dissenting.
My colleagues conclude that, as a matter of law, future profits contingent on taxpayer action are an appropriate component of the economic substance calculus only when that action comports with the taxpayer’s actual past conduct related to the transaction in question. I disagree. In my opinion, there is no such precedential rule of law and no warrant for creating one in this case. The validity of Dow’s COLI plans as investments having economic substance turns on the district court’s findings of fact and the sufficiency of evidence supporting those findings. I would affirm the judgment in favor of Dow because I believe that the district court correctly applied the law of this circuit to factual findings that are not clearly erroneous.
I have no disagreement with the majority opinion’s statement of the complex factual details of Dow’s COLI program, and *606its summary of the general principles of this court’s economic substance doctrine. My disagreement is with the legal principles my colleagues invent and mistakenly attribute to Knetsch v. United States, 364 U.S. 361, 81 S.Ct. 132, 5 L.Ed.2d 128 (1960).
I agree with my colleagues that Knetsch demonstrates that potential future profitability is relevant to the inquiry of whether a long-term investment plan is an economic “sham,” as that unfortunate term has been used by the IRS and some of our federal courts. But my colleagues read into Knetsch far more than the Supreme Court wrote in that case concerning the Court’s refusal to accept the taxpayer’s argument that the transaction would have become profitable in ten years had he paid off the original $4 million loan.
The Supreme Court did not summarily reject the taxpayer’s argument; rather, it stated that the argument was “predicated on the wholly unlikely assumption that Knetsch would have paid off in cash the original $4,000,000 ‘loan.’ ” Id. at 366 n. 3, 81 S.Ct. 132. My colleagues insist that this language implies a general principle of law that future profits are not even relevant to the economic substance inquiry when the taxpayer’s projected future investment in a particular plan is greater than its past investment in that plan, regardless whether the projected future investment is feasible and there is evidence that it is likely to occur. I read Knetsch to indicate only that the Court made a credibility assessment and 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. The Court did not hold that, as a matter of law, a feasible projected future investment of cash in a particular plan is irrelevant to the economic substance inquiry, when that investment is greater than the past investment in that plan. The question is what the taxpayer intended.
If Dow’s potential future cash flows are not arbitrarily excluded from the economic substance inquiry, this case turns on the district court’s findings of fact regarding the intended operation of Dow’s COLI plans. The district court made two crucial findings of fact: (1) Dow intended to operate its plans such that it would realize the benefits of tax-deferred inside buildup; and (2) Dow’s plans transferred mortality risk to the insurers on the aggregate. The court then correctly applied the legal principles set down in the so-called sham-COLI cases, and concluded that, unlike the sham COLI plans, Dow’s plans had economic substance as a long-term investment program.
The sham-COLI cases explain that the two main benefits of life insurance plans other than the potential for tax deductions are: (1) mortality payments to the beneficiary, which are not taxable; and (2) the accrual of tax-deferred interest on policy value (inside buildup), that is, the increase in the policy’s cash value attributable to interest income earned upon the investment of the cash value. The plans in the sham-COLI cases were economic shams because they failed to realize either of the two benefits of a life insurance plan and because they were created solely for the tax benefit of deductions. See Am. Elec. Power Co. v. United States, 326 F.3d 737 (6th Cir.2003) (AEP); In re CM Holdings, Inc., 301 F.3d 96 (3d Cir.2002); Winn-Dixie Stores, Inc. v. Comm’r, 254 F.3d 1313 (11th Cir.2001).
In this case, the district court concluded, on the basis of its two central factual findings, that, unlike the sham-COLI plans, Dow’s COLI plans realized both of the non-deduction benefits of life insurance policies and therefore had economic sub*607stance. Unless both of those factual findings by the district court are clearly erroneous — and I conclude they are not — the court’s judgment for Dow should be affirmed.
