Theory & Writings

Valuing Defined Pension Plans

There are two (general) types of pension plans. The most common are "defined contribution" plans, those where (pre-tax) monies are invested in stocks, bonds, mutual funds or other assets. An IRA is a defined contribution plan. So is a 401(k). While defined contribution plans are often referred to as "retirement plans"---this is the common intended use--- they are really nothing other than pre-tax investment vehicles (generally) not accessible until a certain age (without penalty). Accordingly, the net worth of a defined contribution plan is the gross asset total at valuation date, less some adjustment for taxes. The tax adjustment should be a prediction of taxes that will actually be paid at the time of distribution. Thus, if there is $100,000 in an IRA, and the predicted tax rate is 15% (federal) and 4.4% (state), the net asset value is $80,600.

The second type of plan is the "defined benefit" plan, one where the benefit paid has no (necessary) connection to the assets contributed. Most defined benefit plans pay a monthly amount calculated by the following formula---

Years of Service * A Defined Percentage * Final Average Compensation

The "defined percentage" is typically a number between 1.5% and 2.5%. "Final average compensation" tends to be defined by the highest years of compensation, usually an average of the highest 3 or 5 years. Thus, assuming employment for 24 years, a 2.0% defined percentage, and a final average compensation of $36,000 per year ($3,000 per month), the gross monthly pension would be $1,440 ($3,000*.02*24). Ages of commencement of pension benefits are plan specific, sometimes allowing for a reduced pension if taken at an earlier age.

A variation on this theme is the "two-pronged" pension, one where a specific benefit is paid after a defined number of service years. This kind of plan might allow for retirement after 30 years, without any age limitation. However, until a certain age is reached a monthly payment is fixed regardless of final average compensation or other factors. An example might be a "thirty and out" pension that allows for an ordinary defined benefit at age 60 but, prior to that time, pays the employee a flat amount of $2,000 per month. Thus, if the employee retires at 55, he/she would receive the "flat benefit" until the plan algorithm is applied at age 60.

Valuing a Defined Benefit Plan in Michigan: The seminal case in Michigan is Boyd v. Boyd, 116 Mich App 779, 780 (1982). See Zechin v. Zechin, 149 Mich App 723 (1986); Perry v. Perry, 133 Mich App 453 (1984) and Sullivan v. Sullivan, 175 Mich App 508 (1989) and Pickering v. Pickering, 268 Mich App 1 (2005) following Boyd. As of this writing (December 2010) the basic premises of Boyd remain good law, though there has been some litigation about “how much” of a Boyd value should be distributed to a non-holding spouse. [To be discussed.]

As a preliminary caveat it is noted that the majority (maybe the vast majority) of defined benefit plans are divided by DRO. A DRO [QDRO for qualified federal plans; EDRO for certain State employment plans] eliminates any of the actuarial hypotheses about value and are (generally) preferred, particularly where any litigant would rather opt out of these speculations or prefer not to “buy out” the present value. See Boyd at 782-83. This said, there are still many cases where a “buy out” of value makes sense. And many cases where this is the preference of the parties.

 In these cases Boyd holds that the value of the plan is derived by (1) calculating the life expectancy of the holder at his or her retirement age, (2) reducing the stream of payments (as projected by the number in (1)) to a present value (as of the evaluation date) by using a discount rate, (3) summing the reduced stream of payments and (4) reducing this by the possibility the holder will die prior to the pension going into pay status. This yields a gross value. As with defined contribution plans (assuming the homogenization to net value) it is necessary to reduce the result by an expected tax rate. See the discussion vis a vis defined contribution plans.

Valuing a Defined Benefit Plan in Washington: One of the first cases of consequence in Washington is in re Marriage of Pilant, 42 Wash App 173 (1985) and this case references Boyd v. Boyd (the fundamental Michigan case) and cases from other jurisdictions as being cornerstones for/relevant to the establishment of Washington law. Indeed, Pilant cites the pages from Boyd where the Michigan court suggests the algorithm that was eventually used in the Support pension plan valuator. Pension Plans have often been valued in Washington by reduction to a present lump sum, but the appellate opinions don't seem to lay down any firm methodology; instead, it seems, the cases confront whether a court had sufficient actuarial evidence to justify its valuation of a plan. Pilant does suggest that the method should be to calculate the "pension's future value, expressed in a lump sum, and then discount factors are employed to calculate a present cash value". This is consistent with the Support pension plan valuator method. Also, in re Marriage of Kraft, 119 Wash 2d 438 (1992) it is held that "reducing a pension to its present value is a recognized procedure in the valuation of divorce assets", though no definitive methodology is mentioned. We suggest that the Support pension plan valuator should be consistent with Washington law.

