In the competition among retirement-planning methodologies, systematic withdrawals have been winning the battle against the essential-discretionary approach. But given today’s low interest rates, the essential-discretionary approach may work better for many clients, especially if single-premium immediate annuities (SPIAs) are used.
Systematic withdrawals rely on traditional assets such as bonds and stocks, and all retirement needs are funded from a single portfolio. With the essential-discretionary approach, the client's spending needs are divided into those two categories — essential and discretionary — and separate portfolios are used to fund each one.
The case for systematic withdrawals
Jonathan Guyton and Michael Kitces, two thought leaders in financial planning, both strongly advocate systematic withdrawals. They gave a series of video interviews in 2011 at The American College on withdrawal methods, and Wade Pfau, professor of retirement income at that institution, recently featured them on his blog here and here.
Research on systematic withdrawals originated with planner Bill Bengen's development of the 4% rule in the early 1990s. He used historical data and demonstrated that, if the initial withdrawal rate were set at 4% of savings and the dollar amount of subsequent withdrawals increased with inflation, one could be highly confident that the savings would last for at least 30 years.
However, this approach, which has been used in research, would be too inflexible if applied in practice. In the videos, both Kitces and Guyton make the point that systematic withdrawals should not be viewed as a "set it and forget it" method. Instead, withdrawals should be adjusted based on emerging investment experience. Guyton has helped make the method more practical for planners by developing and testing decision rules to adjust for investment experience as described here.
Among the advantages Kitces and Guyton cite for systematic withdrawals are that the method treats all retirement savings as a single source of funds that can be managed to best meet retirement needs and that the focus can be on managing spending to maintain overall lifestyle rather than on just meeting subsistence needs.
Where the debate begins
Although Guyton and Kitces make a compelling case for systematic withdrawals, there is considerable debate among researchers about whether this is the best approach for generating retirement income. Central to the debate is whether the 4% rule is still viable in an environment of low interest rates and lower-than-historical-return prospects for stocks. Economists tend to favor splitting retirement expenses into essential and discretionary and using products such as SPIAs to fund the essential expenses. SPIAs provide an income that will be available to pay for essential expenses regardless of the length of life.
Guyton and Kitces do not argue that annuities are bad products, but they note that clients often resist them. They also express concern about splitting expenses into essential and discretionary, when in reality people try to maintain a lifestyle that includes both. But at the foundation of their advocacy for systematic withdrawals is their belief that withdrawal rates of 4% and higher are still feasible despite the challenging investment environment.
Kitces argues that, although current return prospects may be somewhat diminished, close to 90 years of historical data support the 4% rule, including periods that were more challenging than the current environment. He views the 4% rule as demonstrating what has worked in worst-case situations. He makes the point that periods of average returns would have allowed for higher withdrawal rates. Guyton mentions his decision rules, which add flexibility that can place initial safe withdrawal rates in the 5.2-5.6% range.
But there is a growing body of recent research that challenges the 4% rule. This 2013 article by Michael Finke, Wade Pfau and David Blanchett placed particular emphasis on bond yields, which were 4% below historical averages when the research was done. They noted that, even if rates were to rise in the future, safe withdrawal rates would be heavily impacted by the yield environment at the time of retirement. Their studies showed safe withdrawal rates as low as 2.5%. Kitces took issue with this study in this blog post, and many others have weighed in on both sides of the viability of the 4% rule.
An example
I'll develop my own viability assessment for systematic withdrawals, and then apply an essential-discretionary approach utilizing SPIAs to see what difference it makes.
First, I need investment return assumptions. For bonds, I am persuaded by Finke, Pfau and Blanchett that, for retirees who will have no future savings contributions, current bond yields largely dictate future returns. Since I will model withdrawals that increase at the inflation rate, I'll use Treasury inflation-protected securities as bond investments. For stocks, I assume they maintain their historical return advantage over intermediate-term government bonds. I assume 0.15% for investment expenses, and my analysis is pre-tax. The result is arithmetic real returns of 0.65% for bonds and 6.25% for stocks.
