A. The Peter Lynch Fallacy
It was less than two decades ago when famed Magellan Fund Manager Peter Lynch believed that retirees invested heavily in stocks could realistically expect to be able to withdrawal an inflation-adjusted 6.5 percent of their portfolio value to cover each year’s living expenses. That’s the average long-term real return for stocks. The Lynch logic was that, if your portfolio is earning that much each year, you should be able take that amount out each year without diminishing the value of the portfolio. It makes intuitive sense, no?
It doesn’t work like that.
Stocks suffered a price drop of 80 percent real in the years following the 1929 crash. Say that you retire with a portfolio valued at $1 million and a plan to live on $65,000 per year. Over the next four years, your portfolio value drops almost to $200,000 as a result of a dramatic price drop. Your withdrawals cause another minus of $260,000. You’re busted! Before Year Five begins!
Today’s retirement calculators (and the retirement planning advice offered by most retirement planners) is rooted in research done to quantify the effect of the critically important factor ignored by Lynch (to his credit, Lynch acknowledged the error when it was brought to his attention) — stock volatility. There is all the difference in the world between an average return of 6.5 percent real and a smooth return of 6.5 percent real. The volatility of stock prices greatly complicates retirement planning.
B. The Old School Safe Withdrawal Rate Research
Only the super-rich can afford to abstain from investing in stocks in retirement. There are times when it is hard to find super-safe asset classes (such as Treasury Inflation-Protected Securities [TIPS], IBonds, and Certificates of Deposit) paying a return better than 2 percent real. The retiree seeking to cover annual living expenses of $60,000 can achieve his goal with savings of $1 million invested in an asset class paying a return of 6 percent real but would require savings of $3 million to do so if he were invested solely in an asset class paying a return of only 2 percent real.
So most retirees need to be open to investing in stocks. But retirees investing in stocks must be wary. Their safe withdrawal rate (the inflation-adjusted amount that they can take out of their portfolios with virtual assurance that their portfolios will survive 30 years) is often not anything close to the average return earned by their investments. To construct successful retirement plans, we need to know how much the price volatility of stocks reduces the safe withdrawal rate from what it would be in a world in which stocks provided their average return smoothly.
This is the question that safe withdrawal rate research — the research on which retirement calculators are based — seeks to answer. The methodology used in the Old School research is analytically invalid and thus the retirement calculators get the numbers wrong. However, before discussing what the Old School safe withdrawal rate studies get wrong, I need to explain the important insights that they developed as these insights laid the foundation for the even more exciting insights being developed today by the New School safe withdrawal rate research used to develop The Retirement Risk Evaluator, the simple retirement calculator that is the focus of this Google Knol.
The most famous of the Old School studies is the Trinity study. Another notable study is the study done by John Greaney and presented at his www.RetireEarlyHomePage.com web site. Yet a third Old School study is the one done by California financial planner Bill Bengen. The most popular Old School calculator is FIRECalc, developed by Bill Sholar, former owner of the Early Retirement Forum.
The studies look at 30-year time-periods on the thinking that a retirement plan that begins when the retiree reaches age 65 has done the job if it remains in effect until he reaches age 95. The studies are often set up not to provide any slack. That is, the withdrawal rate they identify as “safe” is one that leaves at least $1 in the portfolio at age 95. Were the retiree to live past that age or to end up spending even a slightly larger amount than anticipated, the retirement could fail under the assumptions used in the studies. Retirees who want to include slack in their plans would need to employ withdrawal rates lower than those identified as safe in the Old School studies and calculators.
The studies look at each 30-year time-period going back as far as we have good records of stock performance (some examine the historical data going back to 1870, some only examine the historical data going back to the 1920s on the thinking that historical performance dating back farther than that is not terribly revealing as to what is likely to happen to a retirement taking place in the modern era).
