Rhoades, McKee, & Boer v. United States

822 F. Supp. 445 | W.D. Mich. | 1993

OPINION

ENSLEN, District Judge.

Pursuant to Fed.R.Civ.P. 52(e), the Court enters the following Findings of Fact and Conclusions of Law in the above-captioned matter.

Findings of Fact

The conflict between the parties in this case revolves around a defined benefit plan (“the plan” or “the Rhoades plan”). The retirement plan in question was created in 1984 by the law firm Rhoades, McKee, Boer, Goodrich & Titta (“Rhoades, McKee”), and its sole beneficiary was partner Dale Rhoades. The plan received a favorable determination letter from the IRS on its initial qualification on November 4,1985. The level of the firm’s yearly contribution was determined by a variety of actuarial factors, including assumptions concerning retirement age, pre- and post-retirement interest rates, and the life span of the beneficiary. Contributions to the plan were made with funds which otherwise would have gone to Mr. Rhoades as income. Because the law firm was a general partnership until 1987,¹ Mr. Rhoades deducted these contributions on his personal income tax, and was permitted to defer paying tax on the interest this money accrued.²

In 1991, the IRS decided to disallow plan contributions from previous years.³ On April 4, 1991, IRS issued two Notices of Final Partnership Administrative Adjustments which disallowed contributions to the plan for the years 1986 ($151,905) and 1987 ($99,473). On May 6, 1991, the IRS issued a Notice of Deficiency to plaintiff Rhoades, McKee, Boer, Goodrich & Titta concerning its 1988 tax return, disallowing contributions to the plan in the amount of $94,543. On May 29, 1991, Rhoades, McKee deposited $137,619 with the IRS with respect to the disallowance of plan contributions for 1986 and 1987, and it paid taxes, penalties, and interest for the year 1988 in the amount of $77,076. Plaintiffs then sought a refund on taxes paid for 1988, which was denied.⁴

Plaintiffs instituted this law suit to recover these funds. In the joint pre-trial order, the parties defined the contested issues as follows:

Whether the actuarial assumptions used by ■the Plan’s enrolled actuaries to determine the Plan’s cost were reasonable in the aggregate and represented the actuary’s best estimate of anticipated experience under the Plan____ Specifically, the parties dispute (1) the pre-retirement interest rate assumptions ... (2) the 5% post-retirement interest rate assumption ... (3) the retirement age assumption ... and (4) the pre-retirement age mortality table assumptions —

The plan assumptions were as follows. Preretirement interest: 6% (1986); 6% (1987); 7.75% (1988). Post-retirement interest: 5% (1986-88). Retirement age: 60. Mortality tables: 1983-IAM (Female) (1986); 1983-IAM (Male, setback 2 years) (1987); 1983-IAM (Male, no setback) (1988). Mr. Rhoades initially dealt with Robert Bowman when he set up the plan. In 1986, the plan actuary was Ted Retan, and in 1987 and 1988 it was Mike Mojzak.

The Court conducted a four day bench trial, and the closing arguments were filed in brief form. In addition, the Court issued a list of queries which the parties answered. After reviewing all the material submitted in this case, I have concluded that plaintiffs must prevail.

Conclusions of Law

Standard of Review

This is a taxpayer refund suit brought pursuant to 28 U.S.C. § 1346(a)(1) and (e). The taxpayer bears the burden of proving that the assessment was incorrect, and proving the correct amount of tax owed. Phil Smidt & Son, Inc. v. NLRB, 810 F.2d 638, 642 n. 6 (7th Cir.1987).

The IRS regulation which governs the creation of tax deductible defined benefit plans states:

For purposes of this section, all costs, liabilities, rates of interest, and other factors under the plan shall be determined on the basis of actuarial assumptions and methods (A) which, in the aggregate, are reasonable (taking into account the experience of the plan and reasonable expectations) and (B) which, in combination, offer the actuary’s best estimate of anticipated experience under the plan.

26 U.S.C. § 412(c)(3) (emphasis added). Inherent in this statute is a tension which is created by two objectives of Congress. The first is to insure that retirement plans are not underfunded. This is quickly evidenced by the title of § 412, “Minimum Funding Standards.” The second is to insure that they are not overfunded, or, in other words, that tax incentives designed to encourage citizens to provide for their own retirement are not abused as unfair tax shelters. § 412(c)(3) is designed to aid actuaries, the IRS, and courts walk the tightrope between these two prohibited outcomes. In its interpretation and application of this section of the Internal Revenue Code, it is this Court’s mission to define the boundaries of that tightrope.