Valuing a Defined Benefit Plan in Kentucky: We have researched Kentucky appellate cases and, as far as we can tell, there is no reported case concerning the method for placing a dollar value on this quality of asset. The methodologies contained in this part of the Support program reflect Michigan law and is the methodology we have seen employed by accounting experts in Michigan and in other jurisdictions. Our assumption is that Kentucky, like Michigan, prefers the use of DROs (usually a QDRO) as a division theory.

 Valuing a Defined Benefit Plan in Florida: Florida law provides that: “No recitation of formulae, considered in the abstract, could capture the variety of considerations necessary in order to do equity. Valuation of retirement benefits is fact-intensive and varies depending upon the plan, and the trial judge must determine the equitable valuation with the limitation being the valuation is not to include post-marriage contributions.” Boyett v. Boyett, 753 So 2d 451 (Fla 1997) citing Diffenderfer v. Diffenderfer, 491 So 2d 265, 266 (Fla 1986).

Nevertheless, DROs don’t seem to be clearly favored in Florida, at least in relation to the appellate pronouncements. In Egerton v. Van Den Berg, ---Fla App--- (5th District, October 15, 2010) the history was outlined:

In addition, the trial court, having found the Foley and Lardner pension contract to be entirely marital, employed the immediate offset method and converted the Wife's half interest into a lump sum distribution based on a discounted present value applying an assumed life expectancy which recognized a reduction in benefits occurring at age 80. This approach is considered the preferred approach. Diffenderfer v. Diffenderfer, 491 So. 2d 265 (Fla. 1986). At the same time, a lump sum distribution may not be the best way to ensure equitable distribution in every case. Boyett v. Boyett, 703 So. 2d 451 (Fla. 1997).

This court has expressly held that if determining the present value of a pension is too speculative, a trial court should make an award of a portion of the future payments when and if received. Brock v. Brock, 690 So. 2d 737 (Fla. 5th DCA 1997); see also Swanson v. Swanson, 869 So. 2d 735 (Fla. 4th DCA 2004). It is clear, however, that the trial court has a great deal of discretion in making such awards. Canakaris v.Canakaris, 382 So. 2d 1197 (Fla. 1980).

Valuing a Defined Benefit Plan in Maryland: In the early 1980s, Maryland, like Michigan, determined that these plans were marital property, generally subject to division. See Deering v. Deering, 292 Md 115 (1981); Prince George County v. Burke, 321 Md 699 (1991). My research (though not extensive) does not show any Maryland case that looks to any algorithm for calculating a buyout of the asset. Deering does provide that the valuation standards in Maryland are "elastic," and that one method would be to present value the plan, discounting for time (declining value of money), mortality and life expectancies. Id at 130-31. My surmise is that, at least, the issues in Maryland and Michigan aren't variant. My assumption is also that Maryland, like Michigan, prefers the use of DROs (usually a QDRO) as a division theory.

Caveat. My experience is that most litigants bridle at the buyout/present valuation since (a) there aren’t sufficient resources---sometimes, it isn’t vaguely plausible-- and (b) for many the asset feels well too remote to be of current tangible value. This said, there are cases where use of a DRO may be less than preferred. These cases would include (a) 12pt pension plans where a DRO may be overkill of an asset, (b) instances where both parties prefer to retain his/her own plan without the use of DROs and (c) instances where the parties prefer to trade his/her interest in a plan for another asset (a house, for example). The authors stress that the program is not intended as a substitute for an expert evaluation. All DB plans have their own idiosyncrasies and must be closely examined. The point of the program is to give a practitioner a sense of what a plan/interest may be worth, so that planning and negotiation can accrue with a sense of the nature of an asset.