The method for identifying the safe withdrawal rate used in the studies is to examine each withdrawal rate until one is found that does not cause a retirement failure in any of the 30-year time-periods that exist in the historical record. Withdrawal rates above 4 percent produce at least one failure for retirees going with high stock allocations. Thus, these withdrawal rates are deemed “unsafe.” Withdrawal rates of 4 percent and lower generate no failures. Thus, these withdrawal rates are deemed “safe.” It is this finding that is the basis for the famous (infamous?) “4 percent rule” that is commonly cited in the literature and in internet discussion-board threads.
C. Identifying the Risky Safe Withdrawal Rate
The studies generated five important insights, only one of which (the first of those noted below) has been widely recognized.
Insight #1 is that the penalty imposed by stock volatility is high. The safe withdrawal rate for an asset class that provided a smoothly delivered return of 6.5 real would be well in excess of 6.5 percent (the safe withdrawal rate is higher than the return because the safe withdrawal rate calculation assumes that the portfolio balance may be reduced to zero over the course of the 30-year time-period; an asset class providing a zero return would offer a safe withdrawal rate of 3.3 percent under this assumption). But the Old School studies identify the safe withdrawal rate for a high-stock-allocation portfolio as only 4 percent (not much higher a safe withdrawal rate than the one that applies for an asset class paying a long-term return of zero percent).
Insight #2 is that stocks often offer a lower safe withdrawal rate than super-safe asset classes. As noted just above, even an asset class providing a zero return would provide a safe withdrawal rate of 3.3 percent real. In the late 1990s, Treasury Inflation-Protected Securities (TIPS) were offering a guaranteed return of 4 percent real. That translates into a safe withdrawal rate of 5.8 percent real, far higher than the 4 percent safe withdrawal rate claimed for stocks in these studies.
Insight #3 is that it is necessary to remove all slack from a retirement plan to push the safe withdrawal rate for stocks up even as high as 4 percent. The Old School studies are rooted in an assumption that the retirement portfolio may be entirely depleted at the end of 30 years. It is of course possible that a retiree could live beyond age 95 and might want to be sure that something remained in his portfolio account in that event. Or an early retiree might want to use these studies to identify the safe withdrawal rate for a retirement that would need to last 40 or even 50 years. Or a retiree might be concerned that circumstances could develop that would cause him to need to or to want to spend more than he planned to spend on the day he developed his plan. Planning for such contingencies would require using a lower withdrawal rate. Even for those who accept the Old School studies as analytically valid (I do not, for reasons described below), there is reason to question whether the withdrawal rate identified as “safe” truly satisfies the demands of the concept. A truly safe retirement plan contains some slack to cover unexpected developments. It could fairly be said that what the Old School retirement studies and retirement calculators identify is the risky safe withdrawal rate.
Insight #4 is that it is necessary to employ unreasonable assumptions to push the safe withdrawal rate for stocks as high as 4 percent. The Old School studies assume that retirees will not withdrawal even one dollar from their stock portfolios even in the event of a devastating stock crash. Stocks suffered an 80 percent price drop in the years following the 1929 crash. A price drop of 80 percent would cause a high-stock-allocation portfolio of $1 million to fall in value almost down to $200,000 (the loss would be less than $800,000 because the portfolio would not be entirely comprised of stocks) and five years of $40,000 withdrawals would eat away the remaining funds. In the two cases in the historical record in which the issue came up, stocks recovered before retirees following The 4 Percent Rule went under. But it is unlikely that one retiree in 100 would not lower his stock allocation in such circumstances (and these are the very sorts of circumstances that retirees seeking to learn what it takes to plan a safe retirement are seeking to address in their retirement planning efforts!).
This always dubious assumption was shown to be patently dangerous with the reaction of several financial planners to the crash of late 2008. William Bengen, the author of one of the Old School studies, advised his clients to go to stock allocations of zero in the wake of the crash. Taylor Larimore, the author of the book The Bogleheads Guide to Investing, reversed himself on years of preaching that investors should stick with their stock allocations during a price drop, saying that he all along had a “Plan B” of going in such circumstances to a zero stock allocation; never once in his book or in the tens of thousands of posts he put to discussion boards in the years prior to the crash had Larimore let those planning their retirements pursuant to his investment advice know about “Plan B”.
