Fred Williams' Sept. 1, page 38 article ("Bernstein diversification") on Bernstein's all-bond approach to stable value management raises a very difficult question: How does one determine the relative attractiveness of bonds to guaranteed investment contracts? Often answers to this question are based on unsound analysis, biased in favor of either GICs or bonds. While several analytical problems are easy to correct, other issues are complex and subject to debate.
Our research shows that with careful consideration, interested parties (consultants, managers, plan sponsors) can make a fair comparison by breaking down both the assets (GICs and bonds) and the management styles (buy & hold and active) to their fundamentals. The net result is that GICs and bonds are very similar securities. The approaches to managing them, however, are very different. These differences can be analyzed to help determine which style is most attractive. However, a definitive answer is not yet available.
The assets
GICs are like bonds are like GICs are like bonds.
Most GICs and bonds promise a fixed interest rate and have stated maturities. At maturity, both return the investor's principal. Ignoring for a moment its benefit responsive features, a bullet GIC is identical to a zero-coupon bond. However, due to differing conventions in the GIC and bond market, comparing these securities can be a bit confusing. The main differences are:
*GIC rates are quoted using annual effective yields while bonds quotes are based on semiannual yields (called bond-equivalent yields). This understates a bond's true yield by about 10 basis points.
*Ratings agencies rate both GICs and bonds. A proper comparison requires similarly rated securities in similar sectors. GIC rates should be compared to finance sector bonds, which tend to have yields that are about five basis points higher than other corporate bonds.
*GICs are illiquid, except, of course, for responsive withdrawals. Since illiquid bonds (such as private placements) typically have higher yields than similarly rated liquid instruments, some adjustment seems appropriate.
*The benefit responsive aspect of GICs must be considered. GIC rates should be for non-benefit responsive contracts, or bond yields should be reduced by the cost of benefit responsiveness (current wrapper pricing is between 10 and 15 basis points).
*Because most bonds pay semiannual coupons, a proper comparison should adjust for differences in average life (or duration in bond terms) between GICs and bonds.
In making these adjustments, we find there is little difference today in the pricing of GICs and bonds.
Historical return comparisons also are chock-full of complications. For example, on Dec. 31, 1986, the 5-year Treasury rate (annualized) was about 6.9%. The return of the Ryan 5-year GIC index for January 1987 (the first period reported) was about 11.8% on an annualized basis. Naively, this implies the GIC portfolio enjoyed a nearly 500 basis point spread to 5-year Treasuries. Of course, this is true only because high-yielding GICs from 1981 to 1986 are included in the GIC portfolio's return. For this reason, comparing the returns of a GIC index to the returns of a bond index is not an apples-to-apples comparison. In our opinion, a fair comparison is possible only on a marked-to-market basis -- and marked-to-market returns for GICs are not readily available. Our analysis shows such comparisons offer mixed results -- GICs sometimes beat bonds, and bonds sometimes beat GICs.
Management styles
With no discernible differences in either current asset pricing or historical index returns, we are left with the differences in management style. Two questions must be addressed:
1. Which type of manager is more likely to find mispricings and add value in terms of return?
2. Does one style hold an edge over the other in providing a cheaper form of liquidity for participant withdrawals?
With regard to the first point there are more bonds available than GICs -- both in terms of the number of issuers and the types of securities. Bond managers can run more diversified portfolios and have more opportunities to find values. On the other hand, many GIC managers are increasing their expertise in fixed-income securities and are beginning to purchase bonds directly. Furthermore, while smaller, the GIC universe appears to provide opportunities as well. For example, higher quality issuers often offer better rates than their lower quality competitors. We believe this is inherent to the nature of the insurance business, and such mispricings are likely to persist.
An important observation is that the GIC manager's buy & hold strategy is not "inactive," or an "index" strategy. They buy (and hold to maturity) the bonds or GICs they believe offer the best value. Unfortunately, there has been relatively little progress made on choosing the proper benchmark for buy & hold stable value portfolios. Once such benchmarks are determined, GIC managers likely will face the greater performance scrutiny applied to their fixed-income brethren.
Finally, there is the issue of liquidity and benefit responsiveness. GIC managers typically have relied on the regular maturity of a GIC ladder to provide liquidity for participant withdrawals. This allowed them to avoid accessing GIC contracts except under catastrