Vinson & Elkins v. Commissioner

Vinson & Elkins, a Partnership, J. Evans Attwell, Tax Matters Partner, Petitioner v. Commissioner of Internal Revenue, Respondent; Vinson & Elkins, J. Evans Attwell, Tax Matters Partner, Petitioner v. Commissioner of Internal Revenue, Respondent Vinson & Elkins v. Commissioner Docket Nos. 12030-90, 12412-91 United States Tax Court 99 T.C. 9; 1992 U.S. Tax Ct. LEXIS 54; 99 T.C. No. 2; 15 Employee Benefits Cas. (BNA) 1811; July 14, 1992, Filed *54 Decisions will be entered for petitioner. P is a general partnership engaged in the practice of law. Effective Sept. 1, 1984, P adopted individual defined benefit plans for a majority of its partners. Each plan created a trust to provide for the investment and administration of the plan assets. For the taxable years at issue, 1986 and 1987, all contributions to the plans were made within the time required by sec. 401(a)(1) and (6), and each of the plans and related trusts was qualified under sec. 401(a) and was exempt from taxation under sec. 501(a). The appropriate Forms 5500 and Schedules B were filed with the Service. R challenged various funding assumptions made by the plans' enrolled actuary in determining the amount of tax-deductible contributions for each plan for plan years 1986 and 1987. Held: The actuarial assumptions made by the plans' enrolled actuary were reasonable in the aggregate and represented the actuary's best estimate of anticipated experience under the plans, as required by sec. 412(c)(3); accordingly, as the assumptions used were not substantially unreasonable, R is precluded from requiring a retroactive change of assumptions. Thomas P. Marinis,*55 Jr., David T. Harvin, Sarah A. Duckers, and Miriam M. Burke, for petitioner. Marion S. Friedman, Lillian D. Brigman, and David B. Mora, for respondent. Clapp, Judge. CLAPP *9 *2*TABLE OF CONTENTS Factual Background [11] Law [13] Experts [21] Interest Rate [24] Background [24] Petitioner's Actuarial Experts [29] Respondent's Actuarial Experts [33] Discussion [36] Retirement Age [41] 1958 CSO Table for Preretirement Mortality [49] Postretirement Expense Load [53] Best Estimate [56] Tax Motivation [57] Evidentiary Matters [58] Conclusion [59] *10 OPINION Clapp, Judge: These consolidated cases are partnership actions for readjustment of partnership items under section 6226. Vinson and Elkins (V&E) is a general partnership engaged in the practice of law. On April 25, 1990, and April 15, 1991, respondent issued notices of final partnership administrative adjustment (FPAA) to disallow deductions in the amounts of the $ 7,953,115 and $ 3,496,517, respectively, claimed on the V&E partnership tax returns for the years ended September 30, 1986, and September 30, 1987. The issue for decision in these cases is whether actuarial assumptions used by the enrolled*56 actuary for the individual defined benefit plans (IDB plans) of V&E's partners to determine the plans' costs were reasonable in the aggregate and represented the actuary's best estimate of anticipated experience under the plans as required by section 412(c)(3). Specifically, respondent determined: (1) The 5-percent preretirement and postretirement interest rate assumption used by the plans' actuary was unreasonably low; (2) the age 62 retirement assumption was unreasonably low; (3) the preretirement mortality assumption based upon the 1958 Commissioner's Standard Ordinary Mortality Table (1958 CSO mortality table) was unreasonable; and (4) the 5-percent postretirement expense load was unreasonable, and that those assumptions were not offset by any other assumptions that would make the assumptions in the aggregate reasonable. All section references are to the Internal Revenue Code in effect for the years in issue, and all Rule references are to the Tax Court Rules of Practice and Procedure, unless otherwise indicated. *11 Factual Background The stipulation of facts and exhibits attached thereto are incorporated by reference. V&E is a general partnership organized under*57 the laws of Texas engaged in the general practice of law. V&E has been in existence continuously since 1917. During the years in issue, V&E maintained offices in Houston, Dallas, and Austin, Texas; Washington, D.C.; and London, England. In April 1986, V&E had 168 percentage partners and in August 1987, V&E had 172 percentage partners. During the years in issue, J. Evans Attwell was the managing partner of V&E and the tax matters partner (TMP) pursuant to section 6231(a)(7). V&E has at all times maintained its books and records and filed its Federal partnership returns of income using the cash method of accounting. Effective September 1, 1984, 117 out of 155 V&E partners adopted IDB plans. Eighteen plans were adopted after 1984, and three plans were terminated in 1986. These plans were virtually identical except for the benefit formula and the identity of the participants. Each of the 132 plans created a trust to provide for the investment and administration of the plan assets. Each partner for whom a plan was created was appointed as the investment cotrustee for his or her plan. Each of the partners for whom a plan was created entered into an agreement with V&E, which outlined*58 the acceptable investments for the plan assets and set forth rules regarding the handling of plan funds. Effective September 1, 1986, 49 of the 132 plans were amended to add a new preretirement death benefit, which was to be insured by a life insurance policy purchased by the plan trustee for each plan. Each partner bore the cost of funding his or her plan by making available to V&E sufficient funds to make the annual contributions or by allowing V&E to deduct the minimum funding amount from his or her distributive share of V&E profits. V&E allocated to each partner the full expense of any V&E contribution to the plan and the related income tax deduction. All contributions were made within the time required by section 404(a)(1) and (6). For the taxable years in issue, each of the plans and related trusts was a qualified plan or trust under section 401(a) and was exempt from taxation under section 501(a). *12 The enrolled actuary for the plans was Frederic J. Smith (Smith), an employee of Wolper, Ross & Co. Ltd. (Wolper Ross). Smith is an associate of the Society of Actuaries, a member of the American Academy of Actuaries and has been an enrolled actuary pursuant to the*59 Employee Retirement Income Security Act of 1974 (ERISA), Pub. L. 93-406, 88 Stat. 829, since 1977. Smith used a 5-percent preretirement and postretirement interest rate assumption for the plans. In addition, he selected the later of age 62 or 5 years of plan participation for the retirement age assumption. He used a 5-percent postretirement expense load assumption. For those plans that provided for the funding of a preretirement death benefit, Smith assumed a preretirement mortality based upon the 1958 CSO mortality table. Smith prepared and signed the Internal Revenue Service (the Service) Forms 5500, Schedules B, Actuarial Information (Forms 5500), with respect to each plan for the September 1, 1986, to August 31, 1987 (1986 plan year), and September 1, 1987, to August 31, 1988 (1987 plan year), plan years. He certified on the Forms 5500 that for each plan the assumptions used in the aggregate were reasonably related to the experience of the plan and to reasonable expectations, and represented his best estimate of the anticipated experience under the plan. V&E timely filed its Forms 1065, Information Return of Partnership, for its taxable years ended September 30, 1986, and*60 September 30, 1987, with the Service center in Austin, Texas. Effective December 15, 1988, V&E terminated all of the existing plans and their related trusts, as pursuant to section 401(a)(26) they were to lose their qualified status effective for plan years beginning after December 31, 1988. On April 25, 1990, and April 15, 1991, the Houston, Texas, office of the Service mailed the TMP notices of FPAA with respect to the taxable years ended September 30, 1986, and September 30, 1987, respectively. On June 8, 1990, and June 17, 1991, the TMP filed a petition for readjustment of partnership items under section 6226 with respect to the adjustments set forth in the 1986 and 1987 notices of FPAA, respectively. *13 Law Each V&E plan was an individual defined benefit plan. A defined benefit plan provides a participant at retirement with the benefit stated in the plan. The costs of benefits payable from such plans are funded incrementally on an annual basis over the preretirement period. Secs. 404, 412. Contributions made to the plans, within certain limits, are deductible. Sec. 404(a)(1). V&E funded each plan with money furnished by each partner and passed through the*61 deduction for such contributions to each partner. Earnings on the contributions are not taxed as they accumulate. Sec. 501(a). Plan assets are taxed to participants only as they are paid out as benefits. Sec. 402(a)(1). The amount estimated to fund a defined benefit plan is calculated by the plan's actuary and is determined based upon actuarial assumptions about a number of future events, such as rates of return on investments, the benefit commencement date, future earnings, and participant mortality, among other things. Section 404(a)(1)(A) provides that the amount deductible by a taxpayer in any year shall be determined using the actuarial assumptions that meet the requirements of section 412. At issue in these cases is whether the assumptions made by the plans' actuary, in determining the amount of deductible contributions for each plan for plan years 1986 and 1987, were reasonable in the aggregate and represented the actuary's best estimate of anticipated experience under the plans, as required by section 412(c)(3). Section 412(c)(3) is part of a statutory scheme designed to regulate pension plans, and provided as follows: (3) Actuarial assumptions must be reasonable. *62 -- 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 which, in the aggregate, are reasonable (taking into account the experience of the plan and reasonable expectations) and which, in combination, offer the actuary's best estimate of anticipated experience under the plan. Section 6059 requires that the administrator of a pension plan to which section 412 applies must enlist the services of an enrolled actuary to file an actuarial report for the plan's first year and every third year thereafter. This actuarial report shall contain, inter alia, the actuary's funding methods and assumptions deemed necessary to fund the plan's *14 costs and the actuary's certification that to the best of his knowledge the report is complete and accurate and that he has complied with the requirements of section 412(c). Sec. 6059(b). Section 7701(a)(35) defines an enrolled actuary as a person who is enrolled by the Joint Board for the Enrollment of Actuaries (joint board) established under ERISA. Sections 412(c)(3), 6059, and 7701(a)(35) were added to the Internal Revenue*63 Code by ERISA sections 1013(a), 1033(a), and 3043, 88 Stat. 914, 947, and 1003, to apply for plan years beginning after December 31, 1975. ERISA had a significant impact on pension plans, and one of its objectives was to place reliance upon enrolled actuaries to assist plan fiduciaries in ensuring that pension plans would be able to provide earned benefits to participants when due. ERISA was enacted, in part, because the growth and development of the private pension system in the United States had been substantial since the forties. The legislative history of ERISA provides: In 1940, an estimated four million employees were covered by private pension plans; in 1950, the figure had increased to almost 10 million and in 1960 over 21 million were covered. Currently, over 30 million employees or almost one half of the private non-farm work force are covered by these plans. This phenomenal expansion of coverage has been matched by an even more startling accumulation of assets to back the benefit structure. Today, in excess of $ 150 billion in assets are held in reserve to pay benefits credited to private plan participants. [H. Rept. 93-533 (1973), 1974-3 C.B. 210, 212.] *64 However, regulation of the private pension system's scope and operation had been minimal. Although there have been significant legislative changes since the present basic framework of the tax laws relating to pensions was first adopted -- principally in allowing self-employed people to establish retirement plans for themselves and their employees and in requiring the disclosure of information regarding welfare and pension plans -- it has been more than 30 years since these basic pension provisions were first adopted. It is time for new legislation to conform the pension provisions to the present day situation and to provide remedial action for the various problems that have arisen in the retirement plan area during the past three decades. [H. Rept. 93-807 (1974), 1974-3 C.B. (Supp.) 236, 249.] The enactment of ERISA was a legislative attempt to assure equitable and fair administration of pension plans and to *15 remedy problems that had arisen which prevented many of those plans from achieving their full potential as a source for retirement income. One of the most important matters of public policy facing the nation today is how to assure that*65 individuals who have spent their careers in useful and socially productive work will have adequate incomes to meet their needs when they retire. This legislation is concerned with improving the fairness and effectiveness of qualified retirement plans in their vital role of providing retirement income. In broad outline, the objective is to increase the number of individuals participating in employer-financed plans; to make sure to the greatest extent possible that those who do participate in such plans actually receive benefits and do not lose their benefits as a result of unduly restrictive forfeiture provisions or failure of the pension plan to accumulate and retain sufficient funds to meet its obligations; and to make the tax laws relating to qualified retirement plans fairer by providing greater equality of treatment under such plans for the different taxpayer groups concerned. [H. Rept. 93-807 (1974), 1974-3 C.B. (Supp.) 243.] The enactment of ERISA mandated a cooperative effort whereby the Department of Labor (Labor) and the Department of the Treasury (Treasury) would jointly oversee the participation in and vesting and funding of pension plans, *66 the reporting and disclosure requirements for pension plans, fiduciary standards, and plan termination insurance. In enacting ERISA, Congress was well aware that actuaries would play a major role in ensuring that retirement plans would be sufficiently able to provide retirement income when due. Throughout the reports of the Senate Finance Committee and the House Ways and Means Committee, Congress