Behavior of which even big name experts are not capable should not be assumed of all investors in studies purporting to identify safe withdrawal rates. The numbers in the Old School studies do not apply in the event that the retiree sells even a single share of stock in the wake of a price crash. It could fairly be said that what the Old School retirement studies and retirement calculators identify is the very risky safe withdrawal rate.
Insight #5 follows from consideration of the earlier four insights: Investing “experts” are loathe to acknowledge the extent of the dangers of investing in stocks in “studies” published in the heat of out-of-control bull markets. The true safe withdrawal rate is clearly sometimes a number a great deal lower than 4 percent.
The authors of the conventional studies are to be applauded for highlighting the Peter Lynch fallacy; it is generally accepted today that the safe withdrawal rate for retirees heavily invested in stocks can be as low as 4 percent. But I think it is fair to say that the Old School studies are the product of a huge pro-stock bias on the part of the researchers (presumably one of which they are not entirely conscious). Choices made in the development of the methodology used in the studies were consistently ones that push the safe withdrawal rate for a high-stock-allocation portfolio higher than what it it would be if more reasonable assumptions were employed.
The finding that the safe withdrawal rate for stocks is 4 percent was shocking. Had the studies been developed in more scientific ways, the findings generated would have been knock-your-socks-off alarming for Buy-and-Hold stock investors.
D. Science Goes Out the Window in Bull Markets
I come not to bury stocks but to praise them.
And I come not to bury the experts responsible for the Old School retirement planning studies but to praise them too.
I know that I sound critical. I need to sound critical because it is imperative that these studies be corrected. There are millions of middle-class investors who are likely going to suffer failed retirements in days to come as a result of the analytical errors (both those noted above and the even more serious one to be discussed below) driving them. But I also believe that in fairness it needs to be said that the researchers were constrained from doing accurate research by a powerful force — the widespread belief in the Buy-and-Hold Model for understanding how stock investing works.
Buy-and-Hold became dominant in the 1960s and 1970s. Yale Professor Robert Shiller published research in 1981 showing Buy-and-Hold to be the opposite of what works (Shiller’s research shows that the key to long-term investing success is not avoiding stock allocation changes but being willing to make those allocation changes demanded of those aiming to keep their risk profiles roughly constant when valuations change dramatically) . Leaders in the field found it impossible to accept the implications of Shiller’s research and have held back from doing so to this day (please see my Google Knols entitled “Why Buy-and-Hold Investing Can Never Work” and “The Bull Market Caused the Economic Crisis” for background). During this time-period serious questioning of the Buy-and-Hold Model was viewed as grounds for social ostracism (Shiller has pointed out that those who understand the effect of valuations often fail to share all they know because the many numbers-obsessed experts who work in this field view it as “unprofessional” to consider emotion-based factors [all overvaluation and undervaluation is caused by investor emotion]). The watered-down findings of the Old School research were viewed as bad news from a marketing perspective by many in the environment created by the most out-of-control bull market ever seen in U.S. history. Had the safe withdrawal rate researchers followed their examinations of the historical data where the science led them, their findings would have blown the roof off of the house. They should have done so. They would have been heroes had they done so. But asking perfection of any of the flawed humans (that’s all of us) is asking a lot. Many of us would have felt temptations to employ methodologies that slanted the results in favor of stocks had we been put in similar circumstances.
Stock investing is an intensely emotional endeavor. Scientific research is theoretically an objective endeavor. The effort to employ science in the development of a better-informed investment analysis was an important and promising one but one comprised of traps for the researchers participating in it that were not recognized at the time the effort got underway. The short version of my message here is — Had the researchers told the blunt truth about what the historical data says about the safety of stocks in a retirement portfolio in the middle of an insane bull market, they would have been hung from a tree! Please read up on the reaction that I have seen to the work I have done in this field if you have thoughts that I am exaggerating. I know whereof I speak re this matter.