The first legal issue to be resolved is the standard of review. The U.S. Tax Court has recently decided that the actuary’s assumptions under § 412(c)(3) should be afforded great deference when reviewed by a court, and that those assumptions “will not be changed retroactively unless they are found to be substantially unreasonable.” Vinson & Elkins v. Commissioner, 99 T.C. No. 2, 1992 WL 162641 (1992); Wachtell, Lipton, Rosen & Katz v. Commissioner, RIA T.C. Memo. 1992-392, 64 T.C.M. 128, 1992 WL 162645 (1992); Citrus Valley Estates v. Commissioner, 99 T.C. No. 21, 1992 WL 238873 (1992). As the Tax Court stated in Wachtell, Lipton, “[u]pon audit by respondent, these actuarial assumptions may be determined to be unreasonable in the aggregate, and Wachtell, Lipton has the burden of proving that the assumptions are'reasonable in the aggregate. However, these actuarial assumptions will not be changed retroactively unless they are found to be substantially unreasonable.” Id. at 18.

Defendant urges the Court to reject the “substantially unreasonable” standard. It first argues that by restraining the IRS and the courts in their review of decisions, the standard delegates too much authority to actuaries. Secondly; defendant argues that the statute improperly shifts the burden of proof to the government, and it inappropriately raises the standard of proof from “reasonableness” to “substantially unreasonable.”

Decisions of the Tax Court do not bind district courts. They are, however, entitled to considerable weight. Humphrey v. United States, 245 F.Supp. 49 (1965). Tax Court judges possess expertise in this area of law.⁵ Further, uniform administration of the Tax Code is enhanced if district courts conform to the precedents set by the Tax Court.⁶

Two additional factors increase the persuasiveness of the relevant Tax Court decisions in this case. First, they consider the exact regulation at issue here, virtually the exact same factual scenario, and the same expert testimony.⁷ A closer match of circumstances is hard to imagine. Secondly, the Citrus Valley decision is the product of “test” litigation. The Tax Court had many cases concerning individual defined benefit plans and § 412(c)(3) pending before it. It chose to consolidate 12 cases, which both parties agreed were representative, and hear them together. These cases were heard by a fifteen judge panel, and fourteen judges joined the majority opinion.

As a result, I am inclined to follow the Tax Court standard. The factor which propels this inclination into action is my belief that the standard is a logical and appropriate one.

The House Ways and Means Committee report that accompanied an early draft of § 412(c)(3) to the floor contains the following statement:

Since the actuarial assumptions used must be reasonable in the aggregate, it is anticipated that, on audit, the Internal Revenue Service will (as presently) require a change of assumptions where they do not meet this standard. However, unless the assumptions used are substantially unreasonable, it is contemplated that generally the Service will not require a change of assumptions to be made effective for years prior to the year in which the audit is made.

H.R.Rep. No. 93-807, 93d Cong., 2d Sess. 95 reprinted in 1974-3 C.B. (Supp.) 236, 330 (emphasis added). I interpret this passage to suggest that assumptions which the IRS deems unreasonable will not be retroactively altered unless they are substantially unreasonable. However, a plan could be required to prospectively alter its assumptions if they are simply unreasonable.

The government argues that the Tax Court’s reliance on this passage was misplaced because it accompanied a version of the bill which differed from that which was passed. However, the only significant revision in the version which was enacted into law was the addition of the “best estimate” language. Therefore, I believe that this piece of legislative history remains helpful.

Furthermore, there are two reasons why the fact that a reviewing court will not retroactively alter actuarial assumptions unless they are substantially unreasonable does not undermine Congress’ dictate that actuaries must choose reasonable assumptions. The first focuses on the distinction between individual assumptions and the assumptions in the aggregate. In a decision denying summary judgment in this case, I remarked that I found it impossible to pass on whether any individual assumption was “reasonable” as a matter of law, because the statute instructs the courts to consider the assumptions in the aggregate. Perhaps the Tax Court’s standard provides a remedy to this problem. These decisions applied the substantially unreasonable standard to individual assumptions. Therefore, one interpretation of the Tax Court’s decisions is that it used this standard as a sort of “rule of thumb.” If no individual assumption is “substantially unreasonable,” then the aggregate of the assumptions must be “reasonable.”

The second reason this standard does not contradict the statute focuses on the retroactivity language, and the distinction between the reasonableness of the assumptions used year-to-year and the reasonableness of the long-term result of those assumptions, which is a properly funded plan. The IRS seems to forget that proper actuarial assumptions are not an end in themselves, but a means to an end. The fact that assumptions used in past years which are unreasonable, but not substantially so, will not be retroactively altered does not mean that the government is without recourse. § 412(c)(3) requires actuaries to formulate assumptions “taking into account the experience of the plan and reasonable expectations.” I believe that an IRS determination that past plan assumptions were unreasonable would certainly be relevant past experience which an actuary should factor into the choice of future assumptions. Therefore, future assumptions could be selected with an eye to balancing out unreasonable past assumptions. The long-term result will be a correctly and reasonably funded retirement plan.⁸

Finally, I address defendant’s argument that the “substantially unreasonable” standard delegates too much authority to actuaries. The IRS asserts that because actuaries are always under pressure from their employers to get the “right” numbers, their decisions do not deserve deference, and should be scrutinized. However, many other professionals are under pressure from clients to achieve a particular result. This issue is commonly addressed by licensing procedures. In the case of actuaries, Congress identified intellectual and ethical standards and created a rigorous procedure for licensing “enrolled” actuaries. After reviewing the history of ERISA, the Wachtell, Lipton court concluded that Congress enacted its primary regulations in this field in its certification requirements for enrolled actuaries. Thus, it appears that Congress set out to regulate actuaries rather than assumptions. Therefore, I agree that once an actuary has attained this status, judges may comfortably defer to his or her professional judgment in the context of the “reasonableness” analysis, required by the first prong of § 412(c)(3).

In summary, for the reasons given above, as well as a healthy dose of deference to the Tax Court due to its expertise and an eye toward uniform enforcement of the tax code, I will not order modification of actuarial assumptions made in the past unless they are “substantially unreasonable.”

The Objective: The Reasonableness of Assumptions Requirement

Section 412(c)(3) contains an objective component, the “reasonableness” of the actuarial assumptions, and a subjective component, whether the assumptions represent the “best estimate” of the enrolled actuaries. I will first review the objective component.

However, before I reach the three sets of assumptions, I must comment on an issue raised at trial and in the parties’ briefs: the form of payout. The government contends that the actuaries should have funded for a lump sum payout. Plaintiffs contend that the form of payout is irrelevant, because it is not specified as a disputed issue in the pretrial order. In this case, I believe that the form of payout is not properly considered an actuarial assumption. It is a plan provision, and actuarial assumptions are designed in response to such provisions. Therefore, I will take the form of payout as a given (in this ease, joint and survivor annuity). It is relevant only in the sense that actuarial assumptions must be reasonable in relation to it. For example, an assumption which might be reasonable if a lump sum payout is planned might be unreasonable if an annuity payout is planned. However, I do not consider the reasonableness of the choice of form of payout itself an issue in this case.⁹

I have concluded that the actuarial assumptions in question in this case were reasonable in the aggregate, and thus are in accord with the first requirement of 26 U.S.C. § 412(e)(3). I will discuss the three areas of assumption in turn.

Interest Rates

In 1986 and 1987, the plan’s actuaries assumed pre-retirement interest of 6%. In 1988, 7.75% pre-retirement interest was assumed. The plan’s actuaries assumed post-retirement interest of 5% for all three years. The government contends that a reasonable assumption for both pre- and post-retirement interest rates was 8%.

As stated above, the reasonableness of assumptions is objective. Therefore, there is no reason why courts across the country should not be able to come to some agreement as to what the range of reasonableness on a factor like interest rates for one or two person plans is for given years. After an extraordinarily detailed analysis, the Tax Court found that 5% was acceptable as both a pre-retirement and a post-retirement interest rate assumption in Vinson & Elkins, Wachtell, Lipton and Citrus Valley. These decisions concluded that for the first three years of a plan’s existence, an actuary must use his or her own judgment. The Vinson & Elkins court explained that

prudent actuaries maintain a long-term conservative view that will insure benefit security for plan participants.... If the assumptions are too conservative, then the plan sponsor will decrease contributions accordingly — an easier task than finding resources that were not budgeted. Use of overly optimistic assumptions therefore increases the changes that the response will ultimately fail to fully fund the plan.

Vinson & Elkins, supra at K 12-13.

On the authority of the Tax Court decisions and the evidence presented at trial, I conclude that the Rhoades plan’s selection of interest rate assumptions was not outside the range of reasonableness.

Retirement Age

The plan’s actuaries assumed that Mr. Rhoades would retire at age 60. The government contends that the proper assumption was 65. The government’s expert, Ronald Hanenberg, identifies the range of reasonableness as 60 to 70.5, and contends that the proper assumption was 63. In Vinson & Elkins, the Tax Court approved a retirement age assumption of 62. In Wachtell, Lipton and Citrus Valley, the Tax Court approved a retirement age assumption of 55.

I see nothing unreasonable about the selection of age 60 in this case. Without speaking with him about his retirement plans specifically, an actuary can learn that Mr. Rhoades engages in a high income, high stress occupation. In addition, the agreements governing his employment do not require that he cease all work in order to receive retirement benefits. These factors alone could render this assumption rational and reasonable. In addition, age 60 is five years older than an assumption approved in two Tax Court cases, and it falls within the government’s own expert’s range of reasonableness.

The Vinson & Elkins court held that an actuary is not charged with the responsibility of determining when a beneficiary will actually retire, because that would be an impossible task. Instead, the court held, an actuary is charged with determining when benefits under the plan could begin. Vinson & Elkins, supra, at K-24. Mr. Rhoades expressed his intent to retire at age 60 when he selected plan provisions, and therefore that is that age benefits could begin. This assumption is reasonable.

Finally, the government argues that even if age 60 was initially reasonable, a plan amendment made in 1987 rendered it unreasonable for later years. Mr. Hanenberg’s estimate of 63 is based primarily upon the fact that, as a result of the 1987 plan amendment, the maximum benefits would not be available unless Mr. Rhoades worked until age 63. However, actuaries Retan and Mojzack testified that they mistakenly believed that the amendment had grandparent-ed the increased benefit levels under the Tax Reform Act of 1986. Therefore, given the premise they were operating from, their conclusions were not unreasonable. I believe actuaries’ “intent” or mistaken beliefs in calculations are more relevant in the best estimate analysis rather than the reasonableness analysis.

Mortality

In 1986, the plan’s actuaries made mortality assumptions using a table called 1983-IAM (Female). In 1987, the plan’s actuaries used the same basic table, but they used the calculations for males, and set it back two years. In 1988, they again utilized the 1983-IAM (Male), except without a setback.¹⁰

The Vinson & Elkins and Wachtell, Lipton decisions found the use of a table established in 1971 for males (1971-IAM) reasonable. The twelve cases consolidated in the Citrus Valley litigation used a variety of tables. In that case, the court found the use of a 1983-IAM table with a four year setback to be reasonable. In addition, it approved the use of the 1983-IAM (Female) table, used for a male and setback seven years. The court held that “although we are not entirely convinced that [this table] is completely reasonable, it is not substantially unreasonable so as to justify a retroactive adjustment. Furthermore, any misgivings we have about his assumption are not sufficient to affect the reasonableness of the actuarial assumptions in the aggregate.” Citrus Valley, supra, at 87.

The IRS asserts that plaintiffs should have used the 1983-IAM (Male) all three years. Because women live longer, the result of calculating the mortality of a male using a female table is to increase it by approximately four years. Therefore, I estimate that the difference between the IRS’s preferred table and that used in 1986 is approximately four years. The difference in 1987 is two years, and there is none in 1988. Given a projected life span of approximately 30 years, the government’s contention that the plan’s choice of mortality assumptions is outside the range of reasonableness must be rejected.

I do not dispute Mr. Hanenberg’s conclusion that the Buck table might be a more accurate predictor of mortality than the 1983-IAM. However, I find the suggestion that plaintiffs should be penalized for their failure to use the Buck table, an unpublished table which is not listed as an option on the relevant tax form, untenable. I believe that the mortality assumptions used in the plan were reasonable.

The Assumptions in the Aggregate

I do not believe that any of the foregoing assumptions were substantially unreasonable. I therefore conclude that they are reasonable in the aggregate, and satisfy the objective requirement of § 412(c)(3).

The Subjective: The “Best Estimate” Requirement

Through the course of the trial, I found the second prong of 26 U.S.C. § 412(c)(3), which requires that the assumptions represent the actuary’s “best estimate of anticipated experience under the plan,” the most difficult. The function of this clause seems to be to keep the actuary’s “eye on the ball.” It puts the reasonableness requirement in context, and it instructs an actuary on what criteria s/he should employ in order to pick one number among several reasonable choices.

The parties’ conflict over the interpretation of the “best estimate” requirement revealed the sharp divergence of their approaches. In a sense, plaintiffs were required to defend generally accepted actuarial practices, and for that reason I am curious why no professional actuarial association chose to submit an amicus brief. Nonetheless, after listening to the actuaries who testified at trial and reviewing the voluminous documents submitted in this ease, I have the impression that actuaries see themselves as operating in a world of numbers, tables and averages, which they use to make forecasts. This statistical information is constructed from interaction with human beings, but the actuaries creating and administering pension plans do not usually interact with those people. As- a result of this perspective, it would not seem at all odd to most actuaries that the actuaries in this case would not conduct a detailed inquiry into Mr. Rhoades’ life, health, investment habits and plans for the future. , Instead, their “best estimate” is made in reference to such statistical forecasting tools.

In contrast, the IRS perceives Mr. Rhoades as a unique individual, who has a tax burden that is tailored to his personal circumstances. A tax audit does not ask what the average lawyer earns per year, it asks what Mr. Rhoades earns. Therefore, the IRS seems to expect actuaries to conduct a sort of individualized “audit” when they create a defined benefit plan. The “best estimate” concept was brought into question through the government’s contention that a plan actuary cannot make a “best estimate” assumption without consulting the participant.

I have found little precedent to guide me in this area. Relative to their lengthy discussion of the reasonableness provision, the three Tax Court cases spend very little time discussing the best estimate concept.¹¹ These decisions repeatedly state that this provision means that actuaries cannot simply “rubber stamp” plan assumptions selected by beneficiaries. However, they do not specifically identify what else actuaries must do to comply with the best estimate requirement. The fact that the Tax Court approved of 132 identical plans in Vinson & Elkins and 41 identical plans in Wachtell, Lipton suggests that it did not interpret the best estimate clause to require a highly individualized inquiry.

The best analogy to the statute at hand was not discussed by either party. It is 29 U.S.C. § 1393(a)(1), which is essentially identical to § 412(c)(3). This statute governs the penalty calculation for an employer who withdraws from a multi-employer pension plan. It requires that “withdrawal liability ... be determined ... on the basis of actuarial assumptions ... which in the aggregate are reasonable ... and which in combination, offer the actuary’s best estimate of anticipated experience under the plan.”

I reviewed ten appellate cases which considered disputes arising under this statute, but only one directly addressed the best estimate portion. This case, Huber v. Casablanca Industries, Inc., 916 F.2d 85 (3rd Cir.1990), found that an actuary’s figures did not represent his best estimate. The foundation of this finding was the fact that a week after the actuary submitted a report valuing the assets of the fund, the Board of Trustees expressed its displeasure with his calculation method, and stated that it should consider replacing him. A few days later, it appears that the actuary recalculated the figures to accommodate the Board. Id. at 93. The clear implication is that the actuary included improper considerations in his calculations.

Although I realize that the Huber court did not necessarily intend to define the entire scope of the best estimate clause it was considering, I agree with its approach. The plain meaning of this phrase suggests that a “best estimate” is the product of an actuary’s professional expertise when s/he has considered all relevant factors, and no irrelevant or improper factors. E.g., Yates v. City of Cleveland,- 941 F.2d 444 (6th Cir.1991) (penalizing army subcontractor because price quote which did not factor in the price-breaks subcontractor knew it could obtain did not represent its “best estimate.”) Therefore, I hold that when it is clear that considerations which are unrelated to the anticipated experience of the plan have been factored into an actuary’s estimate, that estimate does not represent the actuary’s “best” in accordance with § 412(c)(3).

The government contends that one factor which it is improper for an actuary to consider in formulating a best estimate is the element of “conservatism.” The best support I have found for this proposition is Thor Power Tool Co. v. Commissioner, 439 U.S. 522, 99 S.Ct. 773, 58 L.Ed.2d 785 (1979). In Thor the Supreme Court interpreted 26 U.S.Cl § 471. This tax statute governs inventory accounting, and it requires such accounting to “conform[] as nearly as may be to the best accounting practice in the trade or business and as most clearly reflecting income.” The Thor Court held that the first prong of the statute, which anticipates use of generally accepted accounting principles, creates a range of reasonable treatments, and thus management must pick one. Without the second part of the statute, a business “could decide unilaterally — within limits dictated only by its accountants — the tax it wished to pay.” Id. at 544, 99 S.Ct. at 787. The Court held that although ordinary business accounting may preach a conservative view, in order to avoid overestimating assets, tax accounting cannot. Id. at 543-44, 99 S.Ct. at 786-87.

Although I find the Supreme Court’s opinion in Thor instructive, there are two reasons why it does not control my views of the best estimate clause in this case. The first is that the “generally accepted accounting method” Thor employed, the “lower of cost or market” method, was accompanied by regulations' which defined how to “clearly reflect income.” Thor’s practice did not comply with those regulations, and therefore it was easy to conclude that its accounting practices did not comply with the second prong of the test. In this case, there are no statutes which the Court can refer to in order to define “best estimate.” Secondly, in Thor, the sole government interest behind the tax statute is to make sure accountants in private businesses do'not undervalue assets, and therefore cheat the government’s coffers. However, there are two governmental interests propelling § 412(c)(3). The first is to avoid overfunding, which robs the State coffers, and the second is to avoid underfunding, which prevents citizens from providing for their own retirement. ■ In some cases, the second result requires a State bail-out, which also costs taxpayers. In all situations in which estimates are required, planning for eiTor is part of the calculation. In contrast to business accounting, I believe that the intent of Congress in the field of pension funding was for actuaries to err on the side of over-funding rather than underfunding. Therefore, a degree of conservatism in assumptions is warranted in defined benefit plans.

In addition, I reject the IRS’s categorical contention that satisfaction of this clause requires personal consultation in every case, concerning every assumption. While a beneficiary can proviae an actuary with information unique to that individual, the actuary must still employ averages in order to come up with numerical assumptions. Mr. Rhoades could not tell his actuary exactly when he would retire, exactly when he would pass away, and exactly how much he would make from his investments. As further detailed below, I believe that an actuary has an obligation to make personal contact in establishment of the retirement age assumption. A reasonable duty of inquiry exists in other unusual circumstance, such as when a client gives information which the actuary reasonably believes is false. However, no “unusual” information invoking such a duty seem to exist here.