recognized the importance of the assumptions and methods chosen by actuaries in determining plan funding amounts. Congress also explicitly noted that such determinations by actuaries would involve making predictions and would be a matter of judgment involving many factors and producing a range of results. In estimating pension costs, actuaries must make assumptions ("actuarial assumptions") about a number of future events, such as the rate of return on investments ("interest"), employees' future earnings, and employee mortality and turnover. Actuaries also must choose from a number of methods to calculate future plan liabilities. The amounts required to fund any given pension plan can vary significantly according to the mix of these actuarial assumptions and methods. As a result, *67 the assumptions and methods used by actuaries are basic to the application of minimum funding *16 standards for defined benefit pension plans. [S. Rept. 93-383 (1973), 1974-3 C.B. (Supp.) 80, 147.] See also H. Rept. 93-807 (1974), 1974-3 C.B. (Supp.) 262. Congress realized that making predictions and forecasts to determine plan funding costs was the professional duty and function of actuaries. [The House Ways and Means] committee recognizes that the amount required to fund a pension plan is in large part determined by actuaries' estimates of future plan costs, which in turn are based on the actuarial methods and assumptions used for each plan. Consequently, the determination of the amount of contributions that must be made to a plan to adequately fund the plan benefits is significantly affected by the professional decisions of the plan's actuary. * * * [H. Rept. 93-807 (1974), 1974-3 C.B. (Supp.) 309.] Accordingly, Congress decided that accepting a range of reasonableness for funding amounts for retirement plans would be more desirable and more effective than imposing an inflexible legislative*68 standard on actuaries and, therefore, rejected imposing mandatory funding assumptions and methods. Congress acknowledged that actuaries are not charged with the responsibility of determining a "correct" set of assumptions, but with the responsibility of determining assumptions that fall within a range of reasonableness. Conceivably an attempt might be made to secure uniform application of the minimum funding standards by authorizing the Secretary of the Treasury or some other authority to establish the specific actuarial assumptions and methods that could be used by pension plans. That would involve, for example, setting a specific rate of interest that could be used by certain pension plans or by specifying certain turnover rates for specified types of firms. However, the committee does not believe that this would be an appropriate procedure, since the proper actuarial assumptions may differ substantially between industries, among firms, geographically, and over time. Further, in estimating plan costs each actuarial assumption may be reasonable over a significant range and it would appear that the proper test would be whether all actuarial assumptions used together are reasonable. *69 These considerations strongly indicate that any attempt to specify actuarial assumptions and funding methods for pension plans would in effect place these plans in a straitjacket so far as estimating costs is concerned, and would be likely to result in cost estimates that are not reasonable. [H. Rept. 93-807 (1974), 1974-3 C.B. (Supp.) 262.] By deferring to the professional judgment of actuaries in the area of plan funding, Congress expressed its intent that actuaries be afforded a range of latitude in establishing a *17 mix of reasonable assumptions, and that only if such assumptions are "substantially unreasonable" should they be challenged and changed retroactively. 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. Rept. 93-807 (1974), *70 1974-3 C.B. (Supp.) 330.] The language of the conference report adopted the rules relating to actuarial assumptions of the House Ways and Means bill, as amended by the Senate amendment, and the result was as follows: The conference substitute combines the rules relating to actuarial assumptions of the House bill and of the Senate amendment and requires that, for purposes of the minimum funding standard, all plan costs, liabilities, rates of interest, and other factors under the plan are to be determined on the basis of actuarial assumptions and methods which, in the aggregate, are reasonable. Actuarial assumptions are to take into account the experience of the plan and reasonable expectations. These assumptions are expected to take into consideration past experience as well as other relevant factors. In addition, under the conference substitute, the actuarial assumptions in combination are to offer the actuary's best estimate of anticipated experience under the plan. * * * [H. Conf. Rept. 93-1280 (1974), 1974-3 C.B. 415, 445-446.] The conference report expresses the intent of Congress to place with enrolled actuaries the *71 responsibility for arriving at reasonable assumptions regarding plan funding costs. These assumptions are to be based, inter alia, objectively on a plan's past experience and on the subjective professional judgment of an enrolled actuary regarding potential future plan experience. The actuary's assumptions should not be challenged and changed retroactively unless substantially unreasonable. We note that the requirement that assumptions used to determine plan funding costs be reasonable predates ERISA. The regulations accompanying section 404 state: In no event shall costs for the purpose of section 404(a)(1) exceed costs based on assumptions and methods which are reasonable in view of the provisions and coverage of the plan, the funding medium, reasonable *18 expectations as to the effects of mortality and interest, reasonable and adequate regard for other factors such as withdrawal and deferred retirement (whether or not discounted) which can be expected to reduce costs materially, reasonable expenses of operation, and all other relevant conditions and circumstances. * * * [Sec. 1.404(a)-3(b), Income Tax Regs.] Also, Rev. Rul. 63-11, 1963-1 C.B. 94,*72 states the Service's position with respect to various methods of valuing assets and the proper interest rate to be used in valuing liabilities for computing costs in determining the limitations on deductions for employer contributions to pension plans under section 404. Respondent stated in the revenue ruling: However, it should be emphasized that, for the purpose of determining deductible limits, it is not essential that each individual assumption used be reasonable. It is merely required that the combination of all assumptions produces reasonable results. * * * [Rev. Rul. 63-11, 1963-1 C.B. 94, 95.] Although the legislative history of ERISA evidences congressional intent to rely exclusively on actuarial assumptions regarding plan funding costs -- such assumptions will be accepted if reasonable in the aggregate -- Congress did not permit actuaries unfettered liberty. To do so might invite the unprofessional selling of expertise to achieve tax-desired results rather than prudent plan funding. However, instead of imposing uniform and specific assumptions for pension plans, Congress decided to impose uniform standards for the qualification*73 of actuaries. Congress recognized that there was no existing Government regulation of the actuarial profession as there was for other professions such as the medical, legal, and accounting professions. The Committee [on labor and public welfare] is unaware of any significant licensing procedures for actuaries at either the state o[r] federal level and this may, to some extent, explain inadequate funding procedures which have been found to exist. * * * [S. Rept. 93-127 (1973), 1974-3 C.B. (Supp.) 17.] Accordingly, Congress directed the Secretaries of Labor and of Treasury to establish a joint board which would create and administer standards for enrollment of actuaries; enroll actuaries to practice before Labor and the Service; administer standards for disenrollment of actuaries; and write the appropriate regulations. See H. Conf. Rept. 93-1280 (1974), 1974-3 C.B. 523. By creating a system of actuarial certification, *19 Congress could be satisfied that actuaries who performed services for qualified pension plans would meet an appropriate level of competence in choosing their funding methods and assumptions. The *74 joint board is composed of three members appointed by the Secretary of the Treasury and two appointed by the Secretary of Labor. 20 C.F.R. sec. 900.3 (1991). ERISA section 3042(a), 88 Stat. 1002, provides that individuals performing actuarial services and applying for enrollment before the joint board on or after January 1, 1976, must satisfy the following standards and qualifications: (1) education and training in actuarial mathematics and methodology, as evidenced by -- (A) a degree in actuarial mathematics or its equivalent from an accredited college or university, (B) successful completion of an examination in actuarial mathematics and methodology to be given by the Joint Board, or (C) successful completion of other actuarial examinations deemed adequate by the Joint Board, and (2) an appropriate period of responsible actuarial experience. * * * The Federal regulations governing the performance of actuarial services under ERISA prescribe rules for renewal of enrollment. Those regulations provide that as a condition for renewal, an individual enrolled to perform actuarial services under ERISA must periodically complete a minimum number of hours of continuing education courses. *75 The content of such courses must be "designed to enhance the knowledge of an enrolled actuary with respect to matters directly related to the performance of pension actuarial services under ERISA or the Internal Revenue Code." 20 C.F.R. sec. 901.11(f) (1991). Examples of the subject matter of such courses are: Actuarial cost methods under ERISA, requirements with respect to the valuation of plan assets, requirements for qualification of pension plans, excise taxes related to the funding of qualified pension plans, economics, computer programs, pension accounting, investment and finance, and business and general tax law. Id. ERISA section 3042(b), 88 Stat. 1003, provides that the joint board may, after notice and an opportunity for a hearing, suspend or terminate the enrollment of an individual who fails to discharge his duties or does not satisfy the requirements for enrollment. *20 Only persons who have satisfied the standards and qualifications for becoming an enrolled actuary may determine plan funding methods and assumptions under ERISA. ERISA requires that the administrator of a defined benefit plan retain on behalf of all plan participants an enrolled actuary*76 to prepare an actuarial statement to be filed annually with the Secretary of Labor. ERISA sec. 103(a)(4)(A), 88 Stat. 842 (current version at 29 U.S.C. sec. 1023(a)(4)(A) (1988)). This actuarial statement is to contain, among other things, information pertaining to the plan year such as the normal costs, accrued liabilities, actuarial assumptions and methods used to determine costs, and the value of assets accumulated in the plan. The enrolled actuary also is required to make an actuarial valuation of the plan for every third plan year, or more frequently if deemed necessary. ERISA sec. 103(d), 88 Stat. 845, 846 (current version at 29 U.S.C. sec. 1023(d) (1988)). Section 6059, as added by ERISA section 1033, 88 Stat. 947, requires each defined benefit plan to which section 412 applies to file for the first plan year and for each third plan year thereafter (or more frequently if the Secretary of the Treasury determines), a report prepared and signed by an enrolled actuary. Those reports shall contain: (1) a description of the funding method and actuarial assumptions used to determine costs under the plan, *77 (2) a certification of the contribution necessary to reduce the accumulated funding deficiency (as defined in section 412(a)) to zero, (3) a statement -- (A) that to the best of his knowledge the report is complete and accurate, and (B) the requirements of section 412(c) (relating to reasonable actuarial assumptions) have been complied with, (4) such other information as may be necessary to fully and fairly disclose the actuarial position of the plan, and (5) such other information regarding the plan as the Secretary may be regulations require. [Sec. 6059(b)(1)-(5).] In summary, the statutory scheme for satisfying the provisions of section 412(c)(3) requires that an administrator of a pension plan engage the services of an enrolled actuary who determines the methods and assumptions necessary to fund costs under the plan and certifies that to the best of his knowledge such methods and assumptions are reasonable in *21 the aggregate and represent his best estimate of anticipated experience under the plan. Upon audit by respondent, these actuarial assumptions may be determined to be unreasonable in the aggregate, and V&E has the burden of proving that the assumptions are*78 reasonable in the aggregate. However, these actuarial assumptions will not be changed retroactively unless they are found to be substantially unreasonable. Accordingly, we shall now examine the various challenged assumptions. Experts The following experts, whose reports and testimony will be discussed later in this opinion, appeared on behalf of V&E: James F. Rabenhorst (Rabenhorst), B.S., M.B.A., testified as to the validity of V&E's retirement age assumptions established by the plans' actuary. Rabenhorst is the managing partner of the national law firm consulting group at the accounting and consulting firm of Price Waterhouse. He also is chairman of the law firm services group at Price Waterhouse, a post he has held since 1979. He has been a partner at Price Waterhouse since 1976, has been employed by Price Waterhouse since 1968, and has consulted with prominent law firms on a variety of management issues, including retirement policy, since 1969. Rabenhorst also is a member of the Economics of Law Practice Section of the American Bar Association and has written two monographs in that capacity. Richard R. Joss (Joss), B.S., M.A., Ph.D., F.S.A., 1 M.A.A.A., 2*80 E.A., 3*79 testified with regard to the validity of all four actuarial assumptions. He has more than 16 years' experience as an actuarial consultant to pension funds, the *22 last 11 years with the Wyatt Company (Wyatt), which he joined in 1981 as a cofounder of Wyatt's Seattle office. In August 1990, Joss was promoted to resource actuary in Wyatt's Washington, D.C., office. The resource actuary serves as the source for technical and practical advice in dealing with pension laws and regulations for each of the approximately 350 to 400 actuaries that Wyatt employs nationwide. He has coauthored a textbook