 |
|
||||
| Home | News | Help Desk | Membership | Library | About |
|
| Library | |
 |
|
| Stable Times | |
| Papers | |
| Fee Disclosure Template | |
| Key Principles | |
|
Home > Library > Stable Times > Volume 5, Issue 1  The quarterly publication of the Stable Value Investment Association First Quarter 2001 • Volume 5 Issue 1 Stable Value and Asset Allocation ModelsBy Christopher Cutler, Deutsche Bank The SVIA's Asset Allocation Task Force has discussed how to model stable value with several asset allocation modeling firms, with the goal of helping identify how the benefits of stable value can be covered effectively in their models. Key attractions of stable value are that it offers a rate of return comparable to bond funds and higher than money markets, and has risks that are comparable to money market risks, and certainly much lower than bond fund risks. Defined contribution investors widely acknowledge this attraction by allocating the bulk of their fixed income investments into stable value. Precisely fitting stable value within a portfolio asset allocation model is a bit tricky. We understand that stable value has steady returns and low risk, but how do these properties compare with those of other asset classes? How can one model what stable value returns will look like in the future? How are stable value returns correlated with other asset classes' returns? The challenge is to translate stable value's qualities into the quantitative vocabulary of portfolio optimization methods. Answers to these questions depend on the methodology the model-builder uses. Historical Returns and Forecasting One way to address these questions is to look at historical data to forecast long-term future returns. One may think of this approach as being similar to projecting asset growth rates for defined benefit plans, with an asset classes' projected growth rates being dependent, in part, on their historical returns experience. For forecasting stable value, a challenging issue is that stable value return data is auto- correlated, meaning that the change in one period's return is influenced by the change in the previous period's return. The autocorrelation occurs because stable value amortizes differences between current and credited rates over time. Every statistician is trained to "correct" regressions for autocorrelation. Otherwise, the forecasts will have an R-squared and coefficients that are too high (the forecast says it is more accurate than it really is.) However, the result of "correcting" stable value returns for autocorrelation is likely to reveal returns much like an unwrapped bond fund. Tenor of Investment But should the forecasts be corrected in the case of stable value? While the data is autocorrelated, every participant actually receives a stable return over time with very little risk. Some believe the answer to this question depends on the intended tenor of the investment. If the investor intends to hold stable value for the short-term (5 years or less for example), then the investor will receive the full benefit of stable value's autocorrelation (stable returns, very low risk.) If the investor will hold stable value for very long periods of time such as 30 years, then the investor receives very few benefits of autocorrelation, because the long tenor means that the investor's ultimate return depends primarily on the long-term performance of the fixed income markets. Intermediate-term investors receive some of the benefit. That's a nice way to conceptualize stable value. Unfortunately, we know of no models that can do this! None of today's models adjusts an asset class' risk profile for the type of investor. Every asset allocation model assumes that each asset class has the same risk/return characteristics for any type of investor. The models then determine investor's allocations based on the investor's risk appetite. How Modelers Reflect Stable Value There are several ways to resolve this dilemma. One way is to make an asset allocation model where stable value's correlations with other asset classes depend on the tenor of each investor. Unfortunately this approach would be very difficult to imbed into most asset allocation models, and currently no one is doing this. Another way is to take a more qualitative, traditional investment advisory approach: given an investor's risk appetite, set target allocations for each asset class, where stable value is one of the asset classes. Investors close to retirement will have more in stable value, and younger investors will have less in stable value. A third way is to ignore the fact that stable value is stable value, and just assume that it's just as risky as a bond index fund. This approach yields a poor resolution: it fully ignores the benefits that stable value offers, even to investors with a very short-term horizon. Conclusion As you can see from the sampling of modeling issues this essay provides, there are many ways to model stable value. There is no "ideal" asset allocation model, and each model has limitations. Nonetheless, I believe that plan sponsors and managers must conduct their own due diligence of asset allocation models to make sure they capture the basic properties of plans' investment options, including stable value. Plan sponsors and managers now have an additional challenge to see that the modelers commit sufficient effort to "get it right." If you are considering using an asset allocation model for your plan, how can you approach the due diligence process with respect to stable value? There is no comprehensive guide to examining these models. Here are some initial questions to think about. How does an asset allocation model incorporate stable value? Does it sufficiently consider the benefits of stable value, or does it ignore or discount the stable value feature? How do model allocations between stable value and other fixed income options interact? Is the model likely to recommend stable/robust allocations over time? If not, why? A model with unstable allocations might encourage large cash flow movements, which could hurt stable value investors. If you select an advice model, is there enough coverage from wrappers and GIC issuers who are willing to continue underwriting plans utilizing that type of model? I hope that this discussion provides some helpful insights into examining how asset allocation models approach the stable value option. We can be certain that this subject matter will provide a spirited discussion for years to come.
Read Next: Modified Reporting Requirements for Certain Master Trust Stable Value Funds
|
|
