A suitable choice: it is possible to judge objectively whether a given variable-annuity exchange is the right thing to do, based on possible future scenarios.

Hardly a month goes by without the National Association of Securities Dealers' division of enforcement--or some other regulatory organization--accusing a broker/dealer, financial adviser or insurance company of violating suitability requirements by selling a financial product that was not in the best interest of the client. Nowhere has this been more of an issue than in the market for variable-annuity exchanges, also known as Section 1035 transactions, which are named after the relevant portion of the Internal Revenue Code.

Indeed, according to statistics provided by the National Association of Variable Annuities, close to 63% of the $125 billion in sales during 2003 were exchanges of one VA for another. Sales often generate commissions of 3% to 7% of the premium invested and in the event of an exchange, might involve surrender charges to the client in the same order of magnitude. So, it is not surprising that regulators have taken a hard stance on these transactions, and the intense level of scrutiny has filtered through to compliance guidelines and practices.

On the other hand, carelessly dismissing all VA exchanges as being patently unsuitable is not in the best interest of the public either. Many such replacement transactions actually add economic value for the client and their loved ones. If done properly, a transaction can increase the guaranteed death benefits, and more importantly, offer better longevity insurance, which is the raison d'etre of the life annuity in the first place. In fact, one might actually argue that if the old VA has increased in value--and does not have an automatic ratchet--the financial adviser has a fiduciary obligation to exchange the product to step-up the death benefit to market values, assuming the fees do not outweigh the benefits. The case is even stronger when the new VA--the one being exchanged into--has a better overall package of guarantees.

Yet, many broker/dealers complain that the level of scrutiny imposed on VA exchange--especially at the large firms where a clean reputation is most prized--far exceeds the scrutiny that would be imposed on placing an 85-year-old person into a high-tech mutual fund. This will only serve to discourage advisers who rationally trade-off financial compensation against time, effort and potential liability, from recommending a product that might actually be in the best interest of society and the public at large.

So how does one go about striking a healthy balance between the perceived conflict of interest--unnecessary churning--without discouraging legitimate transactions? Is there an objective framework or methodology that can be uniformly applied to test or grade the merits of a given VA exchange?

The answer to this question is yes. And, the key to help the industry resolve this issue can be obtained from another corner of the insurance industry, namely the property/casualty market. Over the years, their actuaries and risk specialists have developed statistical tools, under the label of scenario analysis, that quantify the liability implicit in their diverse coverage. P/C companies routinely stress test their exposure by simulating meteorological and geological scenarios and then quantifying these outcomes in terms of dollar payouts. The same ideas can be applied in the VA suitability arena. Each and every contemplated annuity exchange could conceivably be put through a number of forward-looking capital-market scenarios to quantify the dollar payouts. The objective merits of the transaction would then rest on the total number of scenarios over which the new product dominates the old one.

To understand how this might work in practice, one first must think about the underlying economics of variable-annuity policies.

How Variable Annuities Work

Boiled down to the economic essence, a variable annuity is a collection of stocks, bonds and other securities held in large diversified portfolios. But, in contrast to generic mutual funds--their close cousins--they contain a number of implicit financial guarantees, which can be compared to call and put options traded on the derivative exchanges. Yes, the VA also provides for tax sheltered growth over time, which is another feature not available with mutual funds, but the decision to exchange one for another is not impacted by this factor. These financial options guarantee maturity, death and annuitization benefits that can all be viewed as a package or collection of financial options.

From a practical point of view, a variable annuity involves three possible "exit outcomes," each associated with its own guarantee or unique value:

1. Lapse or surrenders. When the VA is surrendered, the owner receives the market value minus any surrender charges, which depend on the number of years the policy was held.

2. Death: In the event of death, the beneficiary is entitled to the greater of the market value or some pre-specified guaranteed amount. Some policies automatically increase their death benefit each year, depending on market performance and prespecified interest rates.

3. Annuitization: Every VA has a guaranteed annuity rate that is based on the mortality tables and interest rates specific to the policy. Some policies have additional riders, such as guaranteed minimum income benefits, which improve the annuitization value.

Once again, think of each of these as a derivative security on an underlying variable. If stock markets collapse, life expectancy increases or interest rates decline, these options will be in-the-money and payout to either the beneficiary or annuitant. And while it is hard to predict whether or not these options will ever pay out, they do have some ongoing minimal value today.

How to Judge an Exchange

Now back to scenario analysis and how to evaluate annuity exchanges. Imagine for a moment that you knew with absolute certainty the annual returns from any stock or bond sub account for the next 25 years. Furthermore, assume you knew your precise age of death. Both assumptions, of course, are quite far-fetched. But if you were certain both of these future scenarios were to occur, it would be easy to determine which of two annuity products would be better. This is how you would do it. You would grow the current value of the account by the various annual returns minus the fees charged for the guarantees and then compare the death benefit under both policies--in the assumed year of death--against each other. The product with the higher value would win this scenario.