on the application of mathematics to solving business problems. Mary S. Riebold (Riebold), F.S.A., M.A.A.A., E.A., F.C.A., 4 testified as to the validity of all four actuarial assumptions established by the plans' actuary. Riebold has been a managing director of William M. Mercer, Inc., the world's largest actuarial consulting firm, since 1988, where her practice has focused exclusively on pension plans. She has worked in employee benefits for 23 years and in pensions for 17 years. Additionally, she has been a member of the pension section council of the Society*81 of Actuaries since 1989, a member of the board of governors of the Society of Actuaries since 1990, a member of the board of directors of the American Academy of Actuaries since 1990, and a member of the board of directors of the Conference of Consulting Actuaries since 1984. Riebold served as treasurer of the Conference of Consulting Actuaries from 1987 to 1990 and has been president of that organization since October 1991. Steven H. Schechter (Schechter) earned a certificate degree in data processing and system design from Columbia University in 1982. He was employed at Bankers Trust until 1985 when he began working at Wolper Ross as business applications systems manager. In 1987, Schechter was promoted to director of management information systems, which involves *82 the analysis of computer data on a regular basis. Schechter offered testimony as an expert in computer systems business applications, providing analyses of Labor Form 5500 5 data concerning interest rate assumptions established for small *23 pension plans (those with fewer than 100 participants and those with 10 or fewer participants). John W. Peavy III (Peavy), B.B.A., 6 M.B.A., Ph.D., C.F.A., 7 testified on behalf of V&E as a rebuttal witness. Peavy is professor of finance and Vaughan Roucher Professor of Financial Investments at Southern Methodist University. He was engaged in investment banking with Goldman, Sachs & Co. for 6 years and at present manages $ 200 million of clients' funds through his own firm, Diversified Investment Services. *83 He has published several articles and books in the area of finance and has been retained in the past by the Service as an expert witness. He offered testimony to rebut the contentions of respondent's experts, Shapiro and Haneberg, regarding the interest rate assumptions established by Smith, the plans' actuary. The following experts, whose reports and testimony will be discussed later in this opinion, appeared on behalf of respondent: Ronald L. Haneberg (Haneberg), B.A., J.D., F.S.A., M.A.A.A., F.C.A., offered testimony as to the validity of all four actuarial assumptions established by Smith, the plans' actuary. Haneberg has been a self-employed consulting actuary since 1988, prior to which he was a consulting actuary with Buck Consultants*84 (Buck), a prominent pension actuary firm, from 1977 to 1988. His experience with Buck was mostly with multiple-participant plans, the smallest of which might have included between 10 and 50 participants, not individual plans. In preparation for his testimony, Haneberg read "literally tens of thousands of pages of * * * [actuarial] materials and literature" on small plans. Claude Poulin (Poulin), B.A., B. Comm., F.S.A., M.A.A.A., E.A., testified as to the validity of all four of V&E's actuarial assumptions established by the plans' actuary. He has been the president of Poulin Associates, Inc., an actuarial and benefits consulting firm based in Washington, D.C., since January 1980. His clients include individual plans, as well as those of many large corporations and labor unions. From 1976 to 1980, Poulin was assistant director of the Social *24 Security department of the United Auto Workers union, where he had worked since 1969. Alan C. Shapiro (Shapiro), B.S., M.A., Ph.D., testified as to the validity of the interest rate assumptions as an expert in finance and economics, not actuarial science. Shapiro is the Ivadelle and Theodore Johnson Professor of Banking and Finance, *85 University of Southern California (USC). He has been a professor at USC for 14 years, has previously taught at The Wharton School of the University of Pennsylvania, and has been a visiting professor at several universities worldwide. Shapiro conducts seminars in finance for executives, consults with several large industrial and service companies, and serves as a federally appointed director of Lincoln Savings & Loan, a failed thrift institution. He has written four books, including a leading corporate finance textbook, and more than 50 articles in scholarly journals. William S. Borden (Borden), B.A., Ph.D., testified in rebuttal to Joss' testimony regarding actual rates of investment return for small and large plans relying on Form 5500 data obtained from Labor. Borden is a senior program analyst at Mathematica Policy Research (Mathematica), a position he has held since 1979. Mathematica conducts statistical evaluations of the effectiveness of Federal and State programs. In 1984, Borden was retained by Labor to create a research database of Form 5500 data from the original filings. Since that date, nearly all of his professional time has been devoted to analyzing Form 5500*86 data. Jeffrey F. Jaffe (Jaffe), B.A., M.B.A., Ph.D., offered an expert report on behalf of respondent as to the validity of the interest rate assumption established by the plans' actuary; however, he did not testify. Jaffe has been an associate professor of finance at The Wharton School of the University of Pennsylvania since 1980. He is the author of more than 20 articles in scholarly journals and other publications. Interest Rate Background Smith, the plans' certifying actuary, established an interest rate assumption of 5 percent for both the pre and postretirement periods and used that rate to calculate the contributions *25 for the years at issue. Respondent determined that an 8-percent interest rate pre and postretirement assumption should be used for V&E's plans for the years at issue. As with all of the actuarial assumptions, there is no single formula or set of criteria used by actuaries to establish interest rate assumptions. Actuaries consider the facts and circumstances of each plan and use their experience and subjective judgment, all based on actuarial criteria, in establishing this and all of the actuarial assumptions. The phrase "interest rate" assumption*87 is more accurately described as an "investment rate" assumption as it represents the expected aggregate earnings, gains, and losses on all investment vehicles held in trust for the plan during its existence. As a matter of convention, however, this aggregate return concept is referred to usually in the actuarial and pension community as an "interest rate" assumption, and we will refer to it here as such. In general, a defined benefit pension plan is established and maintained by an employer (plan sponsor) who systematically funds a pension plan trust from which a trustee (plan administrator) will make payments of definitely determinable benefits to participating employees (participants) over a period of years, usually for life. The definitely determinable benefits usually take the form of an annuity commencing at retirement and payable over the life of the participant or the joint lives of the participant and his or her spouse. For each participant, these definitely determinable benefits are measured by, and based on, factors such as years of service with the plan sponsor, compensation level, and other factors, all of which are incorporated into a benefit formula. The benefit*88 formula determines what the plan's ultimate benefit obligation will be. However, since many, if not all, benefit formula factors are unknown at the plan's inception, and some are unknown until benefit payout ceases under the plan, the amount of the benefit obligation is indeterminate and will be so until it is satisfied. Thus, the plan's ultimate benefit obligation is an estimate. The pension trust from which benefits will be paid accumulates assets from two sources: contributions from the plan sponsor during the preretirement period and earnings on the trust assets during both pre and postretirement periods. The interest rate assumption serves to predict the rate at which *26 the trust will self-generate growth as opposed to increasing due to the sponsor's annual contributions. As noted above, earnings on the trust assets will accumulate tax free. Secs. 401(a), 501(a). The interest rate assumption is used by the actuary to predict the annual rate of return on assets in the trust over the life of the plan. This assumption is then used along with all of the other actuarial assumptions to calculate the required contributions. The actuarial assumptions do not affect the*89 ultimate pension benefits paid under the plan or actual aggregate cost of those benefits. Rather, the assumptions affect the pattern of asset buildup in the pension trust because they impact the timing and amount of the sponsor's contributions. Actuarial assumptions can be either conservative or optimistic. Conservative assumptions about future events affecting funding, such as lower estimated fund earnings and earlier retirement, tend to create higher required contributions. Optimistic assumptions tend to create lower required contributions. Sponsors of qualified defined benefit pension plans must make timely contributions sufficient to assure that the plans' estimated benefit obligations are satisfied ultimately in accordance with the plans. Sec. 412. The size of the annual contribution will vary under a defined benefit plan, depending on the actuarial cost method chosen and the actuarial assumptions made, but is always actuarially calculated with the objective of having the pension trust attain a funding level sufficient to satisfy the pension obligations as they come due. The enrolled actuary also attempts to choose an actuarial method and assumptions that will provide*90 for a smooth funding pattern over the life of a plan. The actuarial cost method used by Wolper Ross with respect to V&E's IDB plans is not at issue in these cases. The actuary's primary duty to plan participants under the statutory scheme established by Congress is to establish a realistic contribution pattern over the long term so that the plan sponsor will provide adequate funding for the ultimate pension obligation. Thus, in selecting actuarial assumptions, prudent actuaries maintain a long-term conservative view that will ensure benefit security for plan participants. All of the actuaries who testified at trial confirmed this basic characteristic of actuarial science. Once a plan has established "credible experience", selecting actuarial assumptions *27 becomes a less worrisome task in this regard; the actuary can take comfort that this actual experience has established a reliable trend, and such experience is then useful in predicting future plan experience. All of the actuarial experts generally agreed that a plan should sustain at least 3 years, and as much as 5 years, of plan experience before such experience can be considered "credible experience" for purposes*91 of significantly impacting the establishment of actuarial assumptions for the plan. See Rev. Rul. 63-11, 1963-1 C.B. 94, 96 (reasonableness of the originally established actuarial assumptions should be reassessed after a plan has been in effect for a period of about 5 years). Cf. Jerome Mirza & Associates, Ltd. v. United States,692 F. Supp. 918 (C.D. Ill. 1988), affd. 882 F.2d 229 (7th Cir. 1989) (reasonableness of actuarial assumptions reviewed where 73 percent of pension benefit obligation was funded in the first year of plan). Conversely, where the plan is new or lacks such credible experience, the actuary must be cautious in selecting actuarial assumptions that reflect only recent economic or plan experience. Also, general or even specific participant statements of intended investment strategy or even an expressed intention to obtain a specific investment does not provide a reliable basis on which an actuary can base the choice of interest rate assumption. Such intentions often fail to be carried out, and the choice of investment vehicles as well as economic conditions can*92 vary widely in the short run. Experience under the plan should be sustained before it becomes a primary basis for selecting assumptions. This fact is recognized by the Interim Actuarial Standards Board of the American Academy of Actuaries, which recommends as follows: The actuarial assumptions in combination should reflect the actuary's best judgment of future events affecting the related pension obligations. The actuary should consider the actual experience of the covered group but should emphasize expected long-term future trends rather than give undue weight to recent past experience. * * * [Actuarial Standard of Practice No. 4, Recommendations for Measuring Pension Obligations, sec. 7.1 (Interim Actuarial Standards Board 1990 reprint).] In addition, selecting actuarial assumptions that are too optimistic in a new or relatively young plan adversely affects the ability of the plan to "self-correct" without significantly increasing the required contributions. A plan will actuarially *28 self-correct in a mathematical sense through the annual actuarial contribution calculation: the contribution in a subsequent plan year will be higher or lower than originally anticipated*93 for that year to correct for prior years' actual experience under the plan. When the assumptions are too optimistic, future contributions will have to be higher than originally planned. This disrupts not only the desirable smooth funding pattern, but burdens the plan sponsor to provide higher than anticipated contributions. 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 chances that the sponsor may ultimately fail to fully fund the plan. It is common knowledge that many pension funds have fallen prey to just such a fate with the Pension Benefit Guarantee Corp. having to cover the resulting losses. 8 *94 Moreover, if the assumptions remain consistently too optimistic, even for a relatively brief number of years, asset accumulation suffers further because investment income fails to accrue on contributions that, in hindsight, should have been made earlier. This loss of compounded earnings can be a significant factor, especially in tax-exempt trusts. In addition, the impact of and the risk associated with failing to earn the anticipated rate of return is magnified in small plans where actuarial losses cannot be spread among many participants. The Interim Actuarial Standards Board of the American Academy of Actuaries recognizes this fact in guiding actuaries in establishing assumptions for small plans: In a plan where much of the obligation is associated with a small number of participants, special attention should be paid to potential fluctuations and the aggregate effect of the assumptions. * * * [Actuarial Standard of Practice No. 4, Recommendations for Measuring Pension Obligations, sec. 7.3 (Interim Actuarial Standards Board 1990 reprint).] If actual experience for an assumption, or assumptions, varies significantly from the predicted results so that a consistent pattern*95 of gains or losses emerges, the enrolled actuary *29 will adjust the actuarial assumptions accordingly. 