Beginning almost 20 years ago, an obscure advisor in El Cajon, Calif., published a series of papers defining sustainable withdrawal rates from retirement portfolios. Based on a historical reconstruction of retirees' portfolios since 1926 (with respect to asset class returns as well as inflation), these papers recommended a 4% withdrawal rate for the first year, followed by cost-of-living adjustments every succeeding year. This applied to tax-deferred portfolios seeking a minimum 30-year "longevity."

Later on, by using additional asset classes in his analysis, the advisor increased his recommendation to a first-year withdrawal rate of 4.5%. However, the conclusions of the original research persisted, and have been popularly enshrined, for better or worse, under the moniker of "The 4% Rule."

OK, I confess that I was that advisor. I have been observing my research play out in the markets ever since, like Dr. Frankenstein studying his creation. In the late 1990s, as the bull market neared its peak, there was criticism that the recommended withdrawal rate was too low, that retirees were left with huge account balances late in life and could thus have spent far more than 4.5% initially.

Recently, I have heard the opposing concern expressed, namely that the terrible stock market returns since 2000 have so damaged retirees' portfolios that the 4.5% rule must be reduced, perhaps back to the original 4%, or even something less (I have heard as low as 1.8%).

One lesson I learned in my research is never to assume anything. (Which number rule of L.J. Gibbs is that?) Time and time again, as I opened a new door in my studies, I was surprised by what the numbers revealed if I merely let them tell their story.

In this essay, I am going to undertake three tasks:

1. Compare the status of the original "worst case" scenario (the investor who retired in 1969) with a prospective "worst case" candidate, the 2000 retiree. Who was better off after 12 years of retirement?
2. Explore possible future conditions that might lead to a "blowup" of the 4.5% rule.
3. Consider some alternative strategies for a retired client if the 4.5% rule appears it may fail.

1969 Versus 2000
When I conducted my first research in 1993, there were only 38 complete 30-year retirement periods available for study in Ibbotson's yearbook (1926 through 1963). Today, there are 57 periods (from 1926 to 1982). This is hardly an exhaustive sample, and the case can clearly be made that no sample, no matter how large, would provide conclusive results since markets change over time. However, I maintain that the 87 years beginning with 1926 saw a wide range of market and economic conditions, including wars, depressions, oil shocks, "great moderations," etc. Surely they contain something of value as precedent.

The individual who retired on January 1, 1969, was the "big loser" in my analysis. Employing the 4.5% initial withdrawal rate, his portfolio was the only one of the 57 to exhaust itself at the end of 30 years. Chart 1 depicts the annual nominal account balances for a 1969 retiree who began with $100,000 in his (not-yet-invented) IRA account. (See Figure 1.)

As you can see, the portfolio had retained almost its full original nominal value of $100,000 as late as 1989, the 21st year of retirement. But thereafter, the value rapidly collapsed. This seems counterintuitive since a huge bull market began in 1982. Why weren't returns sufficient to sustain the portfolio?

The answer is contained in Chart 2, which records the annual inflation rates (CPI) for the 1969 retiree. Research has confirmed that the "sequence of investment returns" is crucial for portfolio longevity; a retiree with low returns early in retirement will probably have trouble later in retirement. What is often overlooked is that the same reasoning applies to the inflation rate. Since, under the "lifestyle maintenance" model used in my research, the retiree boosts his withdrawal amount each year by the prior year's inflation rate, high inflation early in retirement can result in a rapid escalation of withdrawals and depletion of the portfolio. (See Figure 2.)

That is clearly what happened to the 1969 retiree. Inflation rates from 1969 to 1980 averaged almost 8% per year, far higher than what we are currently experiencing. This forced the retiree to make large increases in his withdrawals almost every year (at a time when investment returns were generally subpar). As we shall see below, the effective withdrawal rate thus spiraled upward at an unsustainable pace. Even the 17% stock market returns of the 1980s and 1990s couldn't rescue him.

Now that we have a more detailed understanding of the "worst-case" scenario, let's compare the 1969 retiree to somebody else who would be most likely to experience sustainability problems: the person who retired on January 1, 2000. This latter retiree has suffered through an unprecedented occurrence: two stock market declines (in the S&P 500 index) of 50% or more, in the same decade.

Unfortunately, we have only 12 complete years of investment returns since the beginning of 2000 to analyze. However, if we accept the general principle that the experience of the first decade is the most important to the future sustainability of the portfolio, we might still reach some useful conclusions.