I. Inside Buildup
The first basis for the district court’s determination that Dow’s COLI plans had economic substance was its finding that Dow would realize the benefits of the tax-deferred inside buildup of the policies. There were no such benefits in the COLI plans found to be shams in AEP, CM Holdings, and Winnr-Dixie. Dow’s pre-purchase projections showed that the plans generated positive cash flows, even in the absence of the interest deductions. The district court noted that although the pre-deduction cash flows were negative for the first 18 years of the plan, positive cash flows, delayed for 18 years, were consistent with Dow’s subjective business purpose of funding retiree benefit expenses in the long term. Dow Chem. Co. v. United States, 250 F.Supp.2d 748, 806 (E.D.Mich. 2003).
I think the district court got it right, in first analyzing each of Dow’s COLI plans as a whole and then correctly concluding that the plans had economic substance, because, among other things, they realized the benefits of inside buildup. That conclusion rested, first, upon the court’s factual finding that Dow intended from the outset to cap policy loans at $50,000, to withdraw funds only to basis, and to invest cash into the Great West and MetLife plans after 17 years. See id. at 802, 809. The district court based this finding not only on the testimony of Dow’s witnesses, but also on contemporaneous documentation introduced by Dow, including Dow’s original 1987 Request for Proposals (RFP) to the insurance industry; analysis of proposals by Dow’s actuarial consultant and his final recommendation; and pre-pur-chase cash flow illustrations. Id. In addition, the district court noted that a strategy of capping policy loans at $50,000 and withdrawing only to basis was more consistent with Dow’s purpose of producing cash flows to fund future retiree benefit expenses than a minimum-payment strategy because the capped-loan strategy produced higher positive cash flows after 18 years. Id. at 801-02. The capped-loan strategy was also more consistent with Dow’s overall goal of tax avoidance than the minimum payment strategy because interest on loans above $50,000 was not tax deductible, and withdrawing cash above basis would have exposed Dow to taxes that could have been avoided if that cash had been paid out as death benefits. Id.
A. Net Present Value (NPV) Analysis
The district court’s conclusion that Dow’s COLI plans had economic substance because they generated inside buildup was based, in part, upon the court’s acceptance of the net present value (NPV) analysis of Dow’s financial expert. That was a finding of fact, and I am satisfied the court did not commit legal error in accepting Dow’s, rather than the government’s, NPV analysis.
Courts have recognized that NPV analysis is an appropriate method to determine the economic effects of transactions designed to yield deferred returns. ACM P’ship v. Comm’r, 157 F.3d 231, 259 (3d Cir.1998). NPV analysis consists of discounting future earnings by a factor intended to reflect the time value of money. See Dow, 250 F.Supp.2d at 806. The discount rate used is the rate of return on an alternate investment.
The district court accepted the testimony of Dow’s expert, Stewart Myers, who stated that the appropriate discount rate for the NPV analysis of Dow’s delayed cash flows is an “after-tax” Moody’s Corporate Average, which is calculated by *608multiplying Moody’s Corporate Average by a factor of one minus Dow’s tax rate. This adjusted discount rate reflects the reality that the death benefits from Dow’s COLI plans were not taxable, whereas Dow would have to pay tax on the income from alternate investments. Dow’s actual return from alternate investments would be reduced by its income tax rate, but its return from the COLI plans would not.
The government argues that, as a matter of law, the proper discount rate is an unadjusted Moody’s Corporate Average because using a discount rate that reflects the tax-free nature of death benefits conflicts with the “pre-tax” analysis required by this circuit and the Third Circuit, and mistakenly puts Dow in a “world without taxes.” See AEP, 326 F.3d at 743-44; CM Holdings, 301 F.3d at 95.
I do not agree with the government’s argument that the “pre-tax” analysis required in this circuit under AEP requires an unadjusted discount rate. As the Third Circuit explained in CM Holdings, this “pre-tax” analysis requires only that there be removed from consideration the challenged deduction and the transaction be evaluated on its merits. CM Holdings, 301 F.3d at 105. It is not necessary, as the government suggests, that we must imagine a “world without taxes,” and we may take into account the reality that some investments are tax favored.
