Home > Library > Stable Times > Volume 4, Issue 2  

Stable Value: Challenging the Reach of Asset Allocation Models

The latest and hottest innovation to hit the defined contribution plan market is proving to be a lukewarm event for the Stable Value industry. The innovation is automated online investment advice for DC plan participants. By considering information about a participant's age, income, investment goals and appetite for risk, these Internet-based services can spit out a custom-tailored asset allocation plan for that person complete with fund-specific recommendations about how to execute them. Nearly all of the major plan providers now make them available.

The problem, according to Stable Value experts, is that these advice services use widely divergent methodologies to model Stable Value Funds, often with less precision than would be ideal.

"Some treat Stable Value as a money market fund, others as a bond fund, others as a combination of the two," observes Wayne Gates, General Director of John Hancock Financial Services and chairman of an SVIA task force looking into the issue. "But none of these approaches gives Stable Value full credit for its key attributes."

Gates says the root of the problem is the belief by many advice providers that standard deviation of participant returns-the risk measure commonly used in their computer models-does not capture the full risk of Stable Value Funds. By virtue of their book-value guarantee, of course, Stable Value Funds generate returns that are much less volatile than those of the typical short-term bond fund, even though they invest in similar securities. But many advice providers discount the value of the wrap contracts used to provide the book-value guarantee. As a consequence, they tweak their models to assign greater risk to Stable Value investing than its returns would warrant.

"The end result is two things," Gates says. "One, investors who use these models often get lower allocations to Stable Value than they would if it was modeled correctly. Two, they probably end up with lower equity exposure than they should have, and more fixed income exposure, because neither cash nor bonds are as good a diversifier relative to equities as Stable Value is."

Paul Lipson, chief investment officer for the Federal Reserve Employee Benefits System, says he's thus far declined to make any of the commercial advice services available to participants in his organization's $2.6 billion defined contribution plan because of his concerns about how they would model its Stable Value Fund.

"What these consultants will tell you is that in their view, Stable Value is the equivalent of a money market fund, and they have the capability for modeling that," Lipson says. "In fact, I do not think Stable Value investing is the equivalent of money market investing, and I think the inputs have to be unique."

The numbers tell the story. According to Lipson, a portfolio of 90-day Treasury bills held for 10 years ended June 30, 1999, would have produced a compounded annual return of 5.5% with a standard deviation of 0.5. During the same period of time, the Lehman Brothers Government/Corporate index of intermediate-term bonds would have earned 8.1% with a standard deviation of 4.4. Meanwhile, the Stable Value Fund in Lipson's defined contribution plan earned 8.9% with a standard deviation of 2.2.

"What we've just done is define Stable Value," Lipson says. "Stable Value over 10 years gives you an intermediate-bond return with much less risk."

Lipson sees another problem with the current crop of advice engines in that the mean-variance modeling methodologies on which they are built were originally developed for use by traditional pension plans, and so employ mathematics that assume very long investment horizons, a single investment objective and a completely tax-free environment. But individual investors who participate in defined contribution plans have complex and multiple financial goals with varying time horizons, and must pay taxes on their withdrawals from those plans.

Gates says the advice providers are making progress toward better modeling of Stable Value Funds, even if they haven't overcome all of the hurdles. "They're getting a lot closer," he says. "For example, the people who run mPower (one of the leading advice providers) have learned a lot about Stable Value in the past few years, and recommend it quite frequently."

Lipson, though, is waiting for still more progress. In the meantime, he has developed an asset allocation modeling program for his own DC plan participants that suggests how the commercial advice providers might want to approach the task.

Like most of the mean-variance models that are at the core of the commercial advice engines, Lipson's asset allocation program is built on three inputs: projected returns for each asset class, a projected risk level for each asset class, and the correlation between the returns of the various asset classes. To come up with projected returns for his Stable Value Fund, Lipson looks at its actual returns for the past 10 years and then determines the spread between that number and the 10-year return for 90-day Treasury bills. He then layers that spread over the current T-bill rate and uses that as the return going forward.

Lipson's model uses the historical standard deviation of his fund as its risk measure, not so much because he considers that ideal but because it remains the industry standard. In the future, he would prefer that such models use a measure that gives a higher weighting to downside volatility and a lesser weighting to upside volatility.

Although most Stable Value managers argue that asset allocation models should use inputs for the specific Stable Value Fund available to each investor using that model-and Lipson is emphatic that this is necessary-Gates' task force is nonetheless working to develop a methodology that would allow for the modeling of a generic Stable Value Fund, too. Gates says the point of the generic isn't to use it as a substitute for actual fund data in those instances where it is available. Rather, it would be available for use in the generic asset allocation models that are found on many financial Web sites, most of which do not address Stable Value products at all.

"I think it's a reasonable thing to do," says Gates. "If someone is using one of those generic sources to get an idea about how to allocate money to their Stable Value Fund, and Stable Value doesn't show up on the menu, they're not going to know what to do with it."

The task force hopes to complete a white paper on the issue during the summer of 2000.

Read Next: What Asset Classes and Funds Should Plan Sponsors Be Adding in 2001?