Finally, I note that § 412(c)(3) dictates that the plan assumptions represent the actuary’s best estimate of anticipated experience under the plan in combination. Therefore, as with the reasonableness of assumptions, the estimates must be evaluated as whole to, determine whether they are the actuary’s “best.”

Interest Rates

The IRS has argued, as has its expert Ronald Hanenberg (Hanenberg Report at 17), that § 412(c)(3) does not permit actuaries to “hedge their bets” with a degree of conservatism; it requires them to pick their best estimate of actual experience out of a range of reasonableness. As I indicated above, I reject this stance. In regard to interest rates, this argument misses the forest for the trees. Yearly assumptions are not an end in themselves, they are a means to an end. That end is a retirement plan which is properly "funded at the time its beneficiary ceases generating income and begins drawing benefits. There is no “cushion” at that point; the money simply must be there. Although the slight over-funding which conceivably could result from conservative interest rate assumptions in the early years of a plan will result in incorrectly inflated tax deferments, the correction of this overfunding will balance out both the funding of the plan and the payment of tax in the long run.

The government takes exception to the fact that the actuaries did not consider Mr. Rhoades’ investment experience in determining the interest rate assumptions. In Wachtell, Lipton, the Tax Court pointed to the fact that the plans did not have professional investment directors managing the funds as one factor supporting a conservative interest rate assumption for the plans’ first years. The court went on to reject the government’s argument that it was unreasonable for the Wachtell, Lipton actuaries to use the same interest rate assumptions for each of 41 plans without evaluating each partner’s investment strategy. It found that all the experts agreed that until a plan establishes sufficient “credible experience” regarding investment strategy and rates of return, such information carries little or no weight. Wachtell, Lipton, supra, 49-50.

I agree with the Tax Court on this issue. Until the plan establishes credible experience, the investor’s personal investment philosophy is irrelevant. Therefore, I do not believe that there is any evidence that the interest rate assumptions do not reflect the actuaries’ best estimate.

Retirement Age

I agree with the government’s contention that an actuary cannot offer his or her “best estimate” of a retirement age without speaking with the beneficiary. The average age of retirement decreases in significance dramatically when one is considering a particular individual. However, the government does not dispute the fact that Mr. Bowman, the actuary who handled the initial design of the plan, did speak to Mr. Rhoades about this issue. During that conversation, Mr. Rhoades told him that he would like to be financially able to retire at age 60. Additional, Mr. Tittel’s deposition testimony is that he met yearly with Mr. Rhoades to consult on plan design.¹²

However, this discussion does not satisfy the government, and it is at this point in its argument that I part company with its views. The government essentially contends that it is the actuary’s responsibility to press ■further and discover if the beneficiary is telling the “truth” about his or her retirement plans, or whether the beneficiary’s projection of his or her retirement age is based on the appropriate factors. Mr. Rhoades provided several rational reasons to explain his choice of age 60 at trial. The government correctly points out that there is no evidence that he shared these reasons with his actuary when he selected age 60.

However, there is some irony in the fact that the government hired an expert to identify a “best estimate” as to Mr. Rhoades’ retirement age, and he made one without ever talking to Mr. Rhoades. At trial, Mr. Hanenberg acknowledged that selecting a best estimate from a reasonable range is a “tough question,” and that his selection was somewhat “arbitrary.” ¹³ His best estimate of 63 was based in part on how a “rational person” would react to the fact that, due to a 1987 plan amendment, the maximum benefits would not be available until that time. The government contends that this fact prowdes objective evidence that a retirement age assumption of 60 could not have been the actuary’s best estimate in 1988. The Court is left to ponder why the government believes that it was permissible for Mr. Hanenberg to speculate as to this response. By the government’s view, Mr. Hanenberg should have phoned Mr. Rhoades and asked how he intended to respond to this fact, for while finances are a significant factor in retirement decisions, they are not the sole factor, particularly for wealthy individuals.

I conclude that an assumption of a retirement age of 60 represented the actuaries’ best estimate.

Mortality Tables

I finally reach the issue of the best estimate requirement and the mortality assumptions used in this case. This is the most troubling area, because the annual changes in mortality tables were made without discussion with Rhoades and/or his spouse. Ted Retan, ■ the plan’s actuary in 1986, used a female mortality table. He testified that he would typically use a female table for a male who was in excellent health, but he stated that he has no memory of discussing Rhoades’ health with Rhoades or anyone else, and has no recollection of why he shifted to the female table. Mike Mojzack, the actuary in 1987 and 1988, testified that he was “uncomfortable” using the female table, and so in 1987 he switched to the parallel male table, with a two year setback. Records seem to reveal that in 1988 he used the male table without the two year setback, but Mojzack testified that if he did not use a setback in 1988, it was not through a conscious choice.

