stripped from the policies (through the partial withdrawals) or encumbered (though the loans against the policies). If Dow injected *598large sums of cash into the policies, however, the plans would accrue significant interest in the long term. Third, both plans limited Dow’s potential mortality gain (i.e., the receipt of greater-than-expected death benefits due to a greater-than-expected number of deaths among the covered employees): Great West could recover its mortality losses through rate adjustments, while MetLife could assess charges to recoup its mortality losses.
In the taxable years 1989 to 1991, Dow claimed deductions totaling $33,004,360 for interest paid on loans used to pay the COLI premiums and for fees related to the administration of the policies.5 The IRS disallowed these deductions and assessed tax deficiencies and interest totaling $22,209,570.6 Dow paid this amount and filed administrative protests that the IRS denied.
B. Procedural Background
Dow filed suit in the district court to recover by refund the sum it had paid under protest, plus interest. The government argued that Dow’s deductions were improper because the COLI plans were economic shams. Following a bench trial, the district court ruled that the IRS had improperly disallowed the deductions because the COLI plans were not economic shams and consequently ordered judgment in Dow’s favor. The government appeals this conclusion and two subsidiary rulings. First, the district court credited the discount rate (and concomitant NPV) offered by Dow rather than the one offered by the government. Second, the district court ex-eluded projections of the COLI plans’ performance that Dow had submitted with its administrative protests because they were statements made in settlement negotiations under Rule 408 of the Federal Rules of Evidence.7
II. ANALYSIS
A. General Principles
“The legal right of a taxpayer to decrease the amount of what otherwise would be his taxes, or altogether avoid them, by means which the law permits, cannot be doubted.” Gregory v. Helver-ing, 293 U.S. 465, 469, 55 S.Ct. 266, 79 L.Ed. 596 (1935). In the instant ease, Dow attempted to decrease its tax liability principally through I.R.C. § 163(a) (26 U.S.C. § 163(a)), which permits a taxpayer to deduct “all interest paid or accrued within the taxable year on indebtedness.” For life insurance contracts, the deduction of interest on indebtedness is circumscribed:
No deduction shall be allowed for ... any amount paid or accrued on indebtedness incurred or continued to purchase or carry a life insurance ... contract ... pursuant to a plan of purchase which contemplates the systematic direct or indirect borrowing of part or all of the increases in the cash value of such contract ....
I.R.C. § 264(a)(3). This prohibition does not apply “if no part of 4 of the annual premiums due during the 7-year period (beginning with the date the first premium on the contract to which such plan relates *599was paid) is paid under such plan by means of indebtedness.” I.R.C. § 264(d)(1). Even under this safe-harbor, however, the Code provision in effect when the instant plans were purchased (1988 and 1991) limited the deduction to $50,000 of indebtedness per insured life. I.R.C. § 264(a)(4) (1988).
Of course, a taxpayer is not automatically entitled to every claimed deduction. “[A]n income tax deduction is a matter of legislative grace and ... the burden of clearly showing the right to the claimed deduction is on the taxpayer.” INDOPCO, Inc. v. Comm’r, 503 U.S. 79, 84, 112 S.Ct. 1039, 117 L.Ed.2d 226 (1992). Moreover, even if a transaction is in “formal compliance with Code provisions,” a deduction will be disallowed if the transaction is an economic sham. Am. Elec. Power Co. (AEP) v. United States, 326 F.3d 737, 741 (6th Cir.2003), cert, denied, 540 U.S. 1104, 124 S.Ct. 1043, 157 L.Ed.2d 888 (2004). “The proper standard in determining if a transaction is a sham is whether the transaction has any practicable economic effects other than the creation of income tax losses.” Rose v. Comm’r, 868 F.2d 851, 853 (6th Cir.1989). If the transaction has economic substance, “the question becomes whether the taxpayer was motivated by profit to participate in the transaction.” Illes v. Comm’r, 982 F.2d 163, 165 (6th Cir.1992), cert, denied, 507 U.S. 984, 113 S.Ct. 1579, 123 L.Ed.2d 147 (1993). “If, however, the court determines that the transaction is a sham, the entire transaction is disallowed for federal tax purposes, and the [subjective] inquiry is never made.” Id.; see also Rose, 868 F.2d at 853 (“This court will not inquire into whether a transaction’s primary objective was for the production of income or to make a profit, until it determines that the transaction is bona fide and not a sham.”).