My example is based on a single 65-year-old female who is a typical upscale financial-planning client and in good health. Based on earlier research, I assume a life expectancy of 90.3 years. I'll assume that she has $1 million in savings and needs to withdraw $30,000 per year, increasing with inflation, to cover essential expenses and an additional $10,000 for discretionary spending. Her total withdrawal needs match up with the 4% rule.1
I do the modeling using Monte Carlo analysis with the standard deviations for stock and bond returns based on historical averages – 20.3% for stocks and 6.8% for bonds. Mortality is also variable in the modeling.
The chart below shows projected results in today's dollars varying the stock/bond allocation in 25% increments.
4% inflation-adjusted withdrawals, no SPIA
Stock allocation |
Median bequest |
Failure rate |
5th percentile bequest |
0% |
$77,464 |
43.2% |
-$573,882 |
25% |
$352,538 |
24.8% |
-$416,330 |
50% |
$661,250 |
16.4% |
-$362,640 |
75% |
$887,912 |
16.0% |
-$420,080 |
100% |
$1,053,674 |
18.6% |
-$537,062 |
Source: Author's calculations
The median bequests increase as stock allocations increase, reflecting the return premium of stocks over bonds. The failure rates (the percentage of simulations where the individual depletes savings before death) are minimized with stock allocations in the 50-75% range. This finding is consistent with results produced by Bengen and others since. However, because I am using returns that are lower than historical averages, the failure rates are above 16%, whereas returns based on history would bring failure rates down to 5% or below. I also include a column of 5th percentile bequests, which provides more information than failure rate because it accounts for both probability and magnitude of failure.
This chart highlights the importance of return assumptions. My view is that the average return assumptions that get fed into Monte Carlo simulations should represent the researcher's best estimate for those retiring today. In particular, it's important to recognize the impact of current bond yields on future returns for retirees. This approach lowers return assumptions and significantly raises the downside risk measures.
Adding SPIAs to the mix
I'll now add a SPIA purchase to cover the assumed $30,000 of essential expenses. Inflation-adjusted SPIAs can be purchased from Income Solutions® via Vanguard, and the current annualized monthly payout rate is 4.54%. So it requires about $660,000 ($30,000/.0454) to purchase a SPIA that will cover essential expenses in the example. (This immediately brings up the practical issue of convincing a client to spend two-thirds of his or her retirement savings on an annuity, which I will deal with below.)
The chart below shows the impact of the SPIA purchase that produces $30,000 of inflation-adjusted income and leaves $10,000 to be provided from remaining savings of approximately $330,000.
4% inflation-adjusted withdrawals, with SPIA
Stock allocation on non-SPIA assets |
Median bequest
|
Failure rate |
5th percentile bequest |
0% |
$119,538 |
16.8% |
-$67,894 |
25% |
$235,216 |
4.7% |
$3,716 |
50% |
$440,052 |
3.7% |
$20,308 |
75% |
$425,136 |
5.8% |
-$13,152 |
100% |
$492,460 |
8.6% |
-$57,102 |
Source: Author's calculations
Except for the all-bond allocation, the median bequests are below those shown in the first chart where SPIAs were not utilized. In effect, SPIAs are fixed-income investments, so devoting two-thirds of savings to SPIAs tilts the allocation strongly toward fixed income regardless of the stock allocation on remaining savings. But the good news is that adding a SPIA dramatically reduces risk. Failure rates can be brought down to below 5% and the 5th percentile loss measures are much lower.
Look at the top lines (0% stock allocation) in both charts. SPIAs not only improve the risk measures, but they also raise the median bequest. The common wisdom is that SPIAs reduce risk but also offer lower expected returns than for bonds, because of insurer charges. However, as I point out my article on SPIAs, these products may actually produce better returns than bonds. This result depends on assumptions about both longevity and interest rates, but SPIAs deserve serious consideration in a head-on-head comparison with bonds or bond funds.
