Figuring how much an insurance policy will be worth is still tricky

``Garbage in, garbage out.'' Put another way, say the computer buffs, if a computer makes a mistake, it's probably because a human being fed it some bad information.

The phrase could also be used by people shopping around for life insurance. It's especially apt for policies that feature some sort of earnings from investments. When an insurance salesman or broker is asked how much a particular policy will earn and how much the investment side of it will be worth in 10, 20, or 30 years, the answer depends on what information goes into the equation.

Until a few years ago, all of these projections were made in the insurance companies' home offices. Here, room-size computers (or, in earlier days, rows of actuaries) would come up with standard dividend scales to be used by all the salesmen. For years, this system worked fine; projections were based on years of experience with interest rates, which were fairly stable, and life expectancy, which was getting longer every year.

Today, the picture has changed. Instead of big computers programmed by conservative actuaries, small personal computers can do everything their giant ancestors did. Almost every insurance salesman has one on his or her desk, or at least in the office, and can make policy illustrations based on whatever projections they like. If a broker wants to present a computer printout of policy earnings based on an average interest rate of 9 percent over the next 30 years, even though the parent company is using an 8 percent basis, you'll see the projection based on 9 percent.

The art of making these projections has come into greater demand in recent years with the rise of universal life insurance. These products separate the insurance protection and investment parts of the part of the policy. While part of the premium pays for protection through term insurance, the rest goes into an account like a mutual fund and is invested in stocks, bonds, or money markets, as specified by the policyholder.

If the projections by the insurance company or salesman are not accurate, policyholders could one day find there's not as much in the savings side as had been expected. Or if they were planning to use the proceeds from the investments to pay the premiums one day, there may not be enough money to do this. It's even possible premiums might have to be increased to keep the policy in force.

Bad projections aren't limited to universal life, either. Premiums on whole life, which stay the same every year, are also based on the insurance company's projections of what it can earn with premium payments, estimates of mortality rates, and how much it pays in commissions and salaries. Because whole life doesn't separate the insurance and savings features of the policy, however, customers aren't as likely to see the effect of inaccurate projections.

``This has been a source of confusion for 20 or 30 years,'' says Robert Hunter, president of the National Insurance Consumers Organization. ``It was worse with whole life. Universal life has actually been better.''

In a recent article in a journal published by the American Society of Chartered Life Underwriters & Chartered Financial Consultants, Edward P. Mohoric, an actuary, and Paul Silberberg, an attorney, discuss three major factors consumers need to review:

Interest rates. While the rate is important, it's just as important to know how the company comes up with it. There are two approaches for measuring earnings. One is called the new money rate; the other is the portfolio rate.

The new money approach treats each year's investments as a separate portfolio and credits interest based on that block of business alone. The portfolio method lumps together an entire group of years and credits interest based on an average for the entire period.

One method isn't necessarily better than another, the authors note, but the new money approach gives higher earnings in an increasing interest rate environment, while the portfolio approach does better when rates are falling as they have been for the last year or so.

Mortality. These charges are built into the polcy to cover the deaths of polcyholders. While some companies assume an increase in life expectancy, others have begun to scale down these expectations, particularly those firms that believe they have a significant number of policyholders who are at risk of contracting acquired immune deficiency syndrome.

``AIDS isn't a problem for insurers yet, but it will be,'' Mr. Hunter says. ``It will skew the mortality tables.''

Consumers, however, should ask what happens if life expectancy turns out better than assumed.

Expenses. ``There is no such thing as a no-load insurance product,'' the authors say. Even if a policy has no apparent sales charges, the company has to pay its people somehow, and these expenses come out of the premiums you pay. How much comes out depends on the company.

A company with very low overhead might be able to keep its expenses to 3 or 4 percent, though some companies go as high as 9 or 10 percent, Hunter says.

You should also look at the company's track record of living up to interest rate and earnings projections. In most cases, it's probably best to avoid the highest rate or the best earnings outlook you're offered; this is more likely to be based on faulty projections from a computer that's been fed a diet of bad information.