B. Standard of Review
In conducting the economic-sham inquiry, the district court’s findings of fact are reviewed for clear error. AEP, 326 F.3d at 741 (citing Rink v. Comm’r, 47 F.3d 168, 172 (6th Cir.1995)). The district court’s ultimate conclusion that a transaction is or is not an economic sham is reviewed de novo.8 Id. at 742.
*600C. Economic Substance of Dow’s COLI Plans
Several courts, including this one, have addressed the economic substance of COLI plans, finding the plan to be a sham on each occasion. Id. at 744; IRS v. CM Holdings, Inc. (In re CM Holdings, Inc.), 301 F.3d 96, 102-03 (3d Cir.2002); Winn-Dixie Stores, Inc. v. Comm’r (Winn-Dixie), 254 F.3d 1313, 1317 (11th Cir.2001), cert, denied, 535 U.S. 986, 122 S.Ct. 1537, 152 L.Ed.2d 464 (2002). These courts have looked to several indicators of the COLI plans’ potential economic benefits: (i) projected pre-deduction cash flows, “ ‘[ (ii) ] mortality gains to the beneficiary, who does not pay tax on proceeds, and [ (iii) ] interest-free inside build-up [the accrual of interest on the policy value].’ ” AEP, 326 F.3d at 742 (last alteration in original) (quoting CM Holdings, 301 F.3d at 103). We examine the objective economic substance of Dow’s COLI plans against each of these standards below.
1. Pre-Deduction Cash Flows
In each prior case addressing the economic substance of a COLI plan, the court found that, without the benefit of the claimed interest deductions, the plan generated negative cash flows. AEP, 326 F.3d at 742; CM Holdings, 301 F.3d at 103; Winn-Dixie, 254 F.3d at 1316. Indeed, we have called a COLI plan’s negative pre-deduction cash flows (that become positive when the benefit of interest deductions is considered) a “hallmark[] of an economic sham.” AEP, 326 F.3d at 742.
The district court found that the Great West and MetLife plans would generate negative pre-deduction cash flows during their first eighteen and seventeen years, respectively. Unlike the COLI plans in AEP, CM Holdings, and Winn-Dixie, however, Dow’s plans were not projected to be cash-flow negative for their entire durations. The district court found that Dow intended to infuse large sums of cash into the plans during their middle years,9 which would make the plans cash-flow positive in their later years. Because the plans were projected to have both cash-flow-negative and cash-flow-positive years, the district court applied a NPV analysis to determine the overall cash flow.10 Finally, the district court credited the discount rate utilized by Dow’s experts and found that the NPV of each plan was positive.
The government objects to the district court’s finding that Dow intended to inject *601large amounts of cash into the plans in their middle years. Reversing this finding would have a domino effect favorable to the government: it would undermine the finding that the plans would generate positive cash flows in their later years, which would in turn preclude the finding that each plan had a positive NPV. Review of this factual finding for clear error is only necessary, however, if such highly-contingent cash flows are relevant as a matter of law to the economic-sham analysis. Unfortunately, the prior COLI-plan decisions are unhelpful in resolving this legal question because in each case the court found the pre-deduction cash flow to be negative for every year of the plan and there was no possibility of future profitability upon the materialization of some contingency.11
The Supreme Court has, however, provided useful guidance. In Knetsch v. United States, 364 U.S. 361, 81 S.Ct. 132, 5 L.Ed.2d 128 (1960), the Court reviewed a transaction involving the taxpayer’s purchase of $4 million worth of 30-year bonds and his simultaneous borrowing against them, which generated hundreds of thousands of dollars in interest deductions in just two years. Id. at 362-63, 81 S.Ct. 132. The taxpayer terminated the transaction after three years: he surrendered the bonds (now worth $4,308,000), his loan (now worth $4,307,000) was cancelled, and he received the difference ($1,000). Id. at 364, 81 S.Ct. 132. The Court upheld the IRS’s disallowance of the taxpayer’s loan-interest deductions as the product of a sham transaction. Id. at 366, 81 S.Ct. 132.