First, let's examine the investment returns and inflation data in tabular form for the two periods. At the outset, I want to point out that in all cases I have used a portfolio of 35% U.S. large-cap stocks, 18% U.S. small-cap stocks and 47% intermediate-term government bonds (all the return data is from the annual Ibbotson yearbook). This asset allocation is near optimal for generating the highest withdrawal rate. It is possible that using additional asset classes such as REITs, international stocks and bonds, commodities, etc. might generate higher withdrawal rates. However, the relatively high concentration of small-cap stocks hopefully serves as a fair proxy for these other asset classes. The portfolio is rebalanced once a year. (See Figure 3.)

The "average" investment return in this chart is the geometric, compounded, annualized return, while the average CPI is an arithmetic average of the 12 annual inflation figures. A salient feature of this chart is that the returns for the 1969 retiree were somewhat higher than they were for his 2000 counterpart during the first 12 years, but inflation was correspondingly much higher.

Well, this is interesting, but it's only background. It doesn't really allow us to compare, in an easily understandable manner, the overall portfolio experience of the 1969 and 2000 retirees, although it does hint at the final outcome. Figure 4, I believe, does the job.

Figure 4 illustrates the "current withdrawal rate" at the end of each year in retirement. In contrast with the "initial withdrawal rate," which is measured at the start of the first year, the current withdrawal rate computes the same quantity (the withdrawal amount divided by the beginning-of-year portfolio value) for each succeeding year.

For the 1969 retiree, the current withdrawal rate almost tripled, from 4.5% to 12.5%, during the first 12 years of retirement. A 12.5% initial withdrawal rate is normally associated with a portfolio designed to last only about seven years. That the 1969 portfolio lasted an additional 18 years was due entirely to the huge investment returns of the bull market that began in 1982. This bull market merely postponed the inevitable. The damage done in the first 12 years, primarily from high inflation, was irreversible.

In contrast, the 2000 retiree saw his or her withdrawal rate climb relatively modestly, from 4.5% to 5.9% at the end of 12 years. My conclusion is that, after the first 12 years of retirement, the 2000 retiree is in a much better position than was the 1969 retiree. His much lower current withdrawal rate (5.9% versus 12.5%) gives him far more flexibility-and the foundation for a more optimistic outlook. It appears that the 2000 retiree still has a "fighting chance" of sustaining at least 30 years of withdrawals from his portfolio, although that portfolio may shrink in value over time.

By the way, "portfolio shrinkage" is not unexpected in the face of adverse investment and/or inflation conditions. Unless the client specifically requests that a certain balance remain in his account after 30 years, my research allows for the possibility of the portfolio being completely exhausted after 30 years. Ensuring a non-zero terminal balance requires the adoption of a lower initial withdrawal rate, as might be expected.

The 2000 Retiree: An Uncertain Future
Although in the last section we concluded that the 4.5% rule has not clearly been violated for the 2000 retiree, it is still too early to be certain that it will remain valid. Historically, a portfolio has "blown up" prematurely almost always because of dire investment returns and/or inflation that struck in the early years of retirement, as it struck the 1969 retiree.

In 2012, we are faced with multiple gross market uncertainties. Most of the American stock market (according to the Shiller 10-year cyclically adjusted P/E ratio, the Q ratio, dividend yields, etc.) still appears to be about 35% to 40% overvalued. In fact, the valuation of the S&P 500 index today is very similar to that of 1969.

It is claimed by some that the S&P 500 index today is fairly valued or even "cheap" by historical standards. They usually compute "fair value" by applying some arbitrary multiple (typically 13 to 15) to current or forecast earnings. However, that is not an accurate method by which to compute long-term value, as it ignores the fact that earnings and margins appear to be near peak levels.

Various parties have estimated that nominal returns for S&P 500 stocks, as a group, might be as low as 5% annually (or less!) over the next decade, far below the long-term return (since 1926) of 10% for large-cap stocks. (See Figure 5.)

Furthermore, the current "grand experiment" by central banks, injecting trillions of dollars of liquidity into the financial system since 2007, faces an uncertain conclusion. Some prognosticate that once the deflationary tendencies of our current "deleveraging" cycle recede, we might be faced with much higher inflation, perhaps even "hyperinflation," as a consequence of the extraordinarily loose monetary policies.