9 Also, any actuarial gains are amortized over time with a corresponding reduction in contributions. Sec. 412(b)(3)(B)(ii). Addressing the actuarial experts who testified at trial, we first note that they were well qualified and helpful to the Court. As we discuss below, their individual actuarial experience, which influences an actuary's professional judgment, differed. We note that while different actuaries will use different starting points and each will have his or her own experiences from which to draw, there are several commonalities from which all actuaries apparently draw and consider in establishing an interest rate assumption: (1) Long-term historical economic trends and investment rates of return; (2) the economic*96 components inherent in interest rates; (3) the current rates on long-term investments; and (4) actuarial and plan specific factors for which benchmark rates or ranges of rates are adjusted. In addition, all of the actuarial experts recognized that there will be a range of reasonable rates for any given set of circumstances under which actuarial assumptions are established. Petitioner's Actuarial Experts V&E's expert Joss approached the interest rate assumption question in two parts, first determining an acceptable range of rates, and then determining what a "best estimate" would be within that acceptable range. Joss used four widely accepted methods of calculating assumed interest rates, using the high and low from each to establish an acceptable range. He examined historical trends for the last 20, 30, and 40 years for four different types of investments (common stocks, corporate debt, and short- and long-term Government debt), finding the highest rate of return, 11.8 percent, to be for common stocks bought between 1947 and 1987, and the lowest rate of return, 4.2 percent, to be for long-term Government bonds bought between 1946 and 1986. Thus, he established a range*97 of acceptable rate assumptions of between 4.2 percent and 11.8 percent. The three components of the interest rate are: (1) The risk-free real rate of return; (2) an inflation factor; and (3) a risk *30 premium for lower grade or illiquid investments. Joss examined these three components over 60-, 40-, and 20-year periods, and arrived at a risk-free rate of return averaging about 0.6 percent, an inflation rate averaging about 3 percent (taking into account the fact that in 1986 inflation was widely expected to continue its downward trend), and a risk premium ranging from 0.7 percent for corporate bonds to 5 percent for equities. By aggregating all of these figures, Joss arrived at an acceptable range of interest assumptions extending from a relatively risk-free 3.6 percent to a much higher risk 10 percent. In addition, Joss examined the then-current (1986 and 1987) yields on 20- and 30-year Treasury obligations, finding both types of instruments to be yielding between 7 percent and 9 percent. He added, however, that current yields on Treasury bonds are rarely indicative of longer term interest rates, noting that there is very little short-term variation but significant longer*98 term volatility. For example, he stated that Treasury bond rates in 1979 were 5 percentage points higher than they were in 1986, only 7 years later. Last, Joss examined outside economic data, in this case data provided by the Office of Management and Budget, which forecast that the 1986 8.9-percent yield on 10-year Treasury notes would fall to 4.5 percent by 1991. Thus, an actuary would be justified in assuming an interest rate between 4.5 percent and 8.9 percent. Based on the above analysis, Joss concluded that an interest rate assumption between 3.6 percent and 11.8 percent would have been reasonable. After determining that range, Joss narrowed it by enumerating several factors that an actuary might use to determine a single interest rate assumption. First, an actuary should take into account the standards of the actuarial community. Joss stated that a 5-percent interest assumption was in 1986 "within the actuarial mainstream." 10 Second, the asset base of a new plan is much less stable than that of a well-established plan (i.e., one with substantial funding); thus, a more conservative interest assumption is warranted. *31 Third, plans with a smaller asset base and*99 number of participants, such as IDB plans, will earn substantially less, generally 2.5 percent to 3 percent less, than large plans that can take advantage of economies of scale and professional management. 11 Fourth, long-term interest rates cannot be extrapolated with impunity from current rates. Fifth, a lower interest rate assumption with a higher actual rate of return scenario is preferable to a high assumption and low return scenario in that the latter situation leads to severe instability regarding plan funding needs. *100 V&E's expert Riebold concluded, like Joss, that 5 percent was reasonable. She stated that four factors should guide actuarial interest rate assumptions. Initially, like all the other actuarial experts, she affirmed that there is no single correct assumption; a range of reasonable assumptions should be established. Riebold also concluded that small self-directed investment plans, such as IDB plans, will have lower rates of return than large plans or published indices. Reasons for this include inability to make use of economies of scale, often volatile and ill-timed personal investment decisions by the plan participant, and the possible increased need for liquidity in a personal plan. In addition, she noted the eighties have been an atypical investment cycle and, like any brief segment of past plan experience -- especially the first years of plan experience -- cannot be considered to be determinative of future experience. Finally, with respect to the long-term nature of the interest rate assumption, Riebold explained that predictions are bound to be imprecise. Riebold also concluded that the "select and ultimate" approach (where different rates are selected for succeeding *101 periods) would be too costly for a small plan to employ. She also discounted the notion that if, for example, a 20-year bond were available today earning 8 percent, and the preretirement period was also 20 years, 8 percent would be a good predictor of the rate of return. Such analysis assumes that no future contributions would be made, no bonds would mature or be called during the 20 years, and benefit payments would *32 match investment income exactly. Riebold stated that long-term bond rates cannot be direct predictors of future rates given the realities that: (1) Future contributions will be made to the plan at varied interest rates; (2) bonds eventually mature; (3) bonds may be called early; and (4) benefit payments may not match interest income. Using the facts available in 1986, Riebold constructed a model by which a range of interest rate assumptions might be derived. Her model showed an acceptable interest rate assumption for 1986 of between 5.03 percent and 6.59 percent, based on aggregate economic information including interest rate projections made in 1986 by the Council of Economic Advisors (CEA). In arriving at this range, Riebold projected a 20-year funding*102 period (e.g., a 35-year-old participant retiring at age 55) and a 10-year maturity on all investment instruments purchased, and assumed reinvesting all interest payments. The CEA in 1986 projected an interest rate on 10-year Treasury notes declining from 7.5 percent in 1987 to 4.5 percent in 1991. In addition to economic analysis, Riebold, also utilized what she referred to as the "more intuitive form of analysis" of Ira Cohen (Cohen), who in 1986, as the then-director of the Actuarial and Technical Division of the Service, stated to a meeting of enrolled actuaries that a 4-percentage-point corridor on either side of the prevailing long-term Treasury bond rate was within the reasonable range of interest rate assumptions. In 1986, the rate for long-term Treasury bonds was 7.5 percent; thus, Cohen gave his imprimatur to an interest rate assumption potentially as low as 3.5 percent, much more conservative than the 5 percent in question. Riebold testified that while Cohen's statement was not determinative, it was an additional consideration an actuary would have considered in 1986 in establishing an actuarial interest rate assumption. Riebold stated that all of the same factors *103 should guide an actuary in his or her consideration of postretirement interest rate assumptions as guided him or her in making preretirement assumptions. She added that although it is common practice to use the same rate for both the pre and postretirement periods, postretirement assumptions are even more uncertain than preretirement assumptions, given the longer period for prediction (participant's and spouse's joint life *33 expectancy), warranting yet further conservatism. In addition, since current bond rates are not determinative of bond rates 20 years in the future, Riebold concluded that the CEA estimate of a post-1991 bond rate of 4.5 percent was a useful predictor here, further indicating that a 5-percent interest rate assumption was "eminently reasonable". Respondent's Actuarial Experts Respondent's expert Haneberg began his analysis with an identification of the three types of assets in which pension funds are generally invested: Corporate equities, fixed income securities (both corporate and Government), and cash equivalents. Within each of these three categories of assets, Haneberg determined the long-term (20 year), intermediate (10 year), and current (1*104 year) average yields of "passive" or low-maintenance investments (low-maintenance), as well as the average yields for "managed" or high-maintenance investments (high-maintenance) during the same periods. He made his calculations based on information available in 1984, the year that many of V&E's plans were established. Using the Standard & Poor's 500 Stock Composite Index as a measure of the broad market of equities in the United States, Haneberg calculated that long-term, net low-maintenance and net high-maintenance rates of return for equities were 7.69 percent and 7.97 percent, respectively. Intermediate low-maintenance and high-maintenance returns were determined to be 10 percent and 10.28 percent, respectively. Current yields for equities were considered actuarially irrelevant by Haneberg, given the volatility and short-term fluctuations of equity markets. Haneberg based his low-maintenance fixed income assumption on 20- and 30-year maturity U.S. Treasury bonds, finding long-term, intermediate, and current net low-maintenance yields of 7.74 percent, 9.96 percent, and 11.39 percent, respectively. To calculate a net high-maintenance yield, Haneberg included yields from high-grade*105 corporate and Government agency bonds. Haneberg stated that a high-maintenance bond portfolio should perform 1.25 percent better than one consisting of low-maintenance Treasury obligations and, after subtracting a management fee, found long-term, *34 intermediate, and current managed yields of 8.34 percent, 10.56 percent, and 11.99 percent, respectively. Low-maintenance cash equivalents, according to Haneberg, are 3-month Treasury bills, the average annual long-term, intermediate, and current yield of which was calculated by him to be 7 percent, 8.97 percent, and 8.92 percent, respectively. Haneberg used the return on money market funds as a proxy for the return on high-maintenance cash equivalents, which would include commercial paper, other short-term non-Government debt, and certificates of deposit. While a long-term figure was not available, intermediate and current yields for these instruments were calculated by Haneberg to be 9.6 percent and 8.78 percent, respectively. Using the above figures, Haneberg concluded that in 1984, the year of V&E's plans' inception, a conservative actuary might have legitimately assumed a 7.5-percent interest rate, and a more optimistic *106 actuary might have reasonably assumed a 10.5-percent interest rate. An interest rate assumption of 5 percent, as was made by Wolper Ross, would, in Haneberg's opinion, "require significant further justification." The interest rate assumptions in question, however, were made for the 1986 plan year. While Haneberg recognized that actuaries generally do not alter their interest rate assumptions, he stated that, on the basis of several post-1984 contingencies, 12 he would have broadened his acceptable range of interest rate assumptions to include any figure between 7 percent and 11 percent. Respondent's expert Poulin also concluded that a 5-percent interest*107 rate assumption was too conservative. Poulin, who conceded that 2 years of investment experience carries little weight in predicting future investment experience, reached his conclusion in a manner slightly different than Haneberg. Poulin stated that interest rates generally are composed of three separate components as follows: (1) The "pure" rate of interest in an inflation-free environment, which he posited was about 4 percent; (2) the level of anticipated inflation, which he stated has averaged 5.86 percent annually, as measured by the consumer price index over the last 25 years; *35 and (3) a risk premium to compensate investors for the potential loss of capital, which Poulin calculated to be between 1 percent and 2 percent. After having devised this formula for determining interest rates, Poulin stated that, in practice, the easiest means of arriving at an interest rate assumption is to determine the yield of a specific type of investment which: (1) Bears no risk of loss of capital or fluctuation of income; (2) is available at any point in time with varying maturity dates; and (3) can be purchased at low cost. Poulin concluded that such an investment instrument exists*108 -- a Treasury obligation. Given that a defined benefit pension plan is "at least in theory, a long-term undertaking", said Poulin, the 30-year Treasury bond rate would be an appropriate benchmark riskless interest rate. Poulin then stated that with Forms 5500 prepared at the end of 1987 and signed by the enrolled actuary on January 1988, the appropriate interest rate would be that which was prevailing for 30-year Treasury bonds at that point in time, approximately 9 percent. He added, however, that 1 percentage point might be deducted to take into account the possibility of lower interest rates and another percentage point deducted for unforeseen