The Old School research was published at a time when stocks were (nearly) universally loved with a burning passion. The job that the researchers took on was to report dispassionately on the message of the historical data re the safety of stocks in a retirement portfolio. The message is extremely discouraging for stock enthusiasts. Cognitive dissonance kicked in and they did the best they could in exceedingly difficult circumstances, which was a good bit better than anything that had been done by anyone coming before them but a good bit less than what we need to demand if we are to help aspiring retirees craft retirement plans with good prospects for long-term success.
E. The New School Safe Withdrawal Rate Research
The problem that the researchers faced is that the data says that stock are a truly terrible investment choice for retirees. But that cannot be! We cannot steer retirees away from stocks. To do so will delay their retirements by many years. There must be another way!
There is another way, a way that does not require the dangerous (and — let’s be blunt — ethically dubious) fudging evidenced in the Old School studies. The other way is to reject the premises of the Buy-and-Hold Model and look at the question of the riskiness of using stocks in a retirement portfolio in a fresh way.
Retirees need to invest in stocks. The cost of avoiding the high returns associated with stocks is too great to do otherwise. But the data shows that price volatility is so great that stocks are not suitable for retirees. We need to figure out a way to smooth returns for retirees and thereby become able to recommend that they invest in stocks without continuing to deceive them about what the historical data says.
It can be done! The answer is to let retirees know that they need to avoid stocks when their price volatility presents a real danger while investing heavily in stocks when this is not the case. Buy-and-Hold is rooted in a premise that stock returns are not predictable. This premise was discredited by Shiller’s research but the discrediting was ignored by Buy-and-Hold advocates concerned that acknowledging it would mean rewriting their books and restructuring their calculators and generally acknowledging that they did not know all there is to know about stocks going back to the first day they began studying the matter.
Shiller’s research has been largely ignored by researchers with an emotional attachment to stocks because stocks have been insanely overvalued for almost all of the time-period stretching from 1996 through today. During times of insane overvaluation, taking valuations into account in investment research makes stocks look not as exciting as they would look if the research ignored this critically important factor.
But prices have crashed in recent years and are likely to crash again sometime over the next few years in the event that stocks continue to perform in the future anything at all as they always have in the past (stock prices have dropped to half of fair value in the wake of all previous trips to insane levels of overvaluation — the economic destruction that results when millions of investors have no idea of their true wealth and thus engage in millions of ill-considered spending and saving decisions has in the past always caused enough panic to bring stock prices far lower than what they would have been had they not for a time been pumped up to unsustainable levels). Considering valuations at times when stocks are insanely underpriced has the opposite effect of considering valuations at times when stocks are insanely overpriced. When prices are low, analytically valid research makes stocks look better than they appear using the analytically invalid methodology used in the conventional studies. After the next crash, there will not be much marketing purpose served by misstating the safe withdrawal rate. We may soon be entering a time when stock analysts will feel free for the first time since investment analysis became a scientific endeavor to report accurately the numbers that we all use to plan our retirements.
It will then be socially acceptable for all researchers to consider the effect of valuations. Those already doing so today are the pioneers of the New School of safe withdrawal rate research.
F. The Safe Withdrawal Rate Varies with Changes in Valuation Levels
What if Shiller is right, what if valuations really do affect long-term returns? If that were so and if we gave ourselves permission to report the numbers accurately, our troubles would be solved. If that were so (there is now a mountain of data showing that it is — but sssh! Don’t upset the “experts” by saying this out loud!), it would be possible to tell retirees a way to invest in stocks without taking on much more risk than they would be taking on by investing only in super-safe asset classes like TIPS and IBonds and Certificates of Deposit.
Shiller’s research shows that valuations affect long-term returns. If that’s so, then the value proposition provided by stocks changes with changes in valuation levels — stocks do not provide the same returns in exchange for an investor’s willingness to take on risk at all times but greater returns in exchange for taking on less risk at some valuation levels compared to others. The implications are far reaching, far reaching enough to identify Shiller’s finding as the most exciting breakthrough in our understanding of how stock investing works in history.
