Executive Summary
Although the research on safe withdrawal rates has been replicated many times by various researchers to substantiate a safe, sustainable spending level that can withstand at least anything that history has thrown at a retiree, one significant challenge has always lingered: a safe withdrawal rate recommendation is only as good as the time horizon it's associated with. In other words, while the research may support a 4.5% safe withdrawal rate, it's predicated on a 30-year time horizon. If the client planned to retire over a 35- or 40-year time horizon, the safe withdrawal rate would be different. Unfortunately, though, the client may not know that a 35-year time horizon is needed until it's year 31 and there are still a few years left to go! So what's the outlook for a safe withdrawal rate approach if the client outlives the original time horizon?
The inspiration for today's blog post comes from recent conversations I've been having with researcher Wade Pfau at the National Graduate Institute for Policy Studies in Tokyo, Japan (it's a pain trying to schedule our calls!), about safe withdrawal rates, and in particular about something that safe withdrawal rates pioneer Bill Bengen mentioned in a recent article about his work in Forbes, which in turn is being discussed/debated on the Bogleheads forums. Specifically, Bengen mentioned that even if someone starts at a 4.5% withdrawal rate and goes for 30 years, there is still a 96% chance (based on historical analysis) that the client will still have 100% of his/her original principal remaining to fund the extra years.
Wait, what? The safe withdrawal rate, which we've been taught is how low spending has to be to sustain in bad economic environments, also still leaves over more than the original principal a whopping 96% of the time? Shocking, but true. In fact, Wade verified it (not that we doubted Bengen's word). The results are at the end of this post.
Assuming starting wealth of $100, it is in fact true that final wealth is still over $100 in about 96% of the scenarios; in only 3 instances is wealth lower than the $100 starting amount, including 1937, 1968, and 1969 (the latter being the scenario when the account winds down to $0 by the end, and is thus why ~4.5% is the "safe" limit). On the other hand, as mentioned earlier, in almost 96% of the remaining cases, final wealth is at least $100. In fact, the median wealth level after 30 years is a whopping $460! That's right, 50% of the time you do your lifetime spending at a 4.5% withdrawal rate, and more than quadruple your account balance on top of it!
On the other hand, it's worth noting that these are nominal dollars, not real dollars. While it's nice to more-than-quadruple final wealth, spending has also risen dramatically over this time period. Accordingly, the results below also show final wealth on an inflation-adjusted basis; nonetheless, median final wealth is still about $161 (after starting at $100), and in 69% of the scenarios the final wealth is still more than inflation-adjusted dollars too! That's quite a "cushion" for extra longevity! (Editor's Note: These projections assume withdrawals at end-of-year; if the withdrawals are beginning-of-year instead, starting principal is preserved 89% of the time based on nominal wealth, and 55% of the time in real wealth.)
The bottom line is that while safe withdrawal rates ratchet spending down to the point where a retiree can survive a terrible sequence of returns (and/or a substandard period of total return), in the overwhelming majority of cases, the outcome is not nearly so dire. In point of fact, most of the time the safe withdrawal rates approach is a path to significant wealth accumulation, and/or an adjustment period several years into retirement where spending can be increased to account for rising wealth. Nonetheless, for retirees who do not want to ever face the risk of cutting their spending, safe withdrawal rates provide a rising-floor approach that allows for spending or wealth to rise, without anticipating cuts. But while it's the conservative measure needed to protect in bad markets, be cognizant that in "merely average" - not to mention, good - markets, your client's greatest problem may be what to do with all the extra money.
So what do you think? Does this change your thinking about how conservative or aggressive safe withdrawal rates are? Is a 96% chance of having 100% of principal left over a comforting factor when trying to manage the risk of unanticipated longevity?