contingencies such as the "unexpected market reaction" of October 19, 1987. After stating that the interest rate on the 30-year Treasury bond provides a suitable benchmark, Poulin described an "objective standard" for determining reasonable minimum interest rate assumptions for pension plans as developed in a proposal by the Pension Section Council of the Society of Actuaries (the society). This objective standard has four steps (only three steps as articulated by Poulin), which are as follows: (1) A current riskless rate (CRR) is *109 defined as the annual yield on 30-year Treasury bonds "at the time contribution limits are determined"; (2) an ultimate riskless rate (URR), an estimate of where the riskless rate may tend over time, is defined, as suggested by the Society, at 5 percent; (3) assets on hand are assumed to earn the CRR, and assets not yet available are assumed to earn the average of the CRR and URR; and (4) the figure derived in calculation (3) is multiplied by 80 percent to derive the minimum reasonable interest assumption, which, according to Poulin, "assures that the minimum reasonable interest rate will be very conservative." Applying this test to plans funded between 30 percent and 50 *36 percent, and using a 9-percent 30-year Treasury bond rate (the approximate yield as of January 1, 1988), Poulin estimated the minimum reasonable interest assumption to be between 6.1 percent and 6.4 percent, and the maximum reasonable interest assumption to be between 9.1 percent and 9.6 percent. Discussion Although the actuarial experts differed in their ultimate conclusions about the reasonableness of the 5-percent interest rate assumption in question here, they all agreed on several basic tenets *110 regarding actuarial assumptions. They agreed, inter alia, that a reasonable range exists for each set of circumstances, that actuarial interest rate assumptions are long-range predictions that become increasingly speculative with the length of the prediction, and that an element of conservatism is appropriate in the selection of actuarial assumptions. They also recognized that the existing experience of new, immature plans without "credible experience" should not guide the actuary in the selection of assumptions, and that the actuary's choice of assumptions is guided by his or her professional experience as well as actuarial analysis. They all employed an analytical approach that involved selecting a suitable starting point derived either from historical economic data or a current economic benchmark, and then adjusting it for various established actuarial factors as well as for those plan specific factors that, in that actuary's judgment, produced the appropriate actuarial reasonable range. Respondent's two actuarial experts, Haneberg and Poulin, concluded that reasonable ranges for V&E's plans during the years at issue were 7 percent to 11 percent. V&E's two actuarial experts*111 concluded 5 percent was reasonable. Joss opined on reasonable ranges of 3.6 percent to 11.4 percent for plan year 1986 and 3.6 percent to 11.8 percent for plan year 1987. Riebold reached a conclusion of "reasonable" for the 5-percent assumption. Joss and Haneberg arrived at their reasonable ranges based on an analysis of historical rates of return on the various types of securities in which the plans might invest. Beginning with their differing conclusions drawn from this extrinsic economic data as a starting point, *37 both actuaries then made adjustments for actuarial considerations. Riebold and Poulin arrived at their reasonable ranges by choosing a benchmark rate current or proximate to the period at issue, and then making adjustments. In our view, both Joss' and Riebold's analyses present an appropriate approach to the task of selecting an actuarial interest rate assumption, and we accept their conclusions. Similarly, we find Haneberg's and Poulin's expert reports present valid approaches to selecting an actuarial interest rate assumption. However, we cannot accept their conclusions because they omit several important considerations. Haneberg's historical analysis*112 used only the most recent 20 years of the available data, while Joss' analysis used 40 years. Given the fact that most of V&E's plans were to be funded (both pre and postretirement earnings) over a 30- to 50-year period, the longer-term data are more reflective of the likely long-term rates of return. Haneberg also failed to use "total returns" and instead used "yields" in computing his forecasts. Yields are a measure of annual rate of interest or dividend return on a given investment, while total returns measure the return provided by the yield in addition to gains or losses incurred on the disposition of the asset. As previously discussed, it is likely that self-directed plans may experience ill-timed changes in investment strategy, thus total return rates tend to be lower than yield rates over time. When Haneberg's methodology was adjusted for these two factors and the ranges recalculated in an identical fashion by V&E's financial expert, Peavy, the resulting range of expected returns was 3.6 percent to 9.78 percent. The amended Haneberg range is consistent with Joss' ranges of 3.6 percent to 11.4 percent, and 3.6 percent to 11.8 percent, for 1986 and 1987, respectively. *113 Moreover, Haneberg's primary experience was with large multiemployer plans. He had little experience with small plans and no experience with IDB plans. He failed to make any adjustment for the small size of V&E's plans despite acknowledging that many small plans would end up with lower interest rate assumptions because of their investment policies. Poulin arrived at a reasonable range of 7 percent to 11 percent. The lower endpoint was calculated with a beginning point of approximately 9 percent, which was the rate paid by 30-year Treasury bond yield in January 1988, less a 2-percent *38 adjustment. However, the January 1, 1988, date used by Poulin to set the benchmark rate is not appropriate. In choosing a benchmark rate for his analysis of a plan year beginning July 1, 1986, Poulin selected the latest period that the actuary could have prepared and signed the actuarial report required by section 6059. We consider a more logical date for such a plan year to be the valuation date, in the case of Poulin's analysis, July 1, 1986, the 30-year Treasury bond benchmark rate would have been approximately 7.2 percent. A plan's valuation date is a point in time at which the actuary*114 determines the plan's then-projected benefit obligation and the plan's assets to meet such obligation. This is the mathematical starting point from which that year's contribution will be added. However, since the contribution can be calculated only after the actuarial assumptions for that year are selected, it would be unreasonable to wait 19 months before establishing the assumptions. This is especially true in the case of a cash basis plan sponsor who, like V&E in these cases, must make the contribution within its taxable year to be entitled to flow the deduction through to its partners. Making the downward 2-percent adjustment determined by Poulin from the more appropriate benchmark rate of 7.2 percent produces a 5.2-percent reasonable range lower endpoint. In addition, Poulin's analysis reaching a 2-percent reduction due to actuarial concerns in the benchmark rate was applicable to large plans. Had Poulin accounted for the plans' small size, the additional reduction would have further lowered the endpoint. We also note that Poulin's interest rate assumptions in his own significant experience with small plans are more consistent with V&E's experts' conclusions than his*115 own in this case. He previously had used an interest rate assumption of 5.5 percent for many small plans during the same period at issue here, including his own. The interest rate assumption is a long-term estimate that must accommodate diverse and erratic market conditions. Since the actuary's paramount duty is to create a realistic pattern of contributions over the life of the plan which ensures that the plan sponsor will be able to provide adequate funding of the ultimate pension obligation, funding assumptions must have viability in the long term. Against this background, we find the following factors discussed above particularly important in determining the *39 reasonableness of the 5-percent actuarial interest rate assumption concluded by V&E's experts: (1) The nature of the responsibility Congress entrusted to enrolled actuaries in the statutory scheme enacted for defined benefit pension plans; (2) the conservative nature of the actuarial assumption selection process; (3) the fact that V&E's IDB plans were long term in nature with funding to occur over a 30- to 50-year period; (4) the fact that V&E's plans were self-directed with each participant being a coadministrator, *116 especially since most of the IDB plans did not employ a professional manager; (5) the fact that V&E's plans lacked "credible experience" with respect to earnings, investment strategies, and otherwise; (6) the risk of losing compounded earnings in a tax-exempt trust associated with using overly optimistic assumptions, and the resulting requirement for unanticipated higher contributions in later years; (7) the relative closeness of all the actuarial experts' reasonable ranges; and (8) the fact that most actuaries used interest rate assumptions of between 5 percent and 6 percent for small plans during the years at issue. The combination of these factors weighs heavily in favor of concluding that 5 percent was reasonable. Respondent also proffered the testimony of two nonactuaries (Shapiro testified at trial and submitted a report, while Jaffe only submitted a report) regarding the financial and equity markets in which investments are likely to be made. We received this evidence with the caveat that these persons were not actuaries and, as such, conclusions they drew would have limited application in the determination of the reasonableness of actuarial assumptions. Shapiro and Jaffe, *117 neither of whom had established actuarial assumptions or studied actuarial science or its underlying principles, discussed rates of return they expected the market to bear in the future. Both financial experts concluded that a 5-percent return was an unreasonably low forecast of expected earnings for assets invested during the period at issue. Both of these financial experts used probabilistic economic forecasting techniques that are different from and contrary to actuarial principles in several material respects. For example, Shapiro stated that past investment data cannot be used as a factor to predict future returns, whereas section 412(c)(3) specifically requires the enrolled actuary to take *40 into account the plan's past experience. Sound actuarial principles require that once a plan accumulates "credible experience", the actuary should consider such experience. Shapiro also testified that small plans should be able to achieve the same rates of return earned by large plans, while most actuaries during the period at issue predicted a significantly lower rate of return for small plans. Moreover, all the actuarial experts in these cases agreed that small plans tend *118 to underperform large plans. In addition, Shapiro's methodology was based on "yields" rather than "total returns". As discussed above, this is an unrealistic assumption that inflates the expected rate of return. For example, if an equity or debt instrument returns 10 percent per year on average over the stated period, it does not follow that an actuary should choose 10 percent for the same period as the appropriate actuarial interest rate assumption. In practice, investors self-directing their own plans frequently change investment strategies. Given the fluctuating nature of markets, many of these investors are likely to be buying and selling at the wrong time. Thus, the total return for those investors may be substantially less than that predicted for the broader market, regardless of how well that market performs. A principal distinction between the viewpoint of a financial analyst projecting future investment returns and an enrolled actuary establishing an interest rate actuarial assumption is that the latter projection is made with the objective of assuring that the plan has adequate assets to meet the pension benefit obligations as they come due. If a financial analyst's*119 predicted rate of return is higher than the actual rate earned, the investor simply earns less than he supposed he would earn. If an actuary makes the same mistake, there is a significant risk that the plan will become underfunded and the pensioners' full benefits will be unpaid. With due respect to the financial analysts whose testimony we received, while we have considered the information they presented regarding financial and equity markets, we cannot accept their conclusions. The analysis propounded fails to account for concerns an actuary for a self-directed IDB plan must take into account. Congress did not entrust the Nation's tax-advantaged retirement savings system to hypothetical returns that the markets "should" bear. That task *41 was left to actuaries whose background, training, orientation, and philosophy is well suited to the task. As practitioners specifically enrolled under the scheme established by Congress to create a smooth funding pattern assuring that benefit obligations will be met, they necessarily have a different perspective. The selection of an interest rate assumption is an actuarial judgment made in accordance with actuarial principles materially*120 different from financial market performance forecasting. In sum, after considering all of the evidence presented, we conclude that the 5-percent pre and postretirement interest rate actuarial assumption used to calculate the contributions for V&E's plans was reasonable for the years at issue. Retirement Age Smith established an actuarial assumption of age 62 for the retirement age of the V&E partners adopting an IDB plan. Each IDB plan provided that the partner was entitled to a full pension benefit in the form of a joint and survivor annuity (or an actuarially equivalent optional form) at the "normal retirement date". This date is defined in each plan as the later of the date on which the partner attains age 62 or the fifth anniversary of the date the partner became a partner participating in the IDB plan. These benefits were payable at the normal retirement date regardless of whether the partner continued to practice law, was otherwise employed, or was either fully or partially retired. The plans provided also that a partner who terminated active status at V&E prior to age 62 was entitled to the "accrued" portion of the benefit, reduced actuarially to take into account*121 the early termination. Since the sixties, V&E has maintained an unfunded retirement program