The first DB Valuator: In 1990/1991 an actuary and programmer, Jim Becker (in conjunction with Craig Ross) wrote a program conforming to the calculations suggested in Boyd v. Boyd with two variations, both suggested by Boyd. First, while Boyd states that the “statutory mortality tables [of MCL 500.834] may be used for purposes of calculating present value, they need not be used”. The court is explicit that “other actuarial mortality tables deemed to be more accurate” may be applied. Indeed the court goes on to note that MCL 500.834 does “not distinguish between males and females”. And, “as such, they tend to overstate males’ life expectancies and understate females’ life expectancies.” On the basis of this note, and the fact that the statutory mortality tables date to 1958, Becker chose the 1984 mortality tables used by the Pension Benefit Guaranty Corporation. These tables, unlike that shown in MCL 500.834, distinguished between males and females. See

What the DB 2000 Model did: DB 2000 continued the model of Becker-Ross and the basic algorithm of Boyd v. Boyd. The changes were (a) the adjustment of mortality tables to those published by the Public Health Service (US Dept of Health and Human Services, 1996 data) and (b) a greater definition within the mortality tables, by the addition of race. In that the mortality rates for black Americans and white Americans are markedly different, the addition of this factor is a tame extension of footnote 2 in Boyd v. Boyd. Boyd suggests, where male and female life expectancies are different, that it makes sense to distinguish this characteristic. The author suggests the same is true as to race, where the differences in mortality are also significant. I note, as of this writing, that no appellate (Michigan) court has made this adjustment or spoken to the issue. I did speak with two experts on the matter, Joe Cunningham of Plante and Moran and Gary Rogow of Rogow and Loney. Both are respected experts in the application of accounting/actuarial principles to divorce. They suggest the “race” adjustment is (at the least) rational.

What the Current Model does: Life expectancies are increasing and so the current version has upgraded the life expectancy tables using data (for whites and blacks) from the 2006 data as shown in the June 28, 2010 National and Vital Statistics Report (Vol 58, No. 21) as published by the Center for Disease Control and National Center for Health Statistics. Also, in October of 2010 the CDC/NCHS (Series 2, No. 152) published mortality tables for Hispanics. This data is also from 2006. These tables have been incorporated into Support and older tables have been excised. [The CDC and the NCHS are a part of US Center for Health and Human Services.]

Issues Surrounding The Valuation Of A Defined Benefit Plan:

(1) Discount Rate Assumption. This is as much art as science and is why, in part, the application of any program is not a substitute for an expert view. This said, I think it is possible to make a reasonable interest rate assumption by looking to the amalgam of the federal discount rate, T-Bill rates (over the interim to pay status) and money market rates. As a broad overview, the number has been in the range of 4% over the past few years. But, in the current economy, the guess is more treacherous over any period of time. Right now (12/2010), I might guess around 2% over 5 years, 3% at 12 years, 3.5% over 20 years and 4% over 30 years. The PBGC web site ( can help on this. After looking to these sources my surmise is that most of us can come within 1%-1.25% of what an expert would generally conclude.

(2) Mortality Tables To Be Used. See above discussion.

(3) Tax Rate To Be Applied. Assuming it is relevant (that is, all measured against assets have also been tax adjusted), the rate should reflect the probable tax rate of the payee when the pension is in pay status. This poses a question---does one use the “average” rate or the marginal rate? And, if the latter is used, is the rate chosen by hypothesizing the pension to be the “first” or “last” monies earned? There is no definitive answer on this. My tendency has been to pick a rate(s) all parties can live with. If the parties are not comfortable with the number a DRO is always available as an outcome.

(4) Date When Pension Accrues For Valuation Purposes. This was addressed in Kilbride v. Kilbride, 172 Mich App 421 (1988). That court held that the value of a pension “should be determined” upon the assumption that “the employee spouse will retire at the earliest date permitted under the plan”. Id at 437. This conclusion was premised upon an accounting monograph by Troyan. Troyan argues that since the issue is “totally under the control of the employee spouse” the earliest permitted date is the one that makes the most sense. Troyan admits, however, if the “earliest permitted date” is prior to the valuation date, the “date of the action” is more appropriate. This holding was/has been criticized and in Heike v. Heike, 198 Mich App 289 (1993) where the court refused to follow the rule. The Heike court held that, while the “earliest permitted date” was an acceptable choice, the trial court had a broader discretion and could choose other dates. The court held that “no one method is required”. Id at 292. Justice Gribbs, in a concurring opinion, stated that “the ‘earliest date’ calculations mandated in Kilbride provide a clear direction and a practical rule of thumb that may be appropriate in most cases.” Id at 294. See also Booth v. Booth, 194 Mich App 284 (1992). I think that the Gribbs opinion accurately reflects present practice---that the “earliest date” is a reasonable rule of thumb, but not a mandatory condition. In any event, where DROs are preferred over valuation disputes, an in-kind division can avoid an issue on point.