| Year | Final Nominal Wealth | Final Real Wealth |
|---|---|---|
| 1926 | 456.04 | 304.47 |
| 1927 | 427.14 | 273.11 |
| 1928 | 264.89 | 160.98 |
| 1929 | 158.79 | 93.91 |
| 1930 | 289.77 | 169.18 |
| 1931 | 451.36 | 244.03 |
| 1932 | 987.94 | 480.07 |
| 1933 | 937.35 | 403.68 |
| 1934 | 499.35 | 212.62 |
| 1935 | 530.35 | 227.66 |
| 1936 | 306.54 | 132.96 |
| 1937 | 95.67 | 40.64 |
| 1938 | 558.57 | 237.39 |
| 1939 | 350.31 | 138.22 |
| 1940 | 341.19 | 126.27 |
| 1941 | 476.46 | 168.77 |
| 1942 | 800.57 | 301.01 |
| 1943 | 776.34 | 308.49 |
| 1944 | 474.14 | 178.7 |
| 1945 | 308 | 105.65 |
| 1946 | 186.54 | 61.14 |
| 1947 | 470.45 | 173.8 |
| 1948 | 548.02 | 206.71 |
| 1949 | 618.29 | 219.68 |
| 1950 | 595.95 | 183.5 |
| 1951 | 561.04 | 162.59 |
| 1952 | 533.47 | 150.24 |
| 1953 | 604.74 | 165.43 |
| 1954 | 779.9 | 206.84 |
| 1955 | 442.45 | 112.32 |
| 1956 | 376.21 | 92.36 |
| 1957 | 476.52 | 118.98 |
| 1958 | 638.48 | 157.29 |
| 1959 | 372.96 | 89.55 |
| 1960 | 404.22 | 94.15 |
| 1961 | 395.47 | 88.1 |
| 1962 | 268.44 | 58.41 |
| 1963 | 445.95 | 95.45 |
| 1964 | 325.98 | 69.03 |
| 1965 | 185.22 | 38.66 |
| 1966 | 104.83 | 21.75 |
| 1967 | 349.12 | 72.46 |
| 1968 | 43.38 | 9.12 |
| 1969 | 0 | 0 |
| 1970 | 460.07 | 103.04 |
| 1971 | 520.46 | 118.94 |
| 1972 | 392.33 | 91.25 |
| 1973 | 287.83 | 67.62 |
| 1974 | 898.37 | 225.34 |
| 1975 | 1744.98 | 475.63 |
| 1976 | 1275.97 | 359.89 |
| 1977 | 969.95 | 279.64 |
| 1978 | 1200.15 | 354.95 |
| 1979 | 1050.57 | 338.43 |
| 1980 | 1091.23 | 386.76 |
| # of 30-year periods | 55 | 55 |
| # of successful periods | 52 | 38 |
| Probability of Success | 0.945454545 | 0.690909091 |
| Median Final Wealth | 460.07 | 160.98 |
What’s the asset mix?
Eliot,
I believe it was a 60/40 portfolio. Will try to double-check.
– Michael
I was trying to match the asset allocation that Bill Bengen probably had in mind when mentioning that number. I based it on the discussion in his 2006 book, choosing:
40% large-cap, 15% small-cap, 45% intermediate-term government bonds
I think it is a bit misleading to use the 96% number. What people have in mind when they hear “your principal remains intact” is probably something closer to the real terms, not nominal terms. Though I really respect Bill Bengen’s research, I think 55% number is the better number to be mentioning in this context than the 96% number.
A difficulty I have with determining the relevance of this analysis is that if one has a lifetime annuity (esp. if it’s COL adjusted), the risk of ending up with 0 money after 30 years (or 50 years) is close to 0 — no matter what withdrawal rate one uses from the accumulated investments/savings. Thus, if one has SS and/or a partial annuity (preferably also with COLA), sufficient to cover needed expenses til end of life, then one might be more inclined to risk running out of money in one’s “accumulation” before 30 years is up.
Consider for a moment that you know you will die in the next 10 years but you have an equal chance of dying in any of those 10 years. And for simplicity imagine your portfolio doesn’t earn anything. How much should you spend each year?
Clearly you could split your funds into 10 pots and be guaranteed to never have to reduce your spending. But on average this means you will only spend half your money! Economists have developed a formula that based on your risk aversion to a declining standard of living, will front load your spending.
If you use this formula in retirement with an average risk aversion, you don’t see much decline in spending in the sixties and seventies but by the time a person is 100 their spending will be forced to drop in half. Obviously this ignores floors created by SS and other pensions. Since current financial planning programs don’t support this, you can approximate the result by having a client’s “real expenses” drop by 2% a year starting at age 80.
Obviously those with higher or lower risk aversion can adjust these numbers lower or higher.
Michael:
The doctrine of Personal Financial Planning to avoid risks that are too great to bear, so one must use a safe withdrawal method even if it does not allow one to spend down all of one’s net worth. The elderly face a risk of higher than the average person’s risk of inflation because they use more medical services which will go up more than CPI as more baby boomers get old and go to the doctor more frequently.