for partners under its partnership agreement. Since 1966, V&E also has maintained a qualified defined contribution plan, known as the Lawyers Retirement Plan, to which it made contributions with respect to each partner, generally at the maximum deductible levels allowed under the law. At retirement, a partner's account in the Lawyers Retirement Plan was used to offset the promised retirement benefits under V&E's partnership agreement. In 1982, the law changed to permit higher deductible limits on funding *42 retirement benefits for self-employed individuals. This change in the law resulted in V&E's reviewing the feasibility of creating IDB plans for its partners. Gareth W. Cook (Cook), a partner at V&E and chairman of the firm's special retirement committee to review the feasibility of IDB plans for its partners, was involved in the design, adoption, and maintenance of the IDB plans. He was the member of V&E who met with representatives from Wolper Ross and communicated the firm's attitudes and objectives in adopting the plans. Cook testified that in adopting age 62 as the normal*122 retirement date in the plans, he was well aware that the then-current mandatory retirement age of 65 as set forth in the V&E partnership agreement was not especially relevant to what the expected retirement age of the partners would be in the future. He testified that before the adoption of the plans, the partners had been evidencing a desire for earlier retirement and for a mechanism for facilitating the progression of younger, more productive partners. When Cook met with Smith during 1984, he indicated that there was a definite trend within the firm toward earlier retirement and that it was very likely to continue. William H. Drushel, Jr. (Drushel), a partner at V&E and a member of the firm's management committee during the years in issue, testified that during 1984 to 1987, a relatively large number of the firm's partners were between the ages of 36 and 46. He was concerned that as these younger partners would all approach retirement at the same time and with the decreased level of productivity that could be expected generally from partners entering their sixties, it would become difficult to support the partnership financially. Drushel ultimately proposed that the firm *123 begin to consider the sufficiency of its retirement policies. The report of V&E's expert, Rabenhorst, demonstrated a clear trend toward earlier retirement in the general population. Rabenhorst cited Employee Benefit Research Institute briefs from August 1986 and June 1990 and a General Accounting Office (GAO) report from 1985 which confirm this trend. In particular, he noted that the GAO report provides "substantial evidence to indicate that age 62 has replaced age 65 as the retirement age chosen by most private sector workers." Rabenhorst explained that one of the most significant factors influencing the move toward earlier retirement *43 is the ability to establish pension plans that will provide adequate income for retirement. He also cited the growing social acceptance of earlier retirement as a factor influencing the choice to retire at an earlier age. Rabenhorst further explained that the trend toward earlier retirement in the general population was evident also among partners in large law firms and at V&E in particular. He based this conclusion on the changing retirement policies of large law firms. Based on his extensive experience dealing with large law firms, Rabenhorst*124 determined that the firms were decreasing their normal retirement age and establishing early retirement policies to encourage their partners to retire by age 62 or younger. Rabenhorst attributed this trend to several factors which have led to the changing nature of the practice of law. One significant factor was the ever-increasing number of lawyers. For example, he noted, "From the mid 1970s to the early 1980s, the population of lawyers in the United States increased by more than 25,000 lawyers per year." This growth has increased the competitiveness of practicing law and placed greater pressure on lawyers to be more productive and more profitable. Rabenhorst determined that because of the increase in competition in the legal profession, firms were forced to become more formalized and to operate more like traditional businesses, pushing issues such as retirement to the forefront. In addition, he explained that the increased use of sophisticated economic analysis enabled firms to measure the productivity of partners. According to Rabenhorst, "This new emphasis on management often exposed unproductive older partners and put pressure on firms to devise ways of dealing with *125 them." Moreover, Rabenhorst cited an increasing disenchantment among older partners. He noted that most of these older partners had anticipated a decrease in their workload later in their lives; however, instead, they were pressured to continue to produce at high levels because firms felt they could no longer pay high salaries to older partners who were not producing at the same level as younger lawyers. Rabenhorst determined that older partners were increasingly faced with pressures to make room for the younger partners. He concluded that all of these pressures led to the desire on the part of both the firms and the older partners for provisions for earlier retirement. Finally, he determined that, due to the *44 increased earnings levels of large law firms, partners were financially able to establish retirement plans that would provide sufficient income for earlier retirement. Based on his analysis of the trend toward earlier retirement, Rabenhorst concluded that the retirement age assumption of 62 for the V&E IDB plans was reasonable. V&E's expert Joss explained in his report that the retirement age assumption affects the defined benefit funding calculation by determining*126 the period of time over which assets are expected to accumulate and by determining the length of the anticipated payout period. Joss concluded that a retirement age assumption of age 62 was reasonable based on retirement trends in general, on the specific retirement policies of V&E, and on his own retirement planning experience. Joss not only found age 62 to be within the range of reasonable retirement age assumptions, but, in fact, he found age 62 to be the latest age that he would consider reasonable. He cited a 1985 GAO report which stated that most private pension plans allowed employees to retire as early as age 55 with reduced benefits, and that age 62 was the median age of retirement for individuals receiving a private pension. A 1986 GAO report explained that Federal tax policy encouraged retirement at age 62 and younger and noted that individual retirement accounts and Keogh plans allowed funds to be withdrawn for retirement without penalty at age 59 1/2. Joss explained that when a retirement plan provided that a participant was entitled to a full benefit at a particular age, a prudent actuary would not assume a retirement age beyond that age unless there were compelling*127 reasons to do so. A consequence of assuming an incorrect retirement age in a one-person plan is that the promised benefits will not be available for the plan participant at the desired time. There were no compelling reasons to assume a retirement age other than age 62 for the V&E IDB plans because age 62 comported with the retirement philosophy and objectives at V&E that had manifested in the early to mid-eighties. Respondent argued that the retirement age assumption of 62 for plan years 1986 and 1987 was unreasonable and not the plans' actuary's best estimate. Respondent contended: (1) The V&E partnership agreement, the Lawyers Retirement Plan, and the staff-defined benefit plans covering *45 nonattorney employees and certain associates all indicated that retirement at V&E would occur ordinarily at age 65; (2) the partners were not likely to claim the benefits from the IDB plans at age 62; and (3) a change in the law, effective for plan year 1987, served as an incentive to encourage partners to postpone claiming benefits from their IDB plans until past age 62. The 1980 V&E articles of partnership, which were amended several times during 1986, 1987, and 1988 without change*128 to the retirement provisions, provided for a change in a partner's status from percentage partner to a fixed-basis partner on September 30 following his or her 65th birthday and for a second change in status from fixed basis to retired on the fifth anniversary of having become fixed basis. A fixed-basis partner was required to turn over to other partners or associates of the firm the business he was handling and to train those attorneys in the handling of the business. Thus, respondent argued that because the partnership agreement required partners to cease full-time work at age 65, the reasonable retirement age assumption that should have been used by the plans' actuary is age 65. Further, the partnership agreement provided that after reaching fixed-basis status and continuing into retired status for the remainder of the partner's life, former percentage partners would be paid as a guaranteed payment a percentage of their base compensation. This guaranteed payment was to be 75 percent of their base compensation in the first year and 50 percent the second year, and ultimately would reach 25 percent in the fifth year and thereafter. Respondent maintained that as a result of these*129 guaranteed payments, new fixed-basis partners had no reason to commence receipt immediately of the "tax sheltered" benefits due them under the IDB plans, and instead would probably defer actual receipt of the benefits for many years after reaching fixed-basis status. Respondent argued that as the real issue to consider in arriving at a retirement age assumption is benefit commencement -- when it could be expected that the V&E partners would actually claim their benefits from the IDB plans -- the plans' actuary could not have arrived reasonably at age 62. Respondent contended that section 415(b)(2), as amended by the Tax Reform Act of 1986, Pub. L. 99-514, 100 Stat. *46 2085, and effective for plan year 1987, reduced the annual limit on the maximum benefits from $ 90,000 to $ 72,000 or less for those claiming benefits at age 62 and, therefore, encouraged partners to delay receipt of their benefits under their IDB plans until after reaching age 62. Finally, respondent disagreed with Rabenhorst's report. Respondent took issue with the fact that the only empirical data Rabenhorst used related to retirement trends in the general population and not specifically to law firms. Moreover, *130 respondent argued that Rabenhorst's report is "largely irrelevant" because his data regarding law firms dealt with the retirement policy of law firms, not the retirement behavior of partners. Finally, respondent criticized Rabenhorst's use of hindsight data as irrelevant because the actuaries did not have such data available at the time they made their assumptions regarding retirement age. Haneberg, respondent's expert, reported: "There is some statistical basis for the belief that attorneys tend to continue practicing until relatively elderly." He cited statistics contained in a report by the American Bar Foundation which profiled the legal profession in the United States in the eighties. This report was based on information gathered from Martindale-Hubbell, a listing for attorneys. According to this report, only 20.3 percent of the lawyers ages 65-74 were retired or inactive in 1980 and 63.9 percent remained in private practice. In addition to this "apparently clear preference of a large percentage of the legal profession to continue practicing law", Haneberg noted the financial rewards and prestige of a V&E partnership and found it "difficult to see why many partners would*131 retire from Vinson & Elkins unless they either intended to leave the private practice of law or were strongly encouraged to leave the firm." The arguments raised by respondent do not support the conclusion that a retirement age assumption of 62 for the IDB plans in issue was unreasonable. First, the certifying actuary is not charged with the responsibility of determining when a plan participant will actually begin to receive the plan benefits. That would be an impossible task. Further, the fact that a plan participant might choose to, or actually does, delay receipt of the plan benefits beyond the assumed retirement age does not make the retirement age assumption unreasonable. An actuary is charged with looking into the future and *47 making a determination as to, among other things, when benefits under the plan could begin. V&E, through Cook, had expressed to its certifying actuary its objective of moving toward a younger retirement age than currently existed within the firm. A way for V&E to accomplish this objective was to create a mechanism whereby its current partners would financially be able to retire at earlier ages. By adopting IDB plans that provided benefits*132 would be fully available at age 62 and by determining plan funding costs so that benefits would actually be available at age 62, V&E could ensure that its partners would financially be able to retire at earlier ages. Respondent relied heavily on the fact that many of V&E's organizational documents assumed a retirement age of 65. The V&E partnership agreement required a partner to enter fixed-basis status at age 65, and respondent concluded therefore that age 65 was the date at which benefits would commence. However, this same partnership agreement allowed early retirement, albeit with a slightly reduced benefit package, at age 60 or at any age with 35 years of service. The defined benefit plans which covered nonattorneys at V&E had no bearing on the reasonable retirement age assumption of the partners. The Lawyers Retirement Plan defined retirement date as the date on which a partner 65 years or older terminated employment with the firm. Yet the significance of retirement date in a defined contribution plan is to determine when a participant will fully vest, and the Lawyers Retirement Plan provided for full vesting after 10 years of service with the firm, regardless of age. *133 In any event, these documents should not override the provisions of the IDB plans when the IDB plans are in conjunction with and evidence a clear institutional policy to move toward an earlier retirement age. In making an assumption about the retirement age of the partners who were adopting IDB plans, the certifying actuary was to determine retirement experience not as it existed currently, but as it would exist many years into the future. Respondent's argument that the change in the law which reduced the annual limit on the maximum benefits for those claiming benefits at age 62 would probably have encouraged the V&E partners to delay receipt of their benefits is questionable. The language