3-12-01

Everyone is searching for a method to control investment withdrawals so that the investments will support a retiree until death. Presumably the ideal method would:

Maximize the amount of money a retiree could spend each year.

Increase the amount each year to offset inflation.

Provide for needed large purchases or emergencies.

Leave no money (or otherwise a specified amount) to heirs, lawyers, and the government.

Of course the method to accomplish this has not, and never will be, invented. The reason is that no one can foresee the future, so investment returns, taxes, and your own personal needs will vary each year.

We’ll describe some of the common methods used by retirees to establish how much they can spend and offer some comments from the perspective of a 67 year old who has worked with retirees from age 50 to 96.

In fact, this is what I find most people have done in the past and still is the basis for the vast majority of retirees, even those who are unable to articulate their financial budgeting discipline. In the past, this method was conservative because it depended on high dividend stocks, e.g., utilities. Although it sounds contradictory, it is still a conservative approach in the long run if there is a significant stock component in the retiree’s portfolio.

The problems with this approach are as follows:

It doesn’t apply to 401(k)s, IRAs, and other deferred tax investments.

Stock dividends have generally gone down as a percent of stock price over the years as the companies put more emphasis on growth which requires more internal investment and therefore leaves less cash for the stockholder.

For that reason, retirees continually seek fixed interest investments with higher interest rates. Fixed investments (bonds, mortgages, CDs, bank accounts) really take it on the chin as inflation compounds over the years. As retirees get older, they get more conservative which further aggravates the problem.

Mutual fund dividends also include capital gains dividends—even on bond funds. Then there is the question whether the retiree ought to spend those distributions. If they do, they can be in real trouble because such dividends often come with a fall in the mutual fund prices, so their portfolio is doubly cursed.

2. Withdraw the same percentage from the fund every year.

This approach often was recommended in short and simple retirement articles. It had more validity when the world turned a lot slower, inflation was extraordinarily low, and investments were largely bonds or other fixed income investments. In today’s world, it is unlikely to preserve relatively constant purchasing power. It exaggerates the problems associated with running out of money too early if you choose too large a percentage or leave too much to your heirs if you choose too small a percentage.

In fact, most planners effectively recommend a small percentage when you first retire and a much larger percentage as you get up into the 80s or 90s. At 65 you might just draw 4%. At 90, you might draw 15%.

Vanguard’s "Plain Talk" says the following about 3% and 4% annual withdrawals:

"3% You're in good shape. . . You may have to cut spending or return to work if investment returns are unusually low or inflation is unusually high. Some retirees would find this situation uncomfortable."

"4% You're close to the limit of safe withdrawals. . . .be prepared to reduce spending or return to work. . .Many retirees would find this situation uncomfortable."

Note that it doesn’t say anything about withdrawing different percentages at different ages. This is the same concept recommended by many charities that offer trusts. If you use a small initial percentage, the amount you can withdraw annually increases if you are a shrewd investor. However, when you die, you leave a lot in the bank unspent—which is great for the charity but shortchanged you if you are interested in using all of the precious money for yourself.

The other problem with this approach is that people tend to go to a more conservative allocation when they get older, so what might have been an acceptable percentage for the first few years of retirement is out of whack with the allocations in later years.

This is probably worse than the previous alternatives because annuities usually have such high costs and the payments are devastated by inflation as the years go on. For example, in the first ten years of my retirement, the real purchasing power of my fixed pension (annuity) reduced by over 30%-- and that’s in a period of low inflation. That said, there are people who are completely financially inept that can benefit from this—if they will also save a little each year so there is a kitty to help purchase some things after inflation’s deadly toll.

4. A very simple approach that almost anyone can use.

Each year you make a new calculation to determine how much you can spend by adding the following three components:

a) all of your after-tax social security or a COLA pension,

b) your after-tax fixed pension multiplied by your age as a percentage, e.g., 60% at age 60, and

c) each year you can spend your previous year’s investment balance divided by an estimate of the years you have yet to live. This is essentially the same budget you would get with the old IRA minimum distribution rules with the recalculation alternative, but it can be used much earlier than the mandatory 70 ½ age. The investment balance must be an after-tax value and should not include reserves needed for known large value items and some kind of allowance for emergencies.

(This approach is equivalent to assuming a zero percent real return which, in turn, is equivalent to a higher than 80% success rate for a fairly wide range of portfolio allocations.)

