Monte Carlo modeling yields magic number for retirement
BY TERRY SAVAGE SUN-TIMES COLUMNIST Jul 14, 2006
Updated: May 3, 2013 12:14PM
Originally published: July 8, 2004
Are you worried about The Magic Number? Two weeks ago, I wrote about Fidelity’s new Retirement Income Advantage program, which helps you calculate the amount of money you need to retire -- the Magic Number -- as well as how to allocate your investments during retirement, and calculate withdrawals to make sure you don’t run out of money before you run out of time.
That’s the focus of my next book, The Magic Number, from John Wiley & Sons next spring. The concept of “Monte Carlo” modeling used to make those forecasts has generated intense interest. I’ve been asked whether computers are “predicting” the future, or “guaranteeing” the future, or simply gambling with the odds on the future. The answer: none of the above. And it’s not magic, either.
Monte Carlo modeling is simply the use of a sophisticated computer program to assess multiple variables and come out with a range of probable outcomes. It’s more sophisticated than just using averages of past performance to make future predictions.
Sam Savage, Stanford University professor and Monte Carlo simulation expert (and no relation), explains the concept in an article called “The Flaw of Averages: Plans based on average assumptions will be wrong on average” (his italics).
Savage gives a graphic demonstration. Picture a drunk walking down the center of a two-lane highway filled with speeding traffic. He weaves back and forth across the center line. On average, he’ll be right on the line. In real life, he’ll be dead! Obviously averages can be very misleading -- and especially so when it comes to retirement planning.
Joe’s ‘average’ retirement plan
Several years ago, when T. Rowe Price introduced Monte Carlo retirement simulations for its mutual fund clients, it gave an example that I use in every speech. I call it the story of Joe, who retired at age 65 in 1969, with $250,000.
Back then, Joe went to a financial planner who advised him to put 60 percent of that retirement money into a diversified portfolio of stocks, 30 percent into bonds, and 10 percent into a money market fund -- a fairly conservative allocation. The planner took a look at the historical average return of these asset classes, and told Joe he could expect an average annual return of between 10 and 12 percent over the next 30 years -- by which time Joe would be 95 years old, if he lived that long.
With the benefit of hindsight, we know the planner was correct in his projection. The portfolio actually returned an annual average of 11.7 percent between 1969 and year-end 1999. But based on that estimate, the planner had told Joe he could withdraw 8.5 percent of his money every year -- and not run out for 30 years. That was based on the average returns -- which did come through as predicted.
But in real life there was a bear market early in Joe’s retirement, and the Dow fell from over 1,000 in 1969 -- and stayed below 1,000 until 1983. With Joe withdrawing 8.5 percent of his account every year, he ran out of money in 11 years!
Getting to Monte Carlo
Sam Savage calls Monte Carlo modeling “a computerized cure for the Flaw of Averages.” If you go to his Web site, www.AnalyCorp.com, you can actually “test-drive” a Monte Carlo simulation and understand the process, which he explains in such a simple form that even non-geeks, like me, can follow along. If you are enthralled, you can order his computer program, XLSim, for Excel spreadsheets so you can do your own modeling. It’s like having the magician teach you his tricks.
But you don’t have to do it yourself to take advantage of the benefits of Monte Carlo simulations for retirement planning. Vanguard (800-VANGUARD), T. Rowe Price (800-638-5660) and Fidelity (800-903-6563) all offer this service. At Fidelity, it’s free until year-end. You’ll learn how to diversify your investments in retirement and get a range of withdrawal scenarios to meet your comfort level.
You might be wondering if all these programs will give you the same results and recommendations. Actually, no. A lot depends on the input -- including the length of the historical data that’s used to make the projections in the models. And the recommendations will be different, too.
One thing is sure: You’re a lot more likely to find your Magic Number using these planning tools instead of guesswork or averages. And that’s The Savage Truth.
Terry Savage is a registered investment adviser, and appears weekly on WMAQ-Channel 5’s 4:30 p.m. newscast. Distributed by Creators Syndicate.