Thus, it is possible that for the 2000 retiree, the true test of the sustainability of his withdrawals will not be the early years of retirement, as has been the case for earlier retirees. The crucial period may be the "middle years" of retirement, or the remainder of this decade, from 2012 to 2020. This would be an unprecedented circumstance. The validity of the 4.5% rule might be challenged if troublesome events unfold during those years.

Quite frankly, I am at a loss for a meaningful technique to forecast the upcoming years. My historical, deterministic approach doesn't seem particularly useful, as the time period is short and the uncertainty high. To a certain degree, the same considerations would also seem to plague a stochastic method, such as Monte Carlo analysis.

Thus, we are left with rank speculation, which is an activity I normally prefer not to engage in. In this case, however, it might help us answer a simple question: Are there likely combinations of returns and CPI that might result in a failure of the 4.5% rule for the 2000 retiree?

One approach to answering this question is to merely take 18 years of data chosen from some period in the historical record and "graft" it on to the first 12 years of actual retirement statistics for the 2000 retiree. For example, I wondered, what would happen if the next 18 years match the first 18 years of the 1969 retiree's experience-namely, the investment and CPI data from 1969 to 1986?

Figure 6 illustrates the consequences of assembling this somewhat Rube Goldberg-ish model.

The portfolio of the 2000/1969 composite client expires during the 28th year, 2027. This is short of the initially desired 30 years. However, given that we have cobbled together, back to back, two awful investment periods for retirees (2000-2011 and 1969-1986), this model can be thought of as representing a "worse than worst case" scenario. Nothing in the historical records matches the fearsome combination of investment returns and inflation found in this model. Essentially, it consists of a 25-year secular bear market followed by a five-year bull market.

I view the 27-plus-year longevity of this composite model as a rather hopeful result. In this circumstance, the 4.5% rule would have been violated, but not by much. In fact, a reduction of the initial withdrawal rate to 4.3% would have restored the portfolio's 30-year longevity. That amounts to less than a 5% reduction in withdrawals. Even in the face of such painful circumstances, worse than anything actually previously experienced, the 4.5% rule still turned out to be approximately correct.

Interestingly enough, we could have obtained the same result by taking a slightly different tack. In my research, I computed the portfolio longevities associated with a number of initial withdrawal rates, not just 4.5%.

In 2012, we can think of the 2000 retiree as retiring anew. According to my research, a 5.9% initial withdrawal rate (equal to the current withdrawal rate of the 2000 retiree in 2011) is consistent with a portfolio designed to last a minimum of 16 years. The 2000 retiree's portfolio has already lasted 12 years. Thus, one would expect that, in the worst case, the composite portfolio would last a total of approximately 28 years, without any change in the retiree's lifestyle.

We obtained this same result earlier. Not surprising, since we are essentially using the same framework for analysis, but I wanted to underline another approach advisors can take in studying these issues. Retirement withdrawal plans do not have to be static; they can be modified at any time during retirement. Retirees can be "born again" at any age.

Given all this, I am not ready to abandon the 4.5% rule for my clients. However, considering the gross uncertainties we face, and the possibility that the future could hold surprises more harmful than the "worse than worst case" I examined, I deem it appropriate to consider strategies for dealing with a possible future failure of the 4.5% rule, even a relatively minor one.

Contingency Planning or Retiree Withdrawals
There are only two essential ways to deal with retirement spending problems: either increase income or reduce expenses. I have a bit to say about both.

Reducing Spending
The original intent of my withdrawal research was to find a "comfort zone" in which the retiree could maintain his lifestyle during all of retirement. Thus, it seems anathema to ask a client to reduce his spending. However, as we saw before, in a "worse than worst case," lower withdrawals may be required.

In contrast to the retiree who has stood by his original withdrawal plan and followed a buy-and-hold strategy, there are a considerable number of investors who sold their stocks at the bottom of recent market declines and neglected to reinvest in stocks as they recovered. As a result, these individuals may be experiencing unusually high withdrawal rates, much higher than the rates we forecast above for a buy-and-hold investor.

One critical question presents itself: At what withdrawal rate should a client first consider reducing his spending?
If we return to Figure 4, we see that the 1969 retiree experienced rapid increases in withdrawal rates beginning in the sixth year. Fortunately, this client was bailed out by a huge bull market that began in the middle of his retirement. Despite double-digit withdrawal rates (which persisted from the 10th year on to the end of retirement), he was still able to enjoy 30 years of income from his investment portfolio.