The impact of higher withdrawal rates
In the charts below, I change the example by raising the discretionary spending to $20,000 so that the targeted withdrawals are increased to an inflation-adjusted $50,000 and a withdrawal rate of 5%.
5% inflation-adjusted withdrawals, no SPIA
Stock allocation |
Median bequest |
Failure rate |
5th percentile bequest |
0% |
-$218,210 |
65.3% |
-$952,832 |
25% |
$19,498 |
48.8% |
-$807,396 |
50% |
$317,576 |
34.2% |
-$758,342 |
75% |
$531,288 |
30.2% |
-$815,664 |
100% |
$700,010 |
29.5% |
-$908,746 |
Source: Author's calculations
5% inflation-adjusted withdrawals, with SPIA
Stock allocation on non-SPIA assets |
Median bequest |
Failure rate |
5th percentile bequest |
0% |
-$155,248 |
74.7% |
-$436,844 |
25% |
-$96,790 |
65.6% |
-$391,872 |
50% |
-$5,352 |
50.7% |
-$378,848 |
75% |
$72,592 |
42.7% |
-$386,404 |
100% |
$141,846 |
38.7% |
-$410,216 |
Source: Author's calculations
In the top chart without SPIAs, risk measures are much worse than in the first chart with a 4% withdrawal rate. Even if one applies the decision rules suggested by Jonathan Guyton, there is more than a 30% chance of having to reduce withdrawals over the full course of retirement below the lifestyle goal of a 5% rate.
The last chart shows that introducing a SPIA no longer reduces risk to acceptable levels. Because the SPIA payout rate of 4.54% is below the desired 5% withdrawal rate, using it raises the actual withdrawal rate demanded of savings not used to purchase the SPIA. Therefore, we see higher failure rates when the SPIA is introduced, although the 5th percentile bequests improve because the SPIA provides three-fifths of desired withdrawals.
This 5% withdrawal case illustrates the concluding argument in this article by retirement guru Moshe Milevsky:
The annuity price is actually a market signal of what retirement really costs. And, it is the cheapest and safest way to convert a nest egg into a lifetime of secure income. Market prices convey information and the cost of a life annuity is a hard-drive full of intelligence.
In this case, the 4.54% payout rate on an inflation-adjusted annuity shows that a 4% withdrawal rate – but not a 5% rate – is feasible with the longevity risk reduction provided by a SPIA.
Practical considerations
Unfortunately, a host of behavioral issues stand in the way of using SPIAs (or any other type of annuity) in a retirement income strategy. Just think of the likely reaction from a newly retired client to the recommendation that they use two-thirds of their retirement savings to purchase an annuity. Advisors need to anticipate client reluctance. Chances of success will be improved by starting the SPIA discussion early, many years before actual purchase. Also, it may help to spread purchases of SPIAs over 10 or so years. Financial planner Paula Hogan has had success with this strategy, and she discusses the spreading of purchases and other practical considerations for using SPIAs in this blog post.
Other products may also be worth considering. For example, deferred-income annuities (DIAs) with payments beginning late in life (age 85 or 90) provide longevity protection with a much lower purchase price than SPIAs, although savings must be used to provide retirement income until DIA payments begin. Another option is a variable annuity with a guaranteed lifetime withdrawal benefit (VA/GLWB), particularly a low-cost version like the product offered by Vanguard or available in a few 401(k) plans. VA/GLWBs may provide comfort in that amounts used for purchase do not reduce assets on clients' investment statements, unlike with SPIAs and DIAs.
None of these strategies or products produces outcomes that are superior to purchasing a SPIA at retirement, but there may be practical or behavioral reasons for recommending them.
Systematic withdrawals using regular investments still remains the most popular approach, but given the current investment climate, advisors need to give serious consideration to annuity options like SPIAs to help clients meet retirement goals.
Joe Tomlinson, an actuary and financial planner, is managing director of Tomlinson Financial Planning, LLC in Greenville, Maine. His practice focuses on retirement planning. He also does research and writing on financial planning and investment topics.
Read more articles by Joe Tomlinson