Two aspects of the Court’s opinion shed light on the current inquiry. First, the Court observed that if, instead of terminating the transaction, the taxpayer had held the bonds to maturity and continued to operate as he had done for the first three years, he would have received an annuity worth $1,000. Id. at 364, 81 S.Ct. 132. Second, the Court summarily rejected the taxpayer’s argument that the transaction would have become profitable in ten years if he paid off the original $4 million loan. Id. at 366 n. 3, 81 S.Ct. 132. The first statement demonstrates that potential future profitability can be relevant in assessing whether an ongoing transaction is an economic sham. The contrast between the first and second statements reflects a more nuanced lesson. Courts may consider future profits contingent on some future taxpayer action, but only when that action is consistent with the taxpayer’s actual past conduct. Courts should be skeptical, however, when the asserted future profits hinge on future taxpayer action that seriously departs from past conduct, especially where such departure involves the expenditure of large sums of money.12
These principles have obvious relevance to the instant case. It was proper for the *602district court to consider whether Dow’s plans would be profitable in the future; indeed, in each prior COLI-plan case, the court looked to future cash flows. The same cannot be said, however, for the district court’s consideration of positive cash flows that were contingent on Dow’s eventual spending of significant amounts of cash. Like the transaction in Knetsch, the instant COLI plans would become profitable only upon the taxpayer’s large future outlay of additional cash, and, considering that Dow had heretofore made no similar cash infusions, such additional spending would be a drastic departure from the taxpayer’s past conduct.13 Moreover, there was no contractual provision requiring Dow to make substantial cash infusions in the future. In light of the teachings of Knetsch, the district court erred in including in its cash-flow analysis the highly-contingent positive cash flows projected for later years.14 When the future infusion of cash is properly removed from the analysis, only negative cash flows remain. Therefore, without the benefit of the interest deductions, the COLI plans were cash-flow negative for all relevant periods, which is a “hallmark[] of an economic sham.” AEP, 326 F.3d at 742.
Our holding that the future positive cash flows should have been ignored makes it unnecessary to reach the government’s appeal of the district court’s (i) choice of discount rate to credit in the NPV analysis *603and (ii) exclusion of Dow’s tax protest letters under Rule 408 of the Federal Rules of Evidence. First, the removal of the highly-contingent positive cash flows from the analysis leaves only negative cash flows, which of course yield a negative NPV no matter what the discount rate. Second, the government offered the tax protest letters to cast doubt on Dow’s intent to inject large amounts of cash into the plans in the future, which is irrelevant in light of our holding that this putative additional outlay (or, more precisely, the profits contingent on such spending) should be disregarded as a matter of law.
2. Inside Build-Up
In each prior COLI-plan case, the court found that the plan ensured that at regular intervals net equity was adjusted to zero or some small number. AEP, 326 F.3d at 740, 742; CM Holdings, 301 F.3d at 103; Winn-Dixie Stores, Inc. v. Comm’r, 113 T.C. 254, 281, 284, 1999 WL 907566 (1999). This low or zero net equity precluded the taxpayer from realizing the benefit of inside build-up, because there would be little or no net equity to build up (i.e., earn interest).
The district court found that “on the last day of each of policy years four through seven, the cash value of Dow’s Great West and MetLife policies was fully encumbered, yielding virtually zero net equity.” Dow I, 250 F.Supp.2d at 814. The district court also noted that Gary Lake, an actuarial consultant who advised Dow on the COLI plans, id. at 766, “stated that for some of the years the purpose was to have low net equity,” id. at 774. Based on its finding that in the future Dow would infuse the plans with large sums of cash, however, the district court held that the company “stood to realize substantial economic gain from tax-free inside build-up” with respect to both plans. Id. at 807; see also id. at 810.