Dow’s expert separated Dow’s COLI transactions into their investment and financing components or “legs” and analyzed the cash flows from each leg separately. The “investment leg” cash flows showed the benefits from inside buildup, whereas the “financing leg” cash flows showed the value added from financing — in this case, the interest deductions. The government does not claim that separating the investment and financing legs, as Dow’s expert did, is not standard practice in financial analysis.
Dow’s expert concluded that the investment leg of Dow’s MetLife plan had an NPV of approximately $370 million and the investment leg of the Great West plan had an NPV of $76 million at the after-tax discount rate. By analyzing the investment leg separately, Dow’s expert viewed the transaction as a whole — from beginning to end — but removed the challenged deduction from consideration as AEP requires. By using an after-tax discount rate, he recognized the reality that corporations operate in a world with taxes and that some investments are tax favored. Since Dow’s expert’s analysis is a form of the “pre-tax” analysis required by this court in AEP, I am satisfied that the district court did not commit legal error by accepting Dow’s expert’s NPV analysis.
II. Mortality Gains
The second basis for the district court’s determination that Dow’s COLI plans had economic substance was its finding that, unlike the sham COLI plans, Dow’s COLI plans transferred mortality risk to the insurers on the aggregate and therefore were not mortality neutral. There is no basis for concluding that this factual finding was clearly erroneous, and I disagree with my colleagues that the district court committed legal error in distinguishing the mortality features in Dow’s COLI plans from those in AEP and CM Holdings.
This circuit in AEP and the Third Circuit in CM Holdings found that the plans in those cases were designed and operated to be “ ‘mortality-neutral,’ with neither side making money on the risk of employees dying early or late.” CM Holdings, 301 F.3d at 104. “[T]he chance of mortality gains ever being enough to render the plan pre-tax profitable was essentially nonexistent.” Id. Neither the taxpayer nor the insurance company was able to benefit from the timing of employee deaths be*609cause the insurance company paid the taxpayer “mortality dividends” if the cost of insurance exceeded death benefits and assessed COI surcharges if the taxpayer’s death benefits exceeded actuarial projections. AEP, 326 F.3d at 740; CM Holdings, 301 F.3d at 104-05.
Although, as my colleagues point out, the insurance company in CM Holdings did pay out $1.3 million in mortality gains in the early years of the plan, the Third Circuit found the plan was designed to eliminate retrospectively the taxpayer’s ability to realize mortality gains because, “[rjather than accept this loss as one that may sometimes occur no matter how carefully actuaries attempt to chart the vagaries of life and death, [the insurance company] assessed surcharges to recoup its losses and ensure mortality neutrality in the future.” CM Holdings, 301 F.3d at 104-05. As the lower court in CM Holdings explained, although the insurer may never recoup the taxpayer’s mortality gains, “this is a function of a policy contract provision capping maximum COI charges, not a function of risk shifting.” In re CM Holdings, Inc., 254 B.E.. 578, 635 (D.Del.2000).
The plan in AEP was also designed to eliminate retrospectively the taxpayer’s ability to realize mortality gains. My colleagues selectively quote the advice received by the taxpayer to support their argument that the plan in AEP was not designed to provide a 100% adjustment. However, when that advice is quoted in its entirety, it supports the opposite conclusion:
[A representative] advised AEP that “this mortality mechanism will result in a much closer match between expected cash flow from projected death benefits and claims that are actually paid. This will eliminate any volatility or variation in the cash flow that we are expecting in each year of the plan.”
Am. Elec. Power, Inc. v. United States, 136 F.Supp.2d 762, 777 (S.D.Ohio 2001) (citation omitted).