To a certain extent, the 1982-2000 bull market could be considered a freak occurrence, a product of the final years of a giant debt super cycle. It was the biggest and highest-returning bull market on record. Future retirees might not be so lucky to be rescued by such a white knight. Thus, it seems to me, a client should consider reducing spending if his withdrawal rates approach the double-digit realm, particularly if he's doing it early in retirement.

Nevertheless, that approach alone might not be good enough. Consider Figure 7, which plots the annual withdrawal rates for our composite 2000/1969 client. Only the first 25 years of retirement are shown, as the withdrawal rates escalate dramatically thereafter. Remember that this scenario "expired" during the 28th year.

Withdrawal rates never attain double-digit levels until the 18th year of retirement. However, it appears it would have made sense to have taken some pre-emptive action much earlier in retirement, when withdrawal rates rose well above the initial 4.5% level.

I offer the following informal rule for your consideration:
Take some pre-emptive action, no matter how mild, when the current withdrawal rate first exceeds the initial (or expected) withdrawal rate by 25%.

For the 2000/1969 client, that would have occurred during the 10th year of retirement, long before the double-digit regime was reached. Note that this is not intended to be a precise rule, just a general guide for action.

I also offer the following corollary rule: If, despite initial action, withdrawal rates continue to rise, take more aggressive action.
The corollary is probably not much more than common sense. One can always increase withdrawal rates if favorable investment conditions reassert themselves later in retirement. But it's really tough to deal with double-digit withdrawal rates. Don't let them get that high.

One could point out that this rule would have been unnecessary for the 1969 client, as he eventually would have been "bailed out" by the big bull market. However, one can't be sure what's around the next investment bend. At worst, the retiree will leave behind a bit more money than he expected. I submit that's much better than running short.

Increasing Income
I have several alternative strategies to present under this topic. This list should by no means be considered exhaustive; I am focusing on a few I consider particularly interesting:

A. Invest in immediate fixed (or inflation-indexed) annuities: Harold Evensky, a well-known advisor in Coral Gables, Fla., has done much to draw the attention of the financial planning profession to this option. If a portfolio is suffering (or expected to suffer) from low investment returns, it might make sense to transfer some of the investment risk to an insurance company via an annuity. Current fixed annuity rates of 8% for an 80-year-old, and 10% for an 85-year-old, are attractive in a world of subpar returns. If inflation is the major concern, an inflation-indexed annuity might be considered.

The conversion of the client's portfolio could be accomplished incrementally, over a period of years, to hedge the bet on future poor returns. Markets sometimes do surprise us positively! Some clients may object to leaving less to their heirs (I imagine some heirs would object too!), but the client's current solvency is the critical issue.

B. Employ a reverse mortgage: Michael Kitces published an excellent study of reverse mortgages in the October and November 2011 issues of The Kitces Report. Harold Evensky is also promoting this concept among advisors. Many retirees have unused equity in their homes that they take to their graves. Tapping this equity, through a reverse mortgage, may permit them to reduce the withdrawal rate on their investment portfolios temporarily, until investment conditions improve. That may be all that's required to attain the desired portfolio longevity.

C. Change investment methods: My research employed a buy-and-hold investment approach, primarily because that was the simplest assumption needed to obtain meaningful results. So some readers have inferred that I am an advocate of buy-and-hold investing. In certain circumstances, it does appear superior, but in my view, these are not the best times for it.

One reason that "SAFEMAX" (the maximum "safe" historical withdrawal rate) is only 4.5% is that I assumed in my research the investor would ride every bear market down to its ultimate conclusion. Since 2000, that has been a painful experience for investors, with two declines of 50% or more to deal with in the S&P 500 index.

Investors might need to avoid seriously overvalued assets, like the S&P 500 index of 1965, 2000, 2007 and today. There are other asset classes, such as international stocks, emerging market stocks and high-quality large multinationals that offer far better value at this time. I know; this suggestion smacks of "market timing," a filthy term to some, and its practice could cause an investor to lose some upside, as well as some of the downside.

But I believe the profession needs to consider alternatives to buy-and-hold investing under all conditions. It just inflicts too much pain on clients, and introduces too much volatility into portfolios.

In summary, the 4.5% rule (and its infinite variations for time horizon, tax bracket, current market valuations, etc.) may be challenged in coming years. However, it appears to be working now. Rather than discard it willy-nilly, I recommend that advisors consider other weapons in their arsenal, such as the ones I have mentioned, to extend the life of a client's portfolio.

I know we are all trying to do our best for our clients at all times. I hope that I have expressed some ideas you find useful in this regard.