For the same reasons discussed above in the context of cash flows, the district court erred in considering in its inside build-up analysis the prospect of large cash investments that might be (but did not have to be) made in the future. Instead, given the district court’s factual findings with respect to Dow’s actual conduct of maintaining little or no net equity, a simple application of AEP, CM Holdings, and Winn-Dixie yields the result that Dow would have been unable to realize the benefit of inside build-up. This conclusion further suggests that the COLI plans were an economic sham.
3. Mortality Gains
The district court found that the plans contained features designed to neutralize the taxpayer’s ability to realize mortality gains. The-district court nevertheless found that the COLI plans were not mortality neutral because neither one “containfed] a 100% retrospective adjustment mechanism,” a requirement that the court divined from the three prior COLI-plan decisions.15 Dow I, 250 F.Supp.2d at *604811; see also id. at 808, 810. The district court reiterated its application of this 100%-adjustment requirement when it explained in its amended opinion that “a plan that is not designed to eliminate every dime of mortality profit ... cannot be ‘mortality neutral’ over the long term.” Dow II, 278 F.Supp. at 851 (emphasis added).
In the three prior COLI-plan cases, the mortality provisions were deemed neutral, making the potential for mortality gains a sham. Yet two of the three previously challenged plans clearly did not eliminate 100% of all mortality gains and losses. In CM Holdings, the taxpayer received a total mortality gain of $1.3 million over the first eight years of the plan. 301 F.3d at 101. The Third Circuit noted that the insurance company “assessed [the taxpayer] surcharges ... to recoup its losses and ensure mortality neutrality going forward.” Id. at 103. The CM Holdings district court’s description of the plan reveals, however, that these surcharges did not constitute a 100% elimination of mortality gains. The plan contained a provision capping maximum cost-of-insurance (“COI”) charges, which led the district court to conclude that the insurance company “may never recoup [the taxpayer’s] mortality windfall through increased COI charges.” In re CM Holdings, 254 B.R. 578, 635 & nn. 70-71 (D.Del.2000).
The precise details of the mortality-neutralizing provision in AEP are less clear than in CM Holdings but still suggestive of less-than-complete elimination of mortality gains. The taxpayer was advised that the plan’s mortality mechanism would “result in a much closer match” between expected and actual death benefits.16 Am. Elec. Power, Inc. v. United States, 136 F.Supp.2d 762, 777 (S.D.Ohio 2001) (emphasis added). The use of the phrase “much closer” rather than a word like “complete,” “total,” or “absolute” implies that the plan did not provide for 100% adjustment. Indeed, over a nine-year period, the taxpayer experienced a mortality loss (i.e., there were fewer deaths than expected, so it received less in death benefits than it paid for) of $38.8 million between projected and actual death benefits that apparently went unadjusted. Id. at 788. There is nothing in the opinion to indicate that a mortality gain would have been completely eliminated even though a mortality loss was left uncorrected. Further supporting this conclusion is the fact that the plan in AEP was “the same COLI plan ... offered by the same insurer and sold by the same brokers” as the one in CM Holdings, id. at 768-69, which (as has already been shown) limited the insurer’s ability to recoup the taxpayer’s mortality gains.
Finally, the precise details of the mortality-neutralizing provision in Winn-Dixie are unclear. The plan provided for the creation of a claims-stabilization reserve (“CSR”) that was funded by COI charges and from which death claims were paid. Winn-Dixie, 113 T.C. at 268. If the taxpayer experienced mortality gains in a given year, then the excess was deducted from the CSR. Id. at 268-69. “[T]he function of the claims stabilization reserve was to ameliorate fluctuations in actual mortality experience.” Id. at 285. The Winn-*605Dixie district court’s opinion does not elaborate, however, what would happen if the taxpayer’s mortality gains exhausted the CSR. While one might imagine the insurer recouping the gains through some other mechanism, one might also imagine the taxpayer keeping any mortality gains over and above the value of the CSR. In light of this uncertainty, the district court below could not have concluded that the Winn-Dixie plan provided for the complete elimination of mortality gains.