This is the method taught to all CFPs. To make the calculation you need your current investment balances, life expectancy, an estimate of your long-term real return, and a specialized financial calculator. This doesn’t necessarily give better results than the other approaches unless the returns are appropriate. I find that most planners just use long-term index returns and long-term inflation rates. Worse yet, they may just pick a number out of the air like 7% return and 3% inflation which nets to 4% real return. That means they fail to account for transaction costs that reduce returns and reverse-dollar-cost-averaging. Since few programs account for a portfolio that gets more conservative every year as you get older, you’ll have to estimate the real return for some allocation that will be represent the average allocation for the rest of your life. Even more important, virtually all planners fail to account for reverse-dollar-cost-averaging that takes a minimum of 1% off the returns as commonly quoted. If you want something like an 80% success rate, you’ll have to take off about 3% from a balanced portfolio. (See p. 254 of J. K. Lasser’s Your Winning Retirement Plan.) The problem here is that historical returns are always based on the growth on $1 instead of being based an initial deposit with regular withdrawals.

The sources might be Financial Engines, T. Rowe Price, or a planner using a Monte Carlo analysis. This gives you a better appreciation of the uncertainty in making projections, but does not necessarily give you a better answer than using method 4 or 5. There are several problems that are still common with Monte Carlo analyses as follows:

a) They are all open loop, that is, their models don’t account for the fact that in real life, a new retirement planning calculation should be made each year using last year’s investment balances. Instead, they assume expenses will always go up with inflation and not change as a consequence of whatever happened to investments. Further, Monte Carlo analyses most likely don’t account for an annual allocation changes as retirees get older.

b) Some don’t account for transaction costs.

c) Some don’t use real returns, that is inflation adjusted returns, and therefore ignore the relation of returns to inflation. It is inappropriate to use unadjusted returns that come from years of very high or low inflation and then use inflation values that don’t correspond to the time period for each return.

d) All use fake statistics. Paul Samuelson, one of MITs Nobel economic prize winners is reported to have said that financial Monte Carlo analyses have only have one data point (i.e., only one history) so they can’t really have a statistical representation. Analysts make it look like they have decades of data when many of the funds in the data base are really only a few years old, and they leave out funds that were miserable failures and no longer exist. They ignore the effects of continued political pressures (e.g., Greenspan), previous tax rates, or historical events that tip the market because analysts can’t define Greenspan or upsetting events mathematically. Finally, there is practically no chance that any of the simulations would contain the returns from actual years in the correct order—so in effect, Greenspan, tax history, and historical effects be damned.

This is a correct application of method 5, accounts for reverse-dollar-cost-averaging, and adds a new twist to retirement planning technology. (It’s old hat to aerospace engineers.) Not only do you do a new analysis each year, but you bias your analysis with the prior year’s result and inflation. An aerospace type would say that you added an outside loop of additional feedback. There is great mathematical similarity between an airplane flying in gusty air and a retiree spending in a volatile market. An airplane’s autopilot makes the flight smoother. The Retirement Autopilot makes budgets more stable so there is less year-to-year variation. Otherwise, all of the methods above give unacceptably large variations in annual budgets. It’s not just that a 20% change (for example) in the market means 20% less budget from investments, it’s that without the autopilot, there is a cumulative effect that either gives substantial temporary budget increases or leaves little for late in life. In an airplane, either would lead to a crash.

There are a variety of programs that are good for retirees on this web site. I highly recommend the Post-Retirement Perspective program for most retirees. It’s easy to use and gives good results. You can include more detail than the forms included in J. K. Lasser’s Your Winning Retirement Plan though it uses the same methodology. If you have the time, you can get a step up in sophistication with the Retirement Autopilot Pro. As its name implies, it was intended for professional use, but we have so many other people using it, that you too may want to look at a typical printout on the Web site before you select one or the other. You may benefit from the more detailed yet broader alternatives, and you may like to keep a long-range plan in your file to check your progress as the years go by. We also offer the Strategic Retirement Planning III program, but most people don’t know enough about the theory of returns to use it effectively. It’s definitely better for a pro, especially one who wants to impress clients with detail.

Planners generally fail to point out that you can’t predict the future. All that you can do is to say that if the future is like the past, this is what you should do. In fact, I feel so strongly about this point that all of my recent "perspective" programs on www.analyzenow.com just use two periods of history for retirement simulations. The first is one starting in 1950 that gives what I consider to be optimistic results by historical standards. The second is a simulation starting just ten years later in 1960. That is representative of the bad things that can happen. I recommend that people base their plans on a 1960 case. Remember Murphy’s law: if something can go wrong, it will. That really applies to retirees who have no way to recover short of begging for money from the kids, a charity, or the government.