of section 415 provides for an *48 adjustment to the maximum annual benefit of $ 90,000, depending upon when such benefit commences. If the benefit commences before the Social Security retirement age, a reduction is to be made to the benefit so that it equals the actuarial equivalent of the $ 90,000 annual benefit, which would have begun at the Social Security retirement age. Although a plan participant would receive a reduced benefit, the benefit would be payable for a longer period, *134 and therefore he or she would receive ultimately at least the same total amount of benefits. Next, we address respondent's concerns with the report by Rabenhorst. We agree with respondent that Rabenhorst used data related to retirement trends in the general population. However, this data served as a foundation for his analysis, not as specific evidence of what was happening in the legal profession. Such background material was an appropriate starting point for explaining the trends in the legal community, which Rabenhorst determined paralleled those in the general population. Moreover, although Rabenhorst used no specific empirical data directly related to retirement behavior among law partners, he had over 20 years of experience working with law firms and saw firsthand the changes that were taking place. We find that Rabenhorst's use of data regarding law firm retirement policy is relevant because policy influences behavior. If law firms provide the opportunity for earlier retirement and partners are financially able to retire earlier, it is likely that more partners will retire at earlier ages, especially considering the greater pressures on partners which Rabenhorst noted. *135 Furthermore, although it is true that Smith, the certifying actuary, would not have had access to the hindsight evidence that Rabenhorst considered, such evidence has value as confirming evidence. Rabenhorst testified that the trend toward earlier retirement was evident in 1984 when the certifying actuary determined the retirement age assumption. The fact that later data showed that V&E partners actually did retire earlier reinforces the reasonableness of the actuary's determination. We note finally that respondent's expert, Haneberg, testified that based upon all the information available in 1984, the range of reasonableness for retirement ages for the IDB*49 plans in issue "was probably 63 to 68". This range comes very close to incorporating the retirement age assumption actually used. Given the opinions of Smith, Rabenhorst and Joss, who determined that age 62 was on the high end of the reasonableness range, we hold that the retirement age assumption of 62 used to determine plan funding costs for the V&E IDB plans was a reasonable assumption. 1958 CSO Table for Preretirement Mortality In 1986, 49 of V&E's plans established insured benefits in the event of preretirement*136 death. These death benefits ranged in value from $ 500,000 to $ 4 million. Section 1.412(c)(3)-1(f), Income Tax Regs., provides that the cost of certain "ancillary benefits" may be funded by a qualified pension plan. An "ancillary benefit" is defined as a benefit paid as a result of an event which occurs preretirement and which is "detrimental to the participant's health." Sec. 1.412(c)(3)-1(f)(2)(ii), Income Tax Regs. Death is deemed to be an event detrimental to a participant's health, see sec. 1.412(c)(3)-1(f)(2), Income Tax Regs.; thus, a preretirement death benefit such as that provided for in V&E's plans falls under the rubric "ancillary benefit". The cost of an "ancillary benefit" may be computed "using the same method used to compute retirement benefit costs under a plan", sec. 1.412(c)(3)-1(f)(1), Income Tax Regs. (envelope funding), or "may equal the premium paid for that benefit under an insurance contract", sec. 1.412(c)(3)-1(f)(3), Income Tax Regs. (premium paid method). To provide for the ancillary death benefit, the plan trustee bought a universal life insurance policy for each plan. The death benefit would no longer be provided after the plan participant reached*137 the assumed retirement age of 62. Smith chose envelope funding to compute the premium costs. Smith then amended each of the 49 plans so as to use the 1958 CSO mortality table to arrive at preretirement mortality assumptions. The other plans, which did not provide a preretirement death benefit, used the 1971 Individual Annuity Mortality (1971 IAM table) for funding purposes. For those plans with an insured preretirement death benefit, Smith continued to fund postretirement liabilities based on assumptions derived from the 1971 IAM table. *50 The 1958 CSO mortality table is used by insurance companies to estimate the mortality of purchasers of life insurance policies. The 1971 IAM table is used by insurance companies to estimate the mortality of purchasers of annuities. The 1971 IAM table predicts a lower rate of deaths than does the 1958 CSO mortality table. There are three basic reasons for this disparity: (1) Human life expectancy improved between 1958 and 1971; (2) actuarial experience dictates that purchasers of annuities live longer than purchasers of life insurance policies; and (3) insurers build margins for profit and conservatism into mortality tables in order*138 to ensure solvency when benefits are claimed. This final consideration, commonly known as "loading" the mortality table, necessarily leads to opposite consequences in the sale of life insurance as compared to the sale of annuities. In the case of life insurance, insurers are prone to overestimate the chances of death, but, when selling annuities, insurers will generally underestimate the chances of death, thus increasing available reserves for the required payouts under each kind of policy. Smith's use of the 1958 CSO mortality table in his envelope funding method therefore resulted in an assumption of more deaths among those participants whose plans provided for insured preretirement death benefits than among those participants in plans without insured preretirement death benefits. This in turn led to increased funding needs for, and increased contributions to, those plans with insured preretirement death benefits. Respondent argued that Smith's use of the 1958 CSO mortality table was not reasonable for the plans with an insured preretirement death benefit in that it overstated the probability of each participant's death, unreasonably inflating each participant's tax-deductible*139 plan contribution. In addition, respondent contended that the use of two different mortality tables for the same individual was unreasonable and overstated both pre and postretirement liabilities. Respondent's expert Haneberg stated that he found it "troublesome from a professional standpoint that the * * * [1971 IAM] table could be adopted for one purpose by a V&E Plan while at the same time the plan's enrolled actuary is utilizing a table that predicts between two and three times as many deaths." Haneberg stated that the most accurate *51 predictor of plan participants' preretirement mortality was neither the 1958 CSO mortality table nor the 1971 IAM table, but rather a mortality table developed in 1984 by the actuarial firm Buck, which assumed fewer deaths at all relevant ages than even the 1971 IAM table. Respondent's expert Poulin argued that there should be no preretirement mortality assumption in a one-person plan at all. He characterized mortality in an individual pension plan as an "either/or situation"; that is, the participant either dies before retirement or survives until retirement. V&E argued that Smith did not use the 1958 CSO mortality table to accurately *140 predict the incidence of preretirement mortality among the plan participants, but rather used it as a means of estimating the cost of providing life insurance coverage for the preretirement death benefit. V&E contended that use of the 1958 CSO mortality table was thus a reasonable method of funding the plans. V&E's expert Joss testified that Smith's use of the 1958 CSO mortality table to estimate the cost of the preretirement death benefit was less expensive, and therefore more reasonable than other available funding options. The present value of a preretirement death benefit using an Allstate Insurance Co. quote, and using the P.S. 58 rates, 13 estimated higher costs than those estimated under the 1958 CSO mortality table. Thus, Joss concluded that use of the 1958 CSO mortality table to fund the death benefit as the least expensive of these three options, all of which were available to Smith, was reasonable. *141 V&E's expert Riebold concluded that it was reasonable for Smith to use the 1958 CSO mortality table to estimate the cost of the preretirement death benefit to the plans and the 1971 IAM table for the postretirement mortality assumptions for those same plans. Because an insurance company would use a CSO-type table in its calculation of the premium cost for the preretirement death benefit and this premium would be a portion of the actual costs for the plan as a whole, Riebold contended that Smith was correct in having done the same. Concomitantly, reasoned Riebold, the postretirement *52 benefit to be paid by the plan was in the nature of an annuity, and thus the appropriate table to be used in estimating mortality in that situation was an IAM type table to ensure that sufficient reserves are available to fund the defined benefit during participants' retirement. Riebold stated: "it is not considered irregular to use a death benefit mortality table that is different from that used for the Maturity Values of the pension benefits. The determination of pre-retirement mortality involves considerations and objectives that differ from those involved in determining post-retirement*142 mortality." Respondent acknowledged that V&E may provide for preretirement death benefits in its tax-qualified pension plans and that envelope funding is a legitimate way to pay for these benefits. It follows that Smith had to choose a method by which he could estimate the real cost of a preretirement death benefit premium to the plan. Smith therefore had to use a mortality table such as that used by an insurer to compute a life insurance premium, regardless of what the actual chances of death were. While Haneberg and Poulin may be correct that the 1958 CSO mortality table overstated what the V&E plans' actual mortality experience would be, this has no bearing on the purpose of Smith's use of the 1958 CSO mortality table, which was not to measure accurately potential participant mortality but rather was to estimate the life insurance premium expense to each of the plans. In addition, while the actual chances of any one of the participants dying before age 62 may be so low as to be negligible, as Poulin contended, the fact remains that, by definition, a preretirement death benefit assumes some possibility of preretirement death. We believe that Smith was correct in choosing a*143 CSO table to make his preretirement mortality assumptions, as this was the same type of table which would be used by an insurance company in computing the premium it would charge to the plans. While there was a more recent CSO table available to Smith, the 1980 CSO table which used more current life expectancy figures, respondent's experts did not contend that Smith should have used this newer table. In fact, Haneberg, while testifying before the Court, acknowledged that to make a reasonable mortality assumption, one need not use the most recent table available. *53 A mortality table should be used only for the purpose which it is meant to fulfill. A preretirement death benefit has different requirements than a postretirement annuity. In the case of the former, there must be reserves available if the beneficiary dies unexpectedly soon, and in the case of the latter, there must be reserves available if the beneficiary lives unexpectedly long. What is important is that when the plan participant or his beneficiary required the funds to be disbursed, the funds be available. The dual mortality table method devised by Smith legitimately fulfills the funding needs of both a preretirement*144 death benefit and a postretirement annuity. We hold that Smith was reasonable in his use of the 1958 CSO mortality table to arrive at the preretirement mortality assumption in those of V&E's plans, which provided a preretirement death benefit. Postretirement Expense Load Smith established a 5-percent postretirement expense load factor for V&E's plans and used that assumption in calculating the required contribution for the years at issue. A postretirement load factor is used by actuaries to provide for plan expenses that will be incurred in the postretirement period of the plan. As with all the other actuarial assumptions, there is no single correct percentage for all plans. Smith established the 5-percent factor for V&E's plans to serve two purposes: (1) To provide for postretirement plan expenses; and (2) to account for the mortality improvement that was not reflected in the 1971 IAM table used for the postretirement mortality assumption for the plans. Section 13.05 of each of V&E's plans provided that plan administration expenses "may be paid" by V&E, but if they were not paid, the expenses "shall be paid" by the plan trustee. V&E did not intend to incur or pay the costs*145 of administering the plans postretirement. Joss explained that an expense load is a generally accepted actuarial technique and that it is a technique contemplated by respondent, as Schedule B of Form 5500 specifically provides for disclosure of any expense load. Moreover, Joss noted that for small plans, administrative expenses are significant in relation to assets, and such expenses were on the increase *54 throughout the eighties. Joss concluded, "Certainly, it was appropriate for the actuary to be concerned about payment of these expenses and it was reasonable for him to provide for their payment from plan assets." Joss also agreed that the 5-percent expense load was justified by mortality improvement. Based on the mortality improvement that had occurred between 1951 and 1983, he estimated the amount of mortality improvement that would occur by the year 2000 (the year at or after which most V&E partners will reach 62 years of age). From this analysis, he determined that the use of the 1971 IAM table, combined with the 5-percent expense load, accurately accounted for the expected mortality improvement. Riebold listed the types of expenses the plans were likely to incur, *146 including trustee, investment, actuarial, accounting and legal fees, and costs related to plan amendments and legislative compliance. She noted that in 1984 the certifying actuary would have known that these expenses were on the rise and indicated that, in fact, the 5-percent load likely would be insufficient to provide for all of the expenses the plans would incur. She testified that