What if we identified the valuation levels at which stocks make sense for retirees and advised them to invest heavily in stocks only when they were available at those valuation levels? Our problem would be solved. Retirees could still take advantage of the juicy returns offered by stocks because the extreme valuation time-periods that cause all the trouble are rare and would never apply for the entire length of a retirement. But they could do so by taking on only a fraction of the risk that they are required to take on when investing pursuant to the Buy–and-Hold model (which encourages investors to ignore valuation levels when setting their stock allocations and thus to remain at the high stock allocations that make sense at times when stocks are priced reasonably even when stocks are priced to crash).
Taking the price at which stocks are selling into consideration when setting your stock allocation is a way of smoothing out the return offered by this asset class. An investor who invests heavily in stocks at times when they are selling at low prices will obviously earn a return higher than the 6.5 percent average long-term return, which is the return that applies at times of moderate prices. Obtaining a return higher than the average return during times of low prices counters for the lower return obtained at times when stocks are selling at dangerously high prices and the investor is forced to move to safer asset classes until stocks are again available at prices that do not require him to take on levels of risk that he is not able to tolerate. In contrast, the Buy-and-Hold investor at some times is taking on far less risk than what is suitable for someone of his risk tolerance and at other times taking on far more risk than is suitable for someone of his risk tolerance, obviously not an ideal approach to managing risk.
It turns out that the best way for an investor to manage risk is to be willing to take into consideration the biggest factor affecting risk, the factor that may not be taken into consideration for marketing reasons by those in The Stock-Selling Industry at times when stocks are insanely overpriced.
G. FIRECalc Tells the Tale
Are you up for an illuminating mental exercise? Please open a second window on your internet browser and pull up the FIRECalc retirement calculator in it. Enter a withdrawal rate of 5 percent or 6 percent, one sufficiently higher than the purported safe withdrawal rate of 4 percent to generate a good number of retirement failures. Look at the retirement failures generated and see if you can identify a common theme.
The common theme is — they all are retirements that begin in years of high valuations. It is not price volatility in general that causes retirement failures, it is only price volatility that evidences itself at times of high valuations that causes trouble. Why? Price volatility that takes place at times of reasonable valuations is a harmless phenomena; in the event that volatility causes stocks to drop to prices much below fair value, the Reversion to the Mean phenomenon will in a few years cause them to come back — the losses suffered in price drops from fair value are temporary. In contrast, price drops suffered at times of high valuations are permanent; they are a paying back of artificial price gains obtained by a borrowing from future returns. It is not the ownership of stocks that makes a retirement plan risky. It is the ownership of high-priced stocks that make a retirement plan risky.
This makes perfect sense. Think what stock overvaluation signifies. It signifies that the nominal price being cited is not the accurate price. When stocks are priced at three times fair value, as they were in January 2000, a portfolio of $1 million does not provide $1 million in long-term buying power. It provides about $350,000 in long-term buying power. The other $650,000 is cotton-candy nothingness fated to be blown away in the wind as the stock price works its way back to fair value (Reversion to the Mean is an “Iron Law” of stock investing, according to Vanguard Founder John Bogle).
In 1982, stocks were priced at one-half fair value. Someone retiring at that time with a portfolio nominally valued at $1 million possessed stocks with a long-term value of not $350,000 (as did the retiree who retired with a $1 million portfolio in 2000) or $1 million (as did the retiree who retired at a time of fair-value prices) but of $2 million. A retiree starting with a portfolio value of $2 million is obviously in far better shape than a retiree starting with a portfolio value of $350,000. Is there any argument that can be made that the same withdrawal rate is safe for both of these retirees? My feeble brain is not capable of imagining one. Perhaps I need to spend more time learning about Alpha and Beta and Yabba Dabba Do!
The Old School studies claim that a 4 percent withdrawal is safe for retirements beginning at all possible valuation levels. But the idea that there could be any one safe withdrawal rate that would apply at all valuation levels is a logical impossibility in the event that valuations affect long-term returns. The only way that the Old School studies could be said to get the numbers right is if the Efficient Market Theory were proven out. This theory, which posits that investors always price stocks properly, was popular among academics in the 1960s and 1970s but has been discredited by the last 30 years of research.
