Given the considerable reputations of the collaborating organizations and of the report authors (two respected actuaries and a preeminent retirement researcher/academic), we, like many of our readers, had high expectations for the authors’ report. Suffice to say that we remain underwhelmed and disappointed by their recommended RMD withdrawal strategy. Several of our prior posts have discussed the limitations of the RMD approach. In this post, we will outline our “top ten” reasons why we believe the Smoothed Actuarial Budget Benchmark (Smoothed ABB), most recently discussed in our post of November 6, 2018 and November 19, 2018 Advisor Perspectives article, is superior to the RMD withdrawal approach advocated in the SCL/SOA study. In summary, unlike the Smoothed ABB, the RMD suffers from the following deficiencies:
- RMD was not designed to be a sustainable spending plan
- The assumptions underlying RMD withdrawal factors are questionable and are inconsistent with assumptions used to price inflation-indexed annuities.
- Contrary to the authors’ statements, the RMD withdrawal factors are not based on one’s life expectancy.
- Application of RMD is unclear for ages under 70.
- RMD doesn’t coordinate with other sources of income.
- RMD doesn’t consider non-recurring expenses.
- RMD doesn’t work very well for couple’s budgeting.
- RMD withdrawals can fluctuate significantly from year to year.
- RMD is inflexible and doesn’t accommodate “budget shaping.”
- RMD is expected to “back-load” real-dollar spending over an individual’s retirement lifetime.
Background—RMD Approach
Technically, RMD withdrawals apply only to amounts in tax deferred accounts. The authors propose that the same withdrawal rate used for an individual’s tax deferred account also apply to other accumulated savings owned by the retired individual each year. While the authors imply that the IRS RMD rule uses the distribution periods by age from Table III of Appendix B of IRS Publication 590-B (the Uniform Lifetime Table), we assume that individuals who must use Table I or Table II would determine their annual withdrawals from those tables. Note that while the Uniform Lifetime Table (which most people are eligible to use) starts at age 70, values for ages prior to age 70 determined in the same manner as after age 69 can be obtained from Table II of Appendix B. Mr. Vernon indicates that for ages prior to 70 “you can use a simple withdrawal rate of 3.5%,” but values under age 70 determined using Table II are generally lower than 3.5%. For example, at age 65, the hypothetical Uniform Lifetime Table withdrawal rate would be 3.13% and at age 60 it would be 2.72%.
1. RMD was not designed to be a strategic spending plan.
RMD was originally designed to make sure that U.S. taxpayers with tax deferred accounts eventually pay federal income tax on the amounts deferred. It wasn’t designed to be a strategic withdrawal plan, let alone a strategic spending plan. As a result (as discussed in more detail below), it is not surprising that it didn’t turn out to be an optimal budgeting strategy. Unfortunately, we believe that by endorsing this re-purposed very U.S.-centric approach, the Society of Actuaries, in its position as the preeminent global actuarial organization for education and research, has squandered an opportunity to show the world that basic actuarial principles can be used to develop a more robust solution to the global problem of determining how much an individual or couple can afford to spend in retirement.
2. The assumptions underlying RMD withdrawal factors are questionable and inconsistent with assumptions used to price inflation-adjusted annuities.
The questionable assumptions include:
- The mortality table used for RMD is a unisex mortality table (blended male and female rates) that was derived from the basic 2000 Individual Annuity Mortality Table. The resulting table has not been updated since its original adoption in 2002. By comparison, the IRS updates most mortality tables used for pension plan purposes for mortality improvements on an annual basis.
- The last time we looked males and females still experienced different mortality.
- The assumed rate of future investment return is 0%.
- The assumed rate of future inflation is 0%, and
- As discussed in more detail below, the lifetime planning period is not based on the life expectancy of the individual.
3. The RMD factors are not based on one’s life expectancy
Yes, we could have included this questionable assumption in item 2, but it is so questionable that we wanted to highlight it (also we needed it to bring our total to ten). In his recent article, Mr. Vernon states, “The RMD calculates the minimum withdrawal amount by taking your account balance on Dec. 31 and dividing by your life expectancy, as outlined in IRS Publication 590-B”. This is, at best, an “alternative fact.” In fact, IRS Publication 590-B takes great pains to refer to the relevant divisor in the Uniform Lifetime Table calculation as the “Distribution Period” and not the account owner’s “life expectancy”. The Distribution Period for the RMD calculation is not the life expectancy of the account owner. It is the joint life expectancy of the account owner and a hypothetical beneficiary 10 years younger than the account owner. Therefore, it is actually much longer than the life expectancy of the account owner.