(5) Coverture Notions. Kilbride also held that (a) “pension benefits attributable to the employee spouses’ continued employment following the [JOD]” should not be a part of any sharing by the non-employee spouse and (b) looking again to the Troyan monograph, “the benefit has to be reduced” to “reflect any time for which the employee spouse was a member of the pension system before marriage”. 172 Mich App at 438-439. The foregoing premised a “coverture” percentage that would divide the years of marriage “in the plan” by the years of service “in the plan” and multiply this number by the value of the pension. In other words, if the employee-spouse worked at a position for 20 years, and the parties were marriage for 10 of those years, 50% of the value of the plan would be divided. This has been, as far as I can tell, a common occurrence in Michigan courts, whether via DRO or a plan valuation. This fact aside, the “coverture” notions of Kilbride have been overruled, at least to the extent that such a percentage is mandatory. In Boonstra v. Boonstra, 209 Mich App 558 (1995) it is held that while “pension contributions made during the marriage must be considered”, there is nothing in the law “that contributions made before or after the marriage may not be considered”. Id at 562. Boonstra cites Booth v. Booth, supra, in support of its holding. See also Rogner v. Rogner, 179 Mich App 326 (1989) and Vander Veen v. Vander Veen, 229 Mich App 108 at n. 2 (1998). McMichael v. McMichael, 217 Mich App 723 (1996) also follows Boonstra, stating that “pension benefits accrued before marriage may be the subject of a division of property”. As of today the courts have broad discretion in applying or not applying notions of “coverture”. The key, as with all Michigan property divisions, is whether the result is “fair”.

Coverture Redux. In Vander Veen v. Vander Veen, 229 Mich App 108 (1998) the court was faced with an attack on the “straight-line” notions of coverture. In that case an employee-spouse earned a pension over 32 years of work, but was married to plaintiff for only the last 5.5 years of the period. The circuit judge accepted testimony from an expert who divided 5.5 by 32 and multiplied this percentage times the total pension value (with a secondary adjustment) to reach a divisible amount. Plaintiff contested this methodology, although she did not contest the lack of sharing in the portion of the asset earned “outside” of the marriage. Rather, she contended that the appropriate measure was the gain “in the value of the pension from the time the parties married in 1990 to defendant’s retirement in 1995”. Id at 926. She contended that the pension, like any other asset, should be measured by the “change in net worth” that occurred, that is, the difference in the actuarial values of the pension between date of marriage and valuation. Indeed, under this calculation, 2/3 of the entire pension value accrued during the time of the marriage. The court struggles with this (or so it seems to me) but affirms the lower court where “In light of the relatively short duration of the parties’ marriage in comparison to the many years in which defendant earned his pension, the court did not clearly err in refusing to give plaintiff the benefit of defendant’s previous work before the marriage…” Id at 927. This conclusion seems to be pinioned on strong equitable but shaky analytical grounds, the latter being that work “outside of the marriage” predicated the growth during the marriage. This seems very problematic where this argument could be made to defeat, for example, any ordinary wage/savings accrued during a marriage. For example, the argument that “I earned my degree before we were married and that’s why I was able to earn money during the marriage”, would find support under this sort of analysis---and not anywhere else. In any event, the court goes on to repeat its holding as follows--

…under the circumstances of this case in which the trial court was only determining the portion of the pension that accrued during the marriage, the trial court did not clearly err in using the method for apportionment endorsed in Kilbride.

Id. The foregoing is hardly a “ringing endorsement” of the Kilbride equation in the primary run of cases and implies that a court might not “clearly err” via the methodology suggested by Ms. Vander Veen, at least if the “circumstances of the case” (a longer marriage, children born to the couple) so justified. I would expect to see more litigation on point---most likely in relation to “coverture” percentages as applied to QDROs.

What Ross-Becker, DB 2000 and the Current Incarnations Are Not. They are not a substitute for expert advice. While I believe the programs provide (with the caveat of garbage in-garbage out) a reasonably accurate view of the present value of defined benefit plans---they are not intended as a replacement for an expert/actuarial review of the plans. Moreover, the discount rate assumption has elements of art. So do “coverture” and other issues that are relevant to value.



Return To Top