In contrast, the district court below found that Dow’s COLI plans did not contain retroactive adjustment mechanisms designed to eliminate the transfer of mortality risk by equating the COI with the death benefits paid. The district court found that the Great West plan paid Dow dividends to share favorable mortality experience, but those retrospective adjustments could not equalize COI and death benefits paid because Dow’s plans were part of a pool of COLI plans that shared the mortality experience upon which the dividends were paid. Dow, 250 F.Supp.2d at 808. The Great West plan also had an adjustment feature which increased the COI to account for prior unfavorable mortality experience, but that was a prospective measure “designed to fine tune the risk allocation going forward,” not a retrospective measure designed to eliminate past risk transfer. Id. The MetLife plan also paid mortality dividends, but they did not transfer risk away from the insurance company, and benefitted only Dow. Id. at 810. MetLife also reserved the right to increase COI charges to recoup losses if mortality experience was unfavorable, but that right to adjust was limited by a stop-loss provision, which is common in group life insurance contracts. Id.
The district court concluded that Dow’s COLI plans had features that were designed to reduce but not eliminate the mortality risk transferred to the insurers. The court found that Dow had the potential to realize substantial mortality gains, including catastrophic loss, which was a realistic possibility given that Dow’s employees, unlike those in Winn-Dixie, were located in concentrated geographic areas. See Winn-Dixie Stores, Inc. v. Comm’r, 113 T.C. 254, 284-85, 1999 WL 907566 *610(1999). There is no basis for concluding that the district court’s factual finding that Dow had the potential to realize mortality gains because its COLI plans transferred mortality risk to the insurers, on the aggregate, was clearly erroneous or that its conclusion that Dow’s COLI plans were not mortality neutral was erroneous as a matter of law under AEP.
III. Subjective Business Purpose
A close study of Dow’s COLI plans shows that Dow’s plans were not the same as those found to be shams in AEP, CM Holdings, and Winn-Dixie, and that Dow did not enter into the plans for the sole purpose of tax avoidance. The sham COLI plans shared a number of features that indicated a sole purpose of tax avoidance. First, the taxpayers that operated those plans always chose the highest possible interest rates for their policy loans, or at least the highest rates they thought would pass IRS scrutiny. See CM Holdings, 301 F.3d at 100; Winn-Dixie, 254 F.3d at 1315; Am. Elec., 136 F.Supp.2d at 787. The choice of such high, commercially unreasonable rates was rational because those rates allowed the taxpayers to maximize their interest deductions without increased cost due to the fixed 1% spread between the loan interest rate and the credited rate on the policy value. In contrast, Dow chose a commercially reasonable interest rate for its policy loans based on Moody’s Corporate Average, a typical and regulatory-favored rate for policy loans. Dow, 250 F.Supp.2d at 813.
The sham COLI plans also offered loaned crediting rates that greatly exceeded their unloaned crediting rates. For example, in CM Holdings, the unloaned crediting rate was the greater of 4% or a rate declared by the insurance company, and the loaned crediting rate varied between approximately 9% and 13%; and the Winn-Dixie plan offered an unloaned crediting rate of 4% and a loaned crediting rate of 10.66%. See CM Holdings, 254 B.R. at 594-95, 606; Winn-Dixie, 113 T.C. at 281. These large differentials deviated from standard insurance industry practice and created a very strong incentive to continue taking policy loans throughout the life of the plan. See CM Holdings, 254 B.R. at 585-86. In contrast, there was a much smaller differential between Dow’s MetLife unloaned and loaned crediting rates, and Dow’s Great West unloaned and loaned crediting rates were identical. Dow, 250 F.Supp.2d at 778, 789. Dow’s MetLife unloaned crediting rates were between 4% and 6.81%, and its loaned crediting rates were between 4% and 8.65%. Id. at 789. Dow also negotiated a higher un-loaned crediting rate for the MetLife policies to begin in year 18. Id. at 809. Dow’s unloaned and loaned crediting rate choices encouraged Dow to cap policy loans at $50,000 and invest cash into its COLI plans after the 17th year.