Now imagine doing this for 1,000 different scenarios, each one generated by a computer program that shuffles and randomizes past historical sequences so that they all look different, but provide possible outcomes for the next 25 years. The same randomizing and shuffling would be performed for the date of death, except that an actuarial mortality table would be used to determine the possible outcomes and their respective weights. What are we left with? In some of the scenarios the guarantees of the old product will be better, while in other scenarios the guarantees of the new product will dominate. Compare the numbers and you have a probability that product B is better than product A. The same analysis would be conducted for each of the three possible exit outcomes for the VA. Each scenario would have one product winning the lapse value, death value and annuitization value.

"Old vs. New: Which Variable Annuity Resulted in Better Benefits?" displays a stylized example of how such an analysis would be performed when comparing two VA policies, the old one and the new one. Here is how to read the information. Under 750 of the 1,000 scenarios, the lapse and surrender value of the old VA policy resulted in greater financial value than under the new VA. Most likely this was because the new policy's renewed Deferred Surrender Charges schedule--and possibly higher ongoing insurance fees--meant that the amount of money the policyholder would be able to extract would be lower. In contrast, in the event of death, under these same exact 1,000 scenarios for the future, the new VA provided greater financial value compared with the old. This was a result of more refined (read: valuable) death benefits, ratchets, etc. Finally, if the policy were annuitized, the old VA would result in more value since the implicit rates were better, etc.

In total, if we weigh each of the three possible outcomes by the weighting listed in the table, we arrive at the fact that under 630 scenarios, the old VA was better, compared with 370 scenarios under which the new policy was superior. So, where does that leave us? Is the exchange from old to new a suitable one? The probability the exchange will add economic value is only 37%; therefore the transaction is inappropriate and unsuitable. It should not be done.

If, on the other hand, the weighting on the death outcome is increased to 90%, while the weighting on lapsation and annuitization is reduced to 5% each, simple math leads to 176 scenarios under which the old product is better and 825 under which the new one dominates. Is this suitable? Since 80% of the time the policyholder is better off, a strong case could be made for suitability of the transaction. Ideally we would like to see 100% of the scenarios leading to improvement, but like all things in financial life, risk and reward are linked.

Weighting the Outcomes

Given the critical impact of this outcome weighting, where does one obtain the appropriate numbers to use? In theory they could be agreed upon between the client and the adviser. If this leaves the process open to potential conflicts, then a better approach is to use actual industry experience numbers for the percent of VA policies that are surrendered, annuitized and paid off at death. These weightings would obviously depend on the client's age and health and would be keyed off a relevant mortality table. The older the client, the greater the probability of death and hence the more weight placed on death. In fact, the foregoing numbers roughly correspond with the weights for a typical 60-year-old buying a VA. Of course, a more refined approach would be to measure not only the number of scenarios in which A is better than B, but perhaps sum up the cash flows as well.

A transaction should be analyzed by focusing on the probability the exchange will add economic value. If this number is greater than some mandated threshold--90%, for example--then the transaction passes the statistical litmus test and is approved.

For regulators to buy into this way of thinking about suitability and transactions will take time and education, but can be achieved. In fact, a number of regulatory entities are in the process of exploring this way of thinking.

In sum, the definition of economic suitability is quite complex--and perhaps beyond the scope of a mathematical model--yet all parties have an interest in developing a rigorous and objective basis for determining whether a particular transaction adds or destroys economic value.

Old vs. New: Which Variable
Annuity Resulted in Better Benefits?

Under 750 of 1,000 financial scenarios, the
lapse and surrender value of the old variable-annuity
policy resulted in greater financial value
than under the new variable annuity. In the
event of death, under these same exact 1,000
scenarios for the future, the new VA provided
greater financial value compared with the old.

                      Weighting    Old VA    New VA

Lapse & Surrender:          75%       750       250
Death:                      20%       100       900
Annuitization:               5%       950        50

TOTAL:                                630       370

Source: Moshe A. Milevsky

Key Points

* Although regulatory agencies are carefully scrutinizing the suitability of variable-annuity exchanges, many of those exchanges add economic value for the client and their loved ones.

* Using scenario analysis, it is possible to determine objectively whether the old annuity or the new annuity is better for the client.

* In scenario analysis, variable-annuity end outcomes should be weighted according to the client's age and health and the market value of the client's account relative to guaranteed benefits.

Moshe A. Milevsky is a finance professor at York University's School of Business and the executive director of the Individual Finance and Insurance Decisions Centre at the Fields Institute in Toronto, Canada.