4. Application of RMD is unclear for ages under 70
As discussed in the background section above, since RMD applies to individual account owners for years in which they reach age 70.5 and later, the IRS Uniform Lifetime Table distribution periods start at age 70. While Mr. Vernon says that you can use 3.5% (a distribution period of 28.57) for earlier ages, pre-age 70 distribution periods can be obtained by looking up the appropriate age (and 10-year younger beneficiary) in Table II. Arguably if you like the methodology used to calculate withdrawal rate percentages for ages after 69, you will like this same methodology just as much for ages before 70.
5. RMD doesn’t coordinate with other sources of income
When used as a strategic withdrawal plan, the RMD approach develops an amount to be withdrawn each year by dividing an individual’s (or couple’s) accumulated savings by the applicable distribution period. The amount to be withdrawn is added to income from other sources for the year to obtain the person’s (or couple’s) spending budget for the year. Like all SWPs, it doesn’t coordinate with income from other sources that may be payable in the future or that may differ from year to year. It does not consider all sources of retirement income. The classic example of the shortcomings of failure to coordinate with other sources of retirement income is when a QLAC (deferred annuity) is purchased with a portion of a retirees’ accumulated savings. Common sense might lead you to expect that such a purchase would increase the rate of withdrawal from remaining accumulated savings prior to commencement of the annuity and decrease the rate of withdrawals after commencement of the annuity, but this does not occur under the RMD approach. The rate of withdrawal is unaffected by a QLAC purchase (or by other sources of retirement income).
6. RMD doesn’t consider non-recurring expenses
The RMD approach anticipates using an individual’s assets to cover only annual recurring expenses and does not consider non-recurring expenses in retirement, such as long-term costs, unexpected expenses, car purchases, home improvements, limited period home mortgages, limited period travel expenses, etc. Mr. Vernon acknowledges this deficiency and suggests that retirees consider establishing separate “fun buckets” for these expenses.
7. RMD doesn’t work well for couples budgeting
Many couples develop their spending plans on a combined-asset basis. RMD determines required withdrawals from an individual’s account. There is no coordination between couple’s accounts. The authors are not clear how the RMD approach would apply, for example to a couple consisting of a 72-year old male and 65-year old female. While the 3.91% withdrawal rate applicable to the male would apply to amounts held in his tax-deferred accounts, what withdrawal rate would apply to the female’s tax-deferred accounts and what rate would apply to the couple’s combined non-tax deferred accumulated savings? And would these withdrawal rates change depending on the relative size of each person’s account? The authors are not clear how this would work, but in any event, the withdrawal rate used is unlikely to be entirely consistent with the couple’s spending goals.
8. RMD withdrawals can fluctuate significantly from year to year
As noted by Mr. Vernon, RMD withdrawals can fluctuate from year to year based on investment performance. In addition to investing in less risky assets as suggested by Mr. Vernon, retirees could consider smoothing the results over time, but apparently this is not anticipated under the RMD approach.
9. RMD is inflexible and doesn’t accommodate “budget shaping.”
In addition to discouraging smoothing of spending as discussed in item 8 above, RMD apparently also discourages individuals from front-loading their real-dollar spending even though research has shown that most retirees tend to spend less in real dollars as they age.
10. RMD is expected to back-load real-dollar spending over an individual’s retirement lifetime
Because of the relatively conservative assumptions selected by the IRS for tax recovery purposes discussed in items 2 and 3 above, the RMD approach is expected to actually back-load real dollar withdrawals over an individual’s retirement lifetime. In terms of spending, RMD starts out about 30% below the level anticipated through purchase of an inflation-indexed annuity for a male, and depending on assumed future investment experience, is expected to increase much more than 30% above the inflation-indexed level for the much later years of retirement. Because the SCL/SOA study assumed favorable future investment experience, the RMD approach performed better on average than other approaches they considered in their study. However, as noted in item 9 above, this back-loading of real dollar spending is inconsistent with actual experience and the spending goals of most retirees.
Conclusion
In summary, in case it isn’t clear from the comments above or from several prior posts on this topic, we aren’t big fans of the RMD approach. We believe that RMD sets a pretty low bar for developing spending budgets, and you can do better by applying the basic actuarial principles we advocate in this website to all your assets (not just your accumulated savings) and spending liabilities. Yes, RMD may be easier to calculate, but with a little more effort, individuals and couples can obtain a much better answer.
We are disappointed in the Society of Actuaries for facilitating and promoting the results of this study. The Smoothed Actuarial Budget Benchmark approach that we advocate has none of the ten deficiencies discussed above, and is clearly superior to the RMD approach. Therefore, as Fellows of the Society of Actuaries, we call on the SOA to:
- distance itself from the RMD approach, and instead
- publicize the Smoothed Actuarial Budget Benchmark, a much more robust non-U.S.-centric sustainable spending plan that utilizes basic actuarial principles, reasonable assumptions and the same inflation-indexed annuity pricing model previously employed by the SOA.
