When all is said and done, the single most critical judgment I am called upon to help my clients make is the rate of portfolio withdrawal that will maximize their retirement experience while still providing enough tangible visibility for them to anticipate a lifetime of security, however long that may be.

Traditional approaches to this question involve a set of calculations with respect to the portfolio, which precipitates an appropriate withdrawal figure. In effect, whether the models use Monte Carlo simulations or average rates of return, they say to the client: "You have accumulated this much money, which will permit you to spend so much each year. Have a nice life." I have some technical misgivings with each of these approaches, but more importantly, I think there is a more human, more client-respecting way to go about this difficult and critical task.

I like to begin not with the portfolio, but with the client's lifestyle goals, then see if we can develop a portfolio and a plan for withdrawals that will meet those needs with sufficient certainty.

Real Answers For Real People

Tom and Beverly Stone will retire in January 2001 at age 65. They have accumulated $1.5 million. Except for Social Security benefits, they have no other visible means of support for the uncertain number of years that stretch out before them. We have worked out a retirement spending plan with the Stones that will require an annual cash outlay of $110,000, including their tax obligations. This defines the lifestyle to which they are looking forward. Social Security will chip in about $20,000, and their nest egg has to provide the rest.

Like most of the people I work with who are approaching retirement, Tom and Beverly articulate their main financial goal this way: "We want to maintain our current lifestyle for as long as we live without having to worry about money." What they do not have as a goal is to die with a nest egg that has the same buying power it has today. Leaving money to their children would be nice, they say, but it is not what they would call a goal.

We explained to the Stones that there are two unknowns that make retirement planning complicated. The first, of course, is that we do not know any particular client's life expectancy. I often joke with prospective retirees that if they would just tell me the date of their expected demise, I could draft an absolutely perfect plan. So far, no client has wanted a perfect plan badly enough to commit to an expiration date for his tenure on the planet. So, we must work around this significant unknown. The most accepted approach is to plan so assets will last for some finite period beyond a statistical life expectancy or beyond a statistical life expectancy modified by family history.

A 65-year-old has an actuarial life expectancy of something less than 20 years. The Stones' ancestors have consistently not done as well. So, we agreed that planning for, say, age 90 should be quite conservative. "If we're still kicking at 90, let the kids take over," Bev said playfully.

The second unknown, of course, is the future returns from a diversified investment portfolio.

A decade ago, the standard way of coping with investment uncertainties was to make projections based on the historical average returns for stocks and bonds, blended in whatever proportions the advisor and client considered appropriate. The obvious risk of this approach is that a client could start retirement at the commencement of an extended period of subnormal investment returns. If his or her withdrawals were based on the more optimistic historical average returns, by the time the markets got their energy back, the nest egg would already be significantly depleted and the client would be contemplating a steady diet of peanut butter.

More recently, the declining cost of computing power has permitted the use of Monte Carlo simulations to model the uncertainties of future market returns. Typically, these complex exercises generate two significant numbers: a prescribed dollar withdrawal amount and a percentage expressing the degree of confidence that this will not entirely deplete the asset during the client's lifetime.

Undoubtedly, Monte Carlo is a big improvement over the average return approach. It can examine hundreds of possible portfolios, incorporating the historic correlation of various asset classes as well as hypothetical returns varied over the years across a range suggested by historic experience. Nevertheless, the Monte Carlo approach has some constraints that, I believe, limit its usefulness in advising real people.

One limitation (at least in the versions I have seen so far) is that it presents conclusions such as "invest 55% in stocks and withdraw $1,500 a month" without a graphic presentation that inspires confidence in the client. Remember, there are two issues being addressed: The client wants to withdraw as much as possible, within the constraints of his or her long-term financial security. So, the confidence aspect is extremely important.

Another softness I perceive in Monte Carlo simulations is this business of declaring a percentage of confidence that the withdrawal will be sustainable for "x" years. If an advisor tells her client that there is a 70% chance this will work out, what do you suppose goes through the client's mind? "There's a 30% chance I will run out of money before I stop breathing!" If I told the Stones that they should live on $60,000 a year instead of $90,000 so they could be sure their money would outlast them, would I being doing them a favor or sacrificing their dreams on the altar of academic propriety? We are dealing with the future, the unknown, and I think that assigning mathematical degrees of certainty to the output of a black box may be inappropriate.