This review of the cases demonstrates that a “100% retrospective adjustment mechanism” requirement simply cannot be discerned from the past COLI-plan cases. In fact, a rule that permitted a COLI plan to be deemed mortality neutral only upon proof that “every dime of mortality profit” is eliminated would outright conflict with the facts of two of the three cases. Therefore, the district court erred by imposing such a high hurdle as a prerequisite to finding that Dow’s plans were designed to neutralize mortality gains.
When this erroneously high bar is removed, it is clear that the COLI plans were designed to reduce (even if not by 100%) Dow’s ability to realize mortality gains. The Great West plan permitted the insurer to recoup its mortality losses (i.e., Dow’s mortality gains) through rate adjustments. Dow I, 250 F.Supp.2d at 808. The MetLife plan gave the insurer “the right to increase COI charges to recoup losses if claims for death benefits exceeded COI charges.” Id. at 810. The fact that such increase was limited by a stop-loss provision, id., is no bar to a finding of mortality neutrality, because (as discussed above) the recoupment mechanism in CM Holdings also was capped. It is apparent, then, that these features are sufficiently similar to the other COLI-plan cases for us to conclude that Dow would not significantly benefit from mortality gains. This conclusion provides further evidence that the COLI plans were an economic sham.
4. Summary
Courts have recognized three non-tax benefits relevant to the economic-substance analysis of COLI plans: positive cash flows, inside build-up, and mortality gains. Dow’s COLI plans did not generate any of these benefits. Because the plans did not have “any practicable economic effects other than the creation of income tax losses,” Rose, 868 F.2d at 853, we conclude that they had no economic substance. This conclusion makes it unnecessary to discuss Dow’s subjective motivation. Ules, 982 F.2d at 165.
III. CONCLUSION
For the reasons set forth above, we REVERSE the district court’s judgment and remand for entry of judgment in favor of the United States.
. The policy-loan transactions were structured as follows: "Dow would receive a bill from the insurance company that netted the premium and interest charges against the proceeds of a loan that was made on the first day of the policy year, leaving a relatively small balance to pay in cash. The premium payment created value in the policy, which was used as security for repayment of the loan." Dow I, 250 F.Supp.2d at 812.
. Where, as here, the component figures to be summed do not add to the stated total, it is due to rounding.
. These partial-withdrawal transactions were structured as follows: "(1) the gross premium was deemed paid; (2) the deemed payment of the premium created cash value; (3) Dow made a partial withdrawal of the cash value; and (4) the partial withdrawal was used to offset approximately 90% of the premium and accrued loan interest.” Dow I, 250 F.Supp.2d at 814. Under each plan these steps were carried out effectively simultaneously.
.During the tenth through thirteenth years of the Great West plan, dividends from the policies were another source of funding. For example, Dow paid the tenth year’s premium of $3,702,000 (plus $14,442,000 in interest accrued on the policy loans) using $12,343,000 in proceeds from the partial withdrawals, $3,685,000 in policy dividends, and $2,116,000 in its own cash. Dow did not use policy dividends to make any payments to MetLife.
. In taxable years 1989, 1990, and 1991, Dow claimed deductions for (i) interest expenses of $3,843,813, $12,968,778, and $13,491,825, respectively, and (ii) administrative fees of $168,563, $2,175,160, and $356,221, respectively.
. For taxable years 1989, 1990, and 1991, the IRS assessed (i) tax deficiencies in the amount of $1,367,386, $3,043,326, and $6,836,910, respectively, and (ii) interest in the amount of $1,781,778, $3,235,630, and $5,944,540, respectively.
.Other aspects of the district court's opinion were not appealed and therefore are not discussed here.