The taxpayers in AEP, CM Holdings, and Winn-Dixie also discontinued their plans after Congress passed the Health Insurance Portability and Accountability Act of 1996 (HIPAA) and phased out the interest deductions on COLI plans even though their proposed corporate purposes of funding retiree benefit expenses did not require cancellation of the plans. See AEP, 326 F.3d at 740; CM Holdings, 301 F.3d at 101; Winn-Dixie, 254 F.3d at 1315. In contrast, Dow stopped paying premiums on its MetLife policies, but otherwise kept its COLI plans intact after the passage of HIPAA. Dow, 250 F.Supp.2d at 793-94. There is no reason to conclude that Dow’s COLI plans lacked economic substance.
IV. Evidentiary Rulings
The majority opinion, given its treatment of the “economic substance” issue, does not address the government’s argu*611ment that the district court abused its discretion in excluding Dow’s IRS protest letters as “[evidence of conduct or statements made in compromise negotiations” under Fed.R.Evid. 408. Dow’s IRS protest letters included cash-flow illustrations that did not show loans capped at $50,000. Dow, 250 F.Supp.2d at 802.
The district court’s factual finding that Dow submitted the protests in compromise negotiations is reviewed for clear error. Trans Union Credit Info. Co. v. Associated Credit Servs., Inc., 805 F.2d 188, 192 (6th Cir.1986). The court’s ruling that the protests were inadmissible is reviewed for an abuse of discretion. United States v. Logan, 250 F.3d 350, 366 (6th Cir.2001).
The government argues that Dow’s protest letters were not “made in compromise negotiations” because they were filed with the IRS Examination Division rather than the Appeals Division, whose mission is to resolve tax controversies. Dow, 250 F.Supp.2d at 803. It asserts that the Examination Division’s investigation process is an adversarial one that involves determining whether the IRS erred. Id.
The district court found that, given the undisputed facts that large corporate taxpayers protest the vast majority of IRS disallowances and that the great majority of those protests are settled, Dow reasonably believed that it must first file a protest before it could reach the Appeals Division and compromise its claim. Id. at 805-06. Largely on this basis, the district court found that the protests were “made in compromise negotiations” and excluded them under Rule 408. Id. at 806. I am unable to conclude that the district court’s factual finding regarding Dow’s IRS protests was clearly erroneous.
Federal trial courts have wide discretion in determining whether to admit offers of settlement for “another purpose.” District Judge David Lawson, the trial judge below, is a nationally recognized writer and lecturer in the Federal Rules of Evidence. He would have had in mind the policy objective of Rule 408 to “ ‘weigh the need for such evidence against the potentiality of discouraging future settlement negotiations.’ ” See Trebor Sportswear Co. v. Limited Stores, Inc., 865 F.2d 506, 510 (2d Cir.1989) (citation omitted).
The government also argues that Dow’s protests should not have been excluded because they were being offered to impeach rather than to prove liability or damages. The district court found that there is little difference between using a statement to impeach the denial of liability and offering it to prove liability and therefore refused to apply the “another purpose” exception of Fed.R.Evid. 408 and admit the evidence. Dow, 250 F.Supp.2d at 805.
I agree that using settlement evidence to impeach the denial of liability is virtually indistinguishable from using it to prove liability; both uses would equally discourage settlement negotiations. We do not review the district court’s evidentiary ruling de novo, and I am satisfied the district judge did not abuse his discretion.
Y. Conclusion
I respectfully dissent because I disagree with the majority’s conclusion that, as a matter of law, future profits contingent on taxpayer action are relevant to the economic substance inquiry only when that action comports with the taxpayer’s actual past conduct related to that particular transaction. I also disagree with my colleagues’ conclusion that Dow’s COLI plans were mortality neutral and that the district court erred in distinguishing the mortality features of Dow’s COLI plans from those in the sham-COLI cases.
I would affirm the judgment in favor of Dow because I believe that the district court correctly applied this court’s eco*612nomic substance doctrine, did not commit legal error in accepting the NPV analysis of Dow’s expert, did not abuse its discretion in excluding Dow’s protest letters from evidence, and did not commit clear error in making its central factual findings.