Plaintiffs' expert Klinger points out that the use of a female table for a male beneficiary may make sense when the payout form is joint and survivor 100% annuity, because after Mr. Rhoades’ death, Mrs. Rhoades would continue to receive full benefits if she were living. Therefore, the relevant life-span to measure might be the female’s, because on average women live longer. However, there was no testimony that this is what Mr. Retan had in mind when he switched to the female table.

In Citrus Valley, the government argued that in one individual’s case, a four year setback in the 1983-1AM table was unreasonable because his family medical history suggested early mortality. The court held that the individual’s concerns about his mortality “would not affect his actuarial life expectancy, nor would they decrease the cost of providing an annuity for him. Since the purpose of the mortality assumption is to determine cost and not actual mortality,” it was reasonable to disregard health concerns when establishing the mortality assumption. Citrus Valley, swpra at 87.

I agree that actuaries do not have to consult individual health histories to select mortality tables. My concern in this area is not with the exact table the actuaries selected, but the reasons why they did so. The only explanation for Mr. Retan’s use of a female mortality table seems to be an effort to increase contribution levels which did not properly consider “anticipated plan experience” — in this instance Mr. Rhoades’ life span. An actuary can pick the table most appropriate for the person of the age and gender s/he is funding for, or s/he can consider unique health factors to select a table (e.g. a present terminal disease), but the table chosen must have some relationship to the beneficiary. Therefore, while I can imagine factors which would justify the use of the female table, it is the actuary’s imagination which I must probe. In this case, the taxpayer has not sustained his burden, and I can only conclude that the inexplicable table switching which occurred was motivated by improper considerations. These choices cannot be countenanced as a “best estimate.”

The Assumptions in Combination

Although I cannot conclude that the switch of mortality tables in 1986 represented the actuaries’ best estimate, I believe that the assumptions in combination represent the actuaries’ best estimate of anticipated experience under the plan. Therefore, I believe the taxpayer has satisfied both prongs of § 412(c)(3) for the years in controversy.

Summary Judgment

On October 19, 1992, I denied plaintiffs’ motions for summary judgment and partial' summary judgment. In this Opinion, I stated that I could not resolve the issue of whether the plaintiffs’ assumptions were reasonable in the aggregate on the basis of the Tax Court precedent. I also concluded that I could not address the reasonableness of one assumption on a motion for partial summary judgment because the assumptions must be considered in the aggregate. Although one other court has determined that whether assumptions are reasonable in the aggregate is a question of fact, Jerome Mirza & Assoc’s, Ltd. v. United States, 692 F.Supp. 918, 921 (C.D.Ill.1988), I want to state on the record that I might come to a different conclusion today. I believe that the lengthy analysis above demonstrates that the question of reasonableness in the aggregate may be appropriately resolved on summary judgment in some cases, and perhaps the best estimate analysis could be as well. My statement in the summary judgment opinion was too categorical, and I retract it.

Conclusion

■ In its post-trial brief, the government asserts that even if a beneficiary’s sole intent in setting up a defined benefit plan is to shelter' money from immediate taxation, the plan is permissible if it complies with. § 412(e)(3). However, the tenor of the United States’ , presentation at trial suggested that Mr. Rhoades was charged with being a wealthy person who wanted to retain the maximum amount of his income permitted by law. As the IRS now concedes, whether Mr. Rhoades’ motivation to set up his defined benefit plan was a desire to defer taxation on as much of a unique “income bulge” as possible is irrelevant. What is relevant is whether the plan assumptions were reasonable in the aggregate and in combination represent his actuaries’ best estimate of anticipated plan experience. I conclude that they did, and therefore plaintiffs must prevail.

JUDGMENT

In accordance with the Opinion entered on this date and pursuant to Fed.R.Civ.P. 58;

IT IS HEREBY ORDERED that judgment be entered in favor of plaintiffs.

IT IS FURTHER ORDERED that the deductions taken for the plan in 1986 ($151,-905), 1987 ($99,473) and 1988 ($94,543) are ALLOWED. ■

IT IS FURTHER ORDERED that the penalties proposed by the IRS in their notices of deficiency relating to the disallowed contributions are ABATED.

IT IS FURTHER ORDERED that the IRS’s 1988 refund to plaintiffs is APPROVED.

IT IS FURTHER ORDERED that defendant shall RETURN to plaintiffs the amount of tax in controversy remaining ($163,507) which plaintiff was required to deposit -with the IRS, plus accrued interest from the date of payment calculated at the customary rate.

1 . The firm incorporated in 1988.

