Fine-tuning portfolios with computer models

Your Money

WANT SOME advice? Ask your broker where she's getting hers.

Last month, the NASD began allowing brokers to use computer simulation programs that forecast, for example, the probability that a client's strategy will cover his retirement needs.

That widened an active playing field. Financial planners, advisers and investment firms' Web sites offer a slew of these computer models directly or indirectly to clients.

Today's most popular model, Monte Carlo, was developed decades ago for scientific analysis but has roared into financial planning in recent years as costs have declined.

Monte Carlo-based retirement calculators are available free on some Web sites, including www.troweprice.com. Many more are available on subscription sites and through advisers.

Essentially, the models take a nest egg and a desired monthly spending allowance, and mix in thousands of potential market outcomes to determine the probability that a portfolio will outlast its owner. By manipulating some of the variables, such as the amount of stock versus bonds in a portfolio, investors can boost their odds of not running out of money.

But it is only a model based on past events, not a fortune teller, and to rely too heavily on a Monte Carlo model is as much a gamble as its name suggests.

Officials of the NASD (formerly the National Association of Securities Dealers) emphasize that any claims about the models must adhere to rules about advertising results.

"When you present a tool as having made projections on what a portfolio will do, you need to have a basis for saying that or you've violated the advertising rules," said Marc Menchel, general counsel for the NASD.

That's good advice for investors to keep in mind, whether they use a broker or a financial planner. Continually testing the assumptions that went into the model and updating it with new life events, such as a job change, is vital, advisers say.

"It's like the difference between specific directional maps and pictures of broad territories," said Ross Levin, an Edina, Minn., financial planner and president of Accredited Investors Inc. "Monte Carlo analysis is really about territories, but investors tend to use it like a map."

It's hard to blame them. With so many variables over a lifetime - shifting tax and inflation rates, demographics, economic trends, changing market growth forecasts - the concrete numbers give investors something to grasp, a target to hit.

"The tools help people begin to see how important the size of their nest egg is in determining how much they're going to live on in retirement," said Christine Fahlund, a senior financial planner for T. Rowe Price.

To get the most out of these computerized tools, question the assumptions going in, realize that the calculations change drastically if the assumptions change and use the outcome only as a general guide, planners said.

Susan Bradley, a Palm Beach Gardens, Fla., financial planner and founder of the Sudden Money Institute, recently started a software program for retail investors (and another for planners) that lets users plug in highly detailed assumptions to determine whether their financial plans work.

The program costs about $90 with shipping and can be ordered through Bradley's Web site, www.suddenmoney.com. This summer, the program will be updated to include a Monte Carlo mathematical model, Bradley said, but even without that it offers planning Monte Carlo can't accomplish.

"This will allow you to input variable tax rates on different parts of a portfolio, real estate and business assets and charitable trusts," Bradley said. "It's much more complex" than analysis tools found on the Web.

As an example, Bradley's site runs through a situation based on a 39-year-old client who wanted to take $300,000 from an inheritance and use it to shore up her accounts while she quit her corporate job to pursue a passion in the nonprofit area.

In the end, the simulation recommended delaying her quitting date by five years and then delaying her ultimate retirement date from the nonprofit by another five years and juggling some spending. Trying to mimic that situation through the free T. Rowe Price calculator, I couldn't make the numbers add up to justify her quitting.

The lesson, says Bradley, is to derive broad ideas from the models, not specific targets.

"In this case, we have a grieving woman ready to quit her job," said Bradley. "What the computer models do is to help the planner make a case for building in a little time before the action is taken, and what kind of sacrifice it's going to take."

E-mail Janet Kidd Stewart at [e-mail].