H. The Valuation Level Applying When the Retirement Begins Is the Biggest Factor Affecting Retirement Safety
The valuation level that applies on he day the retirement begins is the single biggest factor bearing on the safety or lack thereof of the retirement plan. The Old School studies, developed in accord with the premises of the Buy-and-Hold Model, ignore this factor. Thus, the studies get all the numbers wildly wrong. And the retirement calculators based on these studies also get all the numbers wrong. We need a newfangled sort of retirement calculator!
Enter The Retirement Risk Evaluator. (Please open The Retirement Risk Evaluator on the window in your internet browser used earlier to open the FIRECalc calculator).
This simple retirement calculator’s default numbers (the numbers that appear before you enter any numbers of your own for the calculator to process) show the safe withdrawal rate that applies in four scenarios. To make the four sets of results comparable, the same assumptions have been chosen for three of the factors considered by the calculator (the return for the non-stock asset class, the stock allocation percentage, and the portfolio balance required at the end of the 30-year time-period examined by the calculator). The only difference is that Scenario One reports the safe withdrawal rate that applies for a retirement that begins at a time when the P/E10 value is 8 (an insanely low valuation level); Scenario Two reports the safe withdrawal rate that applies for a retirement that begins at a time when the P/E10 value is 14 (the fair-value P/E10 level); Scenario Three reports the safe withdrawal rate that applies for a retirement that begins at a time when the P/E10 value is 26 (an insanely high valuation level); and Scenario Four reports the safe withdrawal rate that applies for a retirement that begins at a time when the P/E10 value is 44 (the P/E10 value that applied in January 2000, the highest valuation level on record in the United States).
The safe withdrawal rates are: (1) 9.13 (the low valuation scenario); (2) 5.41 (the fair-value valuation scenario); (3) 3.12 (the high valuation scenario); and (4) 2.02 (the January 2000 valuation scenario). Given these results (obtained by running a regression analysis on the historical stock-return data to determine the effect that valuations have had on long-term returns throughout the historical record), I think it is fair to say that valuations matter when putting together a retirement plan.
I. Retirement Planning Is A Community Endeavor
I also think it would be fair to say that any retirement calculator that fails to include an adjustment for the valuation level that applies on the day the retirement begins is analytically invalid and needs to be corrected before it causes more failed retirements. I urge both all experts and all ordinary investors to insist that the authors of the discredited studies and calculators correct them promptly. I also ask your assistance in publicizing The Retirement RIsk Evaluator. We all benefit when aspiring retirees learn how to plan their retirements more effectively. Please help get the word out.
Please scroll down the calculator page to read background on how it was developed and on how it works. At the bottom of the page, there are links to several articles providing more in-depth guidance. You also might want to check out RobCast #189, “The Retirement Risk Evaluator,” in which I discuss in some depth the workings of the calculator and the benefits it provides. After checking out those materials and working the calculator a bit, you may have questions or comments or suggestions. Please forward them to me. The Retirement Risk Evaluator was developed with help from hundreds of my fellow community members. Our work has not necessarily come to an end. I’d like to see us make the calculator an even more powerful retirement planning tool.
I also would be grateful if you would do what you can to help us reopen the many boards and blogs that have adopted a Ban on Honest Posting on safe withdrawal rates to sincere and helpful and informed discussions of retirement planning and other important investment-related topics. Each failed retirement is a failure not only for the investor whose retirement goes bust, but also for the entire community of investors who failed to demand better of those claiming expertise in this field. Experts should be able to get the basic retirement planning numbers right. I mean, come on!
We all benefit from being able to engage in honest and informed discussions of investing with each other. We all have a role to play in changing the environment in which investing advice is offered so that our understanding of how stock investing works continues to improve over time.
You can help!
Please get involved!
And —
Good luck with your own retirement planning efforts!
[…] a positive but incomplete and thus flawed step forward. Rob Bennett has written an article titled The First Retirement Calculator That Gets the Numbers Right</>. His bio is […]