My experience convinces me that people who are old enough to retire have seen enough of life to understand that risk goes with the territory. Most of them would rather enjoy the early years of their retirement, when their knees still work pretty well, and take some risk that, if the markets are worse than expected, they will need to cut back on something. We usually suggest modeling a reduction in projected living expenses at some point, perhaps age 75. Sometimes, we do it by simply eliminating the inflation factor at that age, in effect beginning a gradual reduction in "real" spending. When it seems reasonable to our clients, we actually reduce projected spending by 15% at some future age to permit greater spending in the early years; it is just more realistic. In all cases, we counsel our clients that it is important to revisit these long-term projections every year or two for a reality check.

The new retiree's greatest financial risk is that the first so many years of retirement will be terrible years for the investment markets. We wanted to create an easy-to-understand model that stared this risk right in the face. If we could get past it, our clients could have the level of confidence they need to make the withdrawal decision. We call the model our Bear Market Endurance ModelĀ©.

Bear Market Endurance Model

Using the Stones' data, we have displayed nearby the spreadsheet model that we used to determine a stock/bond mix capable of providing the $90,000 annual withdrawal that their planned lifestyle requires. One Monte Carlo presentation we looked at would guide them to a first-year withdrawal of about 4.3%, or $64,500, with a 95% certainty. In sharp contrast, our model suggests that the Stones' portfolio can support the 6% ($90,000) beginning withdrawal rate on which our clients have their hearts set.

Our presentation focuses on the annual withdrawal, which is what the client is focused on. It also shows a year-by-year portfolio value in three radically different stock-market scenarios to demonstrate a range of possible outcomes. Our goal is to support their lifestyle even in the worst-case scenario; that way, surprises are more likely to be pleasant ones.

Our worst-case scenario is an entire decade when stock appreciation averages only 2% a year. We purposefully placed the worst-case markets in the first 10 years of retirement, because this is when it would be most disruptive to our clients' plans.

It wouldn't really matter how this 2% average return happened. It could result from a 1973-74-style debacle in years one and two, followed by eight years of recovery. It could come from alternating good and bad years, or just 10 consecutive years of pitiful 2% returns. Or it could surprise us by beginning with a couple great years and then correcting!

The reason we can say that it wouldn't matter how the 2% average return unfolded is that we designed this portfolio with enough fixed-income investments so that withdrawals could be maintained for 10 years without touching the stock portfolio. That means our clients could ride out an extended period of subnormal stock returns, even if it included a monstrous bear market, without having to change their lifestyle!

Since the end of 1930 (we think it is appropriate to leave out 1929 and 1930), overall stock appreciation has averaged 8.8% a year. Total returns (appreciation plus dividends) have compounded at a rate of 11.5% from 1931 to 1999. How did we decide on a decade of 2% average appreciation for a worst-case scenario?

Since 1931, there have been 60 running 10-year periods (1931-40, 1932-41 etc.). The very worst decade for total stock returns since 1930 was 1965-1974. The average annual total return for that awful period was 2.4%. The second worst 10-year period in the past 69 years was 1969-1978, when total returns averaged 4.7% a year.

In our model, we use 1.5% for cash dividends the first year, which gradually rises to 3% over 25 years. Adding to this the 2% annual stock price appreciation, our total return for our worst-case decade averages 3.7% a year. This is nearly midway between the worst and the second-worst of 60 ten-year periods going back to 1931. Since our clients do not want to base their retirement withdrawals on the expectation of another Great Depression, this seems an appropriately sober hurdle for worst-case analysis.

From our client's point of view, the really good news is that they can withdraw $90,000 (6% of their portfolio) the first year and expect with very high confidence to sustain their lifestyle at that level for the rest of their lives.

Models Need To Be Updated

In real life, it is highly unlikely that we would advise our clients to make all their withdrawals from the fixed-income portfolio and never buy or sell stocks for 10 years. This model is designed to see at what stock-allocation level they could have complete confidence in maintaining a certain lifestyle no matter what was going on in the market the first 10 years, which are the years when their lifestyle is at greatest risk.

The likelihood is that, if the decade ahead averages 2%-a-year appreciation, the average will be made up of both significant declines in the stock market and major recoveries as well. Volatility, like the poor, we have always with us. Because we use a rebalancing discipline, volatility in stock prices should actually help improve their overall investment results by providing opportunities to buy low and sell high at the margin.

Finally, there is certainly the possibility that the markets will do better than our worst-case model, perhaps significantly better. While it is our job to worry about the downside, and that does give our clients a quiet confidence, the column on the far right of the model in which stocks keep growing at 10% a year is not without interest. Have a good life!

J. Michael Martin, JD, CFP, is president of Financial Advantage in Columbia, Md.