. A panel of this court once noted in an economic-sham case that "[w]here the judgment below is ultimately a finding of fact, it is well-settled that the determination of the Tax Court is binding on the appellate court unless clearly erroneous.” Ratliff v. Comm’r, 865 F.2d 97, 98 (6th Cir.1989) (alteration in original) (internal quotation marks omitted). Two decisions issued mere months after Ratliff clarified that while the factual findings underlying a sham determination are reviewed for clear error, the legal standards employed and the ultimate conclusion are reviewed de novo. Kennedy v. Comm’r, 876 F.2d 1251, 1254 (6th Cir.1989); Rose, 868 F.2d at 853. Since this trio of decisions in 1989, we have consistently followed the more nuanced formulation of Kennedy and Rose whenever we have explicitly stated the standard of review. AEP, 326 F.3d at 741-42; Rink, 47 F.3d at 171; Smith v. Comm’r, 937 F.2d 1089, 1096 (6th Cir. 1991); Bryant v. Comm’r, 928 F.2d 745, 748 (6th Cir.1991). Although there may appear to be some tension between Ratliff and our subsequent cases, we recognized in AEP that the Ratliff panel "did not specifically hold that the ultimate question of whether a transaction is a sham is to be reviewed under the clearly erroneous standard.” AEP, 326 F.3d at 742. Furthermore, employing different standards of review for the factual and ultimate questions is more consistent with the Supreme Court’s teaching that "[t]he general characterization of a transaction for tax purposes is a question of law subject to review. The particular facts from which the characterization is to be made are not so subject.” Frank Lyon Co. v. United States, 435 U.S. 561, 581 n. 16, 98 S.Ct. 1291, 55 L.Ed.2d 550 (1978). Both parties agree that although factual findings are reviewed for clear error, the ultimate question of whether the transaction is an economic sham is reviewed de novo. Appellant's Br. at 33; Appellee's Br. at 38.
. The district court did not specify how large this infusion would have had to be in order to make the COLI plans profitable. Based on the following evidence and findings, however, we can infer that the figure was quite large indeed — about $315 million.
With respect to the Great West plan, Dow asserts that it relied on two illustrations (Cash Flow # 1 and # 2) in making its purchase decision. Appellee’s Br. at 12. The district court credited "Dow’s intention to cap loans at $50,000 and withdraw only to basis,” Dow I, 250 F.Supp.2d at 802, which reflects only Cash Flow # 1, Appellee's Br. at 9. This illustration shows cash infusions of about $30 million during the middle years. Joint Appendix ("J.A.”) at 1711.
With respect to the MetLife plan, the district court credited Dow’s assertion that it relied on an illustration called Case #23. Dow I, 250 F.Supp.2d at 809; Appellee’s Br. at 16. This illustration shows cash infusions of about $285 million during the middle years (after multiplying the illustration's per-insured infusion by the 17,800 employees that Dow projected to be covered by the plan, Dow I, 250 F.Supp.2d at 784-85; Appellee's Br. at 15). J.A. at 898.
. "In transactions that are designed to yield deferred rather than immediate returns, present value adjustments are, as the courts have recognized, an appropriate means of assessing the transaction’s actual and anticipated economic effects.” ACM P’ship v. Comm’r, 157 F.3d 231, 259 (3d Cir.1998) (citing cases).
. Beyond the COLI-plan context, courts have disfavored the deduction of interest paid on investments with contingent benefits. See, e.g., Coleman v. Comm’r, 87 T.C. 178, 208-09, 1986 WL 22162 (1986), aff'd, 833 F.2d 303 (3d Cir.1987); Rice’s Toyota World, Inc. v. Comm’r, 752 F.2d 89, 92-93 (4th Cir.1985); Smoot v. Comm'r, 61 T.C.M. (CCH) 2897, 1991 WL 97650 (1991); Goldwasser v. Comm’r, 56 T.C.M. (CCH) 606, 1988 WL 118617 (1988); Zegeer v. Comm’r, 54 T.C.M. (CCH) 1203, 1987 WL 49205 (1987).