2 . Due to tax law changes, the plan would have lost its qualified status in 1989, so it was terminated and changed into an IRA in December 1988.

3 . This disallowance appears to be part of a national program. On November 29, 1989, the IRS established "guidelines” for its agents in the evaluation of actuarial assumptions of small defined benefit plans. The guidelines mandate the use of 8% interest rate assumptions and an age 65 retirement age. According to an article published in 1991, the IRS expected to audit 12,000 annual reports of small plans, and it anticipated receiving approximately $800 million in income, excise, and penalty taxes as a result. C. Frederick Reish & Buree L. Ashton, Most Small Plans Face Disallowances Under New Actuarial Audit Program, 75 J. Tax'n 4 (1991).

4 . Plaintiffs have informed the Court that after the commencement of this litigation, the IRS issued a check to plaintiffs for $57,178.83. Plaintiffs explain that this amount represents a full refund for the 1988 tax deposited ($51,188) and interest ($5,990.83). I can only assume that the difference between the $51,188 refunded and the $77,-076 plaintiffs assert they deposited represents penalties.

As a result, $163,507 remains on deposit with defendants pending the outcome of this litigation.

5 . Such expertise is no small advantage, as district court judges trying cases such as this one quickly learn. My sentiments in this case echo those of Learned Hand:

In my own case the word of such an act as the Income Tax ... merely dance before my eyes in a meaningless procession: cross-reference to cross-reference, exception upon exception— couched in abstract terms that offer no handle to seize hold of — leave in my mind only a confused sense of some vitally important, but successfully concealed purport, which it is my duty to extract, but which is within my power, if at all, only after the most inordinate expenditure of time. I know that these monsters are the result of fabulous industry and ingenuity, plugging up this hole and casting out that net, against all possible evasion ..! that they were no doubt written with a passion of rationality; but that one cannot help wondering whether to the reader they have no significance save that the words are strung together with syntactical correctness.

L. Hand, The Spirit of Liberty: Papers & Addresses of Learned Hand, 213 (I. Dillard ed. 1960), quoted in Mirza v. United States, 692 F.Supp. 918 (C.D.Ill.1988).

6 . This concern for uniformity has led at least one circuit to conclude that unless another circuit's decision is "demonstrably erroneous,” or "cogent reasons” support its rejection, that it will follow the precedent of other circuits in tax cases. Delk Investment Corp. v. United States, 344 F.2d 696 (8th Cir.1965) (quoting Birmingham v. Geer, 185 F.2d 82, 85 (8th Cir.), cert. denied, 340 U.S. 951, 71 S.Ct. 571, 95 L.Ed. 686 (1950)).

7 . Mr. Hanenberg, the government's expert witness, testified in Vinson & Elkins and Mr. Klingler, plaintiffs' expert witness, testified in the Citrus Valley case.

8 . I am aware that in practice, the fact that authorization for this type of defined benefit plan was terminated prevents the future corrections I am describing. However, I have no reason to believe that Congress contemplated that this termination would occur so soon when it drafted § 412(c)(3). Therefore, I believe that my interpretation of § 412(c)(3) should not factor in this event.

9 . It is relevant to note that the assumptions in all of the 132 plans considered in Vinson & Elkins were geared to plan provisions of joint & surviv- or annuities.

10 . I note that there is some dispute as to whether the actuaries used a setback in both 1987 and 1988. Whether they did or not is immaterial to my conclusion.

11 . In Vinson & Elkins, the court opined that Since Congress must have recognized that there would be a range of reasonable assumptions, and since actuarial calculations require a single value for each assumption, a particular assumption would have to be chosen from each reasonable range. That choice is to be the actuary’s best estimate. Thus, the ‘best estimate’ language accommodates the mathematical realities of actuarial calculations without imposing a higher or stricter standard on the requirement that the actuarial assumptions be reasonable in the aggregate as respondent implies. The ultimate inquiry remains whether the assumptions chosen are reasonable in the aggregate in combination with all actuarial assumptions.

Vinson & Elkins, supra at K-29-30. Boiled to its essence, this paragraph reads, "the best estimate clause requires actuaries to pick a number within the range of reason, which by definition will be reasonable.” I find this observation unhelpful.

12 . "I would have reviewed with him essentially the plan specifications; the relationship of value of benefits to value of assets; the funding limits; as to whether or not he personally had any changes in his situation that would or could precipitate a change in plan specs; just very general type of things.” Tittel deposition at p. 8, 1.24 — p. 9, 1.4.

13 . In fact, one article the Court reviewed asserts that an internal IRS training manual states "[flor purposes of the actuarial examination program, a (somewhat arbitrary) definition of reasonableness has been adopted: retirement ages of 65 of more are reasonable.” Reish & Ashton, supra, at n. 20, citing "Lesson 5 — Actuarial Update” at 5.