Very fascinating that inflation adjusted success is only 69%, which I consider to be a bit on the inadequate side. The saving grace is that these safe withdrawal methods ignore the value of a home (because it is a use asset not an investment) which could be sold when someone is age 95 and has run out of funds.
The have been articles written showing that the U.S. stock market was much luckier than the rest of world during the past 100 years and there is no guarantee that the next 100 years will be as good for the U.S. So a “conservative” 4.5% withdrawal rate is about as aggressive as I would want to get.
Michael:
We have also been actively following your work and Wade’s work. The terminal wealth is an important second reward metric we introduced in Income Discovery’s retirement income analysis framework last year. In fact, our platform reports that median terminal value in today’s dollars so that it is easily interpretable. This metric can be very helpful in comparing strategies that may give same income level at similar confidence level but very different terminal value characteristics. That helps identify superior retirement income strategies.
As I had explained in our one-on-one conversation (thanks for making time for that last year), the second risk metric of bad case portfolio life (worst 2nd percentile portfolio life) has turned out to be a very useful and pertinent metric with the Advisor population using the beta version of our platform. There are plans with similar confidence level and very different performance on bad case portfolio life.
That is why we encourage the use of 4 metrics – 2 risk and 2 reward metrics – in retirement income plan analysis.
Michael,
My clients don’t want to die with lots of money left over – they want to live to the max that they can afford in retirement and especially while we are young.
I think we need more research that keys in on SWR and its link to stock market valuations (like the P/E10) such as your Feb and April 2009 newsletter content. We can’t just let our retiree clients sit with a 40-60% exposure to overvalued markets…especially, as W. Pfau is revealing, to markets as over-valued as the stock market was in the late 90’s into early 2000.
Not being a researcher myself, I have to rely on those with a knack for it (such as yourself and Mr. Pfau) to help us figure out how to help client not outlast their money without erring too much on the side of them dying rich. Its seems like market valuations hold the key to us being able to get more precise with SWR.
I really value your contributions to our profession!
Kay
Michael, Kay, and others,
I just finished a new paper called, “Getting on Track for a Sustainable Retirement: A Reality Check on Savings and Work.” It’s not closely related to this blog entry. Rather, it was really quite inspired from your blog entry last November 19 about the logic of compounded returns. But it does indirectly incorporate valuations (like my “safe savings rate” paper) by arguing against a need to meet a specific wealth accumulation target by retirement.
This new paper is “safe savings rates” for mid-career individuals, as it indicates what still may need to be done in terms of saving and working before one can reasonably expect a sustainable retirement. I summarize and provide a link to the full paper at:
http://wpfau.blogspot.com/2011/06/getting-on-track-for-retirement.html
I mentioned your name on my blog, as I do wonder whether you find what I am saying to be persuasive, or if you have any other comments.
Thank you and best wishes, Wade
Hi Mike, I am almost certain that Bill Bengen used, in his SAFEMAX calculations, in this case, an allocation of : Tax-advantaged acc with 40% Lrg caps, 20% Sm caps, 40% intermediate Gov bonds.
Mike, what is your opinion on instead of picking an initial SWR and boosting it annually by the CPI, to follow Bill Bengen’s later approach (in his book) to monitor Current Annual Withdrawal Rate vs. Life Expectancy.
By using this approach there is no initial withdrawal rate to be chosen and to be boosted by the CPIs of previous years. Current Withdrawal Rate (CWR) is: end of year dollar withdrawal divided by the beginning of same year account value, both in nominal terms.
Here is an excerpt from the heading of the table which Bengen backtested since 1926 till 2005 on an allocation as I mentioned above:
Age…..A……B………..C…………D………E
Number values are for ages 55 till 100 at gaps of 5 years.
A = App unisex life expectancy
B = Adjust Life Expectancy by adding 10 years
C = SWR% for the adjusted LE
SWR% = safe withdrawal rate (SAFEMAX®)
D = Add 25% to C and get the RED FLAG%. Don’t go over the RED FLAG.
E = equity % suggested by Bill with three asset classes: LCS, SCS, and intermediate term Gov bonds, at the ratio of: SCS allocation = ½ LCS
Best wishes,
Vig