postretirement expenses could reach as high as 12 percent or 15 percent. Respondent argued that since each plan participant is a cotrustee, there would be no trustee expense. In addition, respondent contended that investment expenses should not be considered to be part of the expense load because those expenses merely reduced the investment return. Respondent also asserted that actuarial expenses would be minor because the law only required actuarial valuations every 3 years. Finally, respondent maintained that compliance costs would be minimal, if they existed at all, because all of the plans were identical, and compliance could be obtained through uniform amendments written for all of the plans. Respondent assumed that because V&E would have to amend the plans for nonretired partners, the *147 firm would charge little or nothing to amend the plans of retired partners. Haneberg, respondent's expert, conceded that it would be reasonable to anticipate that V&E would not pay postretirement administrative expenses for the IDB plans. However, he noted that at the time, the law only required actuarial valuation every third year, which reduced possible postretirement administrative expenses. He concluded that the expense load *55 assumption was "on a spectrum between Trivially Understates (for the relatively small asset accumulation) to Overstates (for the plans anticipating relatively large asset accumulations)". Nevertheless, at trial, he conceded a 2-percent expense load to account for postretirement expenses was reasonable. Respondent also contended that the 1971 IAM table sufficiently accounted for future mortality improvement, and therefore none of the expense load could be justified by mortality improvement. Respondent relied heavily on Haneberg's conclusion that the 1971 IAM table was appropriate for predicting postretirement mortality. We find that respondent's contention that there should be no expense load assumption is flawed. There is sufficient evidence*148 to support the conclusion that the plans will incur significant postretirement expenses. Respondent ignored the existence of the other cotrustee who, presumably, would not donate such services free of charge. Furthermore, there was no indication that an actuary would consider investment expenses to be part of the investment return analysis rather than part of the expense load analysis. In fact, Haneberg excluded such transactions costs from his interest rate assumption analysis, thereby indicating that such costs should be accounted for elsewhere. In addition, some allowance for actuarial expenses was appropriate. We note that Smith was making valuations yearly, rather than every 3 years as respondent assumed he would. Nevertheless, even if the actuarial valuations were only made every 3 years, allowance must be made for that expense since the actuary is not likely to perform his service without charge. Finally, respondent presented nothing that compels the conclusion that V&E would always provide free administrative and compliance services. The partners and associates who would provide such services would have no connection with retired partners or incentive to provide such*149 a benefit to those retirees, whom they know only by name and maybe from a few aging "war stories". We believe that it is completely reasonable to assume that the plan will be responsible for administration costs during the postretirement period. *56 We hold that the 5-percent expense load assumption was reasonable and appropriate in light of the significant post-retirement expenses the plans are likely to incur. Moreover, the indications that the 1971 IAM table understates post-retirement mortality reinforce our conclusion that the 5-percent expense load assumption is well within the reasonable range. Furthermore, we find that as 5 percent is not a conservative assumption, an even higher percentage would have been reasonable. Best Estimate Respondent asserted that the "best estimate" test of section 412(c)(3) constitutes a separate and distinct element requiring that selection of assumptions be founded on the actuary's "highest degree of professional judgment". Respondent intimated that the "best estimate" language of section 412(c)(3) imposed a separate and higher duty in the selection of actuarial assumptions. While the argument is not entirely clear, respondent suggested*150 that since each of the assumptions should offer the best estimate of anticipated experience under the plan for which assumptions are being established, the actuary should not rely on comparisons to the actuarial assumptions used by other plans. Respondent also argued that since the actuary must exercise "the highest degree of professional judgment" in choosing actuarial assumptions that are the "best estimate" of anticipated experience under the plan, he or she should not be concerned with actuarial conservatism. Respondent supported this contention by pointing to the lack of express statutory recognition of actuarial conservatism, noting that if Congress were concerned with actuarial conservatism, "it undoubtedly would have said so." We find no merit in respondent's arguments. As discussed earlier, we believe that Congress was concerned with adequate funding of pension plans. In choosing to leave selection of pension plan assumptions to enrolled actuaries, Congress must have been aware of the basic tenets of actuarial science, including actuarial conservatism, and designed the statutory structure with them in mind. Furthermore, we do not read the "best estimate" statutory*151 language as somehow preventing an actuary from making comparisons to other *57 plans that in his or her judgment reflect the actuarial mainstream. 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. Tax Motivation Respondent argued that the principal purpose of the IDB plans was to provide V&E's partners with a tax shelter and that the partners' principal motivation for making contributions was to obtain deductions. Accordingly, contended respondent, the actuarial assumptions made by Smith were strongly influenced by the partners' desire to obtain*152 large tax deductions and should be found to be unreasonable. Respondent, as a basis for these arguments, pointed to the attention given by Wolper Ross and V&E in internal memoranda to the tax considerations involved in the IDB plans. We are persuaded that V&E's principal purpose in adopting the IDB plans was to help provide adequate retirement income for older partners that would enable them to retire gracefully and not have to keep up with the modern-day, high-pressured pace of the law firm, as well as to open up the way for younger partners to advance within the firm. Tax considerations have always been, and presumably will always be, an important factor in retirement plans. V&E and its partners would have been remiss not to take into account the tax aspects of the IDB plans. Furthermore, it is a well-settled and basic tenet of Federal tax law that taxpayers are not precluded from structuring their affairs so as to minimize their taxes. As Judge Learned Hand eloquently stated: Over and over again courts have said that there is nothing sinister in so arranging one's affairs as to keep taxes as low as possible. Everybody does *58 so, rich or poor; and all do right, *153 for nobody owes any public duty to pay more than the law demands: taxes are enforced exactions, not voluntary contributions. To demand more in the name of morals is mere cant. [Commissioner v. Newman,159 F.2d 848, 850-851 (2d Cir. 1947) (Hand, J. dissenting), revg. 5 T.C. 603 (1945).] V&E cannot be faulted for choosing to employ a tax-advantaged technique enacted by Congress to encourage taxpayers to save for retirement. Rather, such behavior is precisely the type desired by, and thus statutorily encouraged by, Congress. Evidentiary Matters Respondent objected to the use of hindsight evidence, particularly with respect to financial and retirement statistics arising after the 1986 and 1987 valuation dates. We agree that the actuaries, in reviewing the actuarial assumptions, should be confined to evidence available at the time the assumptions were made. Later statistics may confirm the validity of the assumptions but are not relevant to the assumptions themselves. In our consideration of these cases, we have confined ourselves to the facts available at the time the actuarial assumptions were made and, therefore, consider*154 this objection to be moot. Respondent also objected to the use by the actuaries of respondent's training manuals, audit guidelines, internal and external correspondence, and transcripts of speeches made by Service employees regarding the matters at issue here. To the extent that these documents have been disseminated publicly, they are part of the actuarial universe within which all actuaries must live, think, and work in arriving at their conclusions as to reasonableness and their best estimates regarding appropriate contributions. These documents are part of the actuarial way of life, and taking them into account does not violate the hearsay rule and does not constitute going behind the notice of deficiency. The foregoing is particularly true of transcripts of speeches by someone such as Cohen, the then-director of the Actuarial and Technical Division of the Service, to professional conferences and seminars by and for actuaries. The purpose of those conferences and seminars was to educate and update the actuaries with respect to the latest developments in *59 actuarial science. A speech made at one of those conferences or seminars by a high-ranking Service employee, heard*155 by many actuaries and disseminated by publication to many more, is not hearsay. The fact that the speech was made under the circumstances in which it was made is important and relevant, and actuaries surely should be entitled to look to such speeches for guidance in carrying out their duties. Any objection to the quality of the transcript of Cohen's February 1986 speech to enrolled actuaries is overcome by the stipulation that the transcript is true and correct. Conclusion We find that each of the four challenged assumptions was reasonable, and the actuarial assumptions and methods used by V&E were reasonable in the aggregate. A fortiori, these assumptions are not substantially unreasonable so as to permit retroactive changes of assumptions for years prior to the year in which the audit is made. We further find that the actuarial assumptions and methods certified by V&E's enrolled actuary in combination offer the actuary's best estimate of anticipated experience under the plans. Decisions will be entered for petitioner. Footnotes 1. F.S.A.: Fellow of the Society of Actuaries. The society is a certification organization for actuaries, which administers exams to demonstrate educational qualifications. The society also conducts and publishes studies concerning mortality and health risks. "Fellow" is the society's highest designation; to be so designated, one must pass a full syllabus of examinations.↩ 2. M.A.A.A.: Member of the American Academy of Actuaries. The highest (and only) designation in that organization. The American Academy of Actuaries issues standards of practice and sets discipline procedure for the actuarial profession.↩ 3. E.A.: Enrolled Actuary. Actuaries may be "enrolled" by the Joint Board for the Enrollment of Actuaries, a certifying organization composed of members appointed by the Secretaries of the Treasury and of Labor, which was established in 1974 pursuant to the Employee Retirement Income Security Act of 1974 (ERISA), Pub. L. 93-406, 88 Stat. 829, 1002. ERISA requires that defined benefit plans be audited annually on behalf of their participants by an actuary so enrolled. This actuary makes funding assumptions for the plans and certifies as to the reasonableness in the aggregate of those assumptions.↩ 4. F.C.A.: Fellow of the Conference of Consulting Actuaries. The highest designation in that organization. Fellowship requires 12 years of responsible actuarial work. All members are consultants, as distinguished from insurance company actuaries.↩ 5. Form 5500 is a report filed with Labor, the Service, and Pension Benefit Guaranty Corp. by all ERISA-qualified (i.e., tax-qualified) pension plans. Form 5500 contains information such as interest rate assumptions, actual rates of return, and plan funding status.↩ 6. B.B.A.: Bachelor of Business Administration.↩ 7. C.F.A.: Chartered Financial Analyst. This designation is awarded after one has taken a series of examinations in portfolio management, security analysis, and investments and has passed certain experience and ethical requirements.↩ 8. LTV, Eastern Air Lines, and Pan American Airlines are among the well-known, underfunded defined benefit pension plans to terminate leaving the Pension Benefit Guarantee Corp. (PBGC) to cover the underfunded benefit obligation. A Nov. 25, 1991, PBGC news release reported that the 50 largest underfunded pension plans were underfunded by $ 14.2 billion in 1990, up 50 percent from 1989.↩ 9. Sec. 412(c)(3) requires the actuarial assumptions to be reasonable in the aggregate "taking into account the experience of the plan" as well as reasonable expectations for the future.↩ 10. Consistent with this conclusion, V&E's computer data expert, Schechter, confirmed that for 76.6 percent of the preretirement assumptions and 82.5 percent of the postretirement assumptions, actuaries established interest rate assumptions of between 5 percent and 6 percent for 1986 plans with fewer than 100 participants. Schechter's conclusions were based on his analysis of data obtained from Labor.↩ 11. Joss provides confirming data obtained from Labor showing that plans with fewer than five participants earn a median return of 5 percent in 1988, while the median return for larger plans (100 or more participants) is upwards of 8 percent for the same period. Respondent's computer data expert, Borden, questioned the magnitude of the difference in rates of returns earned by small and large plans but confirmed that, in general, "smaller plans would tend to have lower rates * * * [of return] than larger plans."↩ 12. The four contingencies in question are as follows: (1) A 1986 congressional warning about 5-percent investment rate assumptions; (2) a 1985 Financial Accounting Standards Board statement regarding "best estimate" assumptions; (3) continuing favorable investment results between 1984 and 1986; and (4) 3 years of very favorable investment experience from V&E's plans.↩ 13. The P.S. 58 rates are those rates deemed by the Commissioner to be acceptable in determining the cost of life insurance protection includable in gross income for a participant covered by a life insurance contract held in a qualified pension plan. Rev. Rul. 55-747, 1955-2 C.B. 228, amplified by Rev. Rul. 66-110, 1966-1 C.B. 12. See also sec. 1.72-16, Income Tax Regs.↩