. The dissent objects to our reading of Knetsch, asserting that the Court simply "determined that the taxpayer did not intend to make the greater future investment by paying off the loan, and therefore, the potential future cash flows were not relevant to the economic substance analysis.” Dissent at 605. Our reading of Knetsch is consistent with this statement. What the dissent ignores (and we recognize) is the reason behind the Court's determination: the large future investment would be altogether inconsistent with his past conduct. This is what made paying off the loan "wholly unlikely.” Knetsch, 364 U.S. at 366 n. 3, 81 S.Ct. 132.
. As the facts recited in the text demonstrate, the policy premiums had been funded principally through loans against and withdrawals from the policies, with Dow using little of its own cash. Yet the $315 million in projected future investments purportedly would have been coupled with Dow limiting its borrowing and withdrawals. Dow I, 250 F.Supp.2d at 801-02, 807, 809-10. On these facts, there can be no doubt that the projected cash infusions would have required Dow to depart drastically from its past conduct. Thus, we leave for another day the consequences of less drastic departures. This obviates the dissent's concern that our holding prohibits the consideration of contingent future profits whenever it depends on future investment that is merely "greater than the past investment." Dissent at 605.
Also, without citing any authority, the dissent argues that we should also consider "whether the projected future investment is feasible and there is evidence that it is likely to occur.” Dissent at 605. While we agree that the in feasibility of a future investment is indicative of a sham transaction, the fact that an investment is feasible is less likely to provide useful information because large future investments will almost always be feasible for large corporations or other resource-rich taxpayers. A factor that is too easily met is no factor at all.
A likelihood factor presents problems of its own. How would one determine likelihood, i.e., what types of evidence would one consider? (The dissent does not say.) One possibility is the profitability of the transaction, because a taxpayer is more likely to make a future investment if it will make the transaction profitable. But this would be circular, requiring us to look to the profitability of the transaction to determine the likelihood of the investment, which we would then use to determine the economic substance (i.e., profitability) of the transaction all over again. Another possibility is the taxpayer's plans, because internal deliberations might show that a future investment is likely to occur. But that crosses over to the subjective inquiry that we do not conduct until the transaction is determined to have economic substance in the first place. liles, 982 F.2d at 165. In light of these difficulties, the better course is to remain true to Knetsch by limiting the inquiry to whether the future investment drastically departs from past conduct.
. We do not hold that future profits contingent on new expenditures by the taxpayer may never be considered. For example (as we suggest in the text), if a taxpayer were contractually obligated to infuse additional cash into an investment at some point in the future, such spending would not constitute a serious departure from past conduct that should be disregarded under Knetsch. Instead, the eventual outlay would be consistent with actual past conduct, i.e., the obligation that existed all along. Of course, Dow was under no such obligation in the instant case.
. The district court read the three cases as follows:
[Cjourts found that the COLI plans were flawed because of devices designed into the plans that eliminated the transfer of risk. Those courts thus adopted the government’s terminology and found that the plans were "mortality neutral” when, after the relevant period, usually a year, the cost of insurance (COI) and death benefits were "trued up” retrospectively. That is, when there is an agreement contained in the plan to make payments after the conclusion of a policy year so that the COI equals the amount paid out in death benefits, risk is eliminated and, as insurance, the COLI transaction is economically meaningless.
Dow I, 250 F.Supp.2d at 807-08 (citing Winn-Dixie, 113 T.C. at 268-69, 285; IRS v. CM Holdings, Inc. (In re CM Holdings, Inc.), 254 B.R. 578, 632-35 (D.Del.2000); Am. Elec. *604Power Co. v. United States, 136 F.Supp.2d 762, 777, 787-88 (S.D.Ohio 2001)).
. Contrary to the dissent’s assertion, Dissent at 609, the next sentence of the advice given the taxpayer — "This will eliminate any volatility or variation in the cash flow that we are expecting in each year of the plan” — does not undermine the sentence quoted in the text. Indeed, the use of the qualifier "that we are expecting” to modify "volatility or variation” suggests that the taxpayer’s mortality gains would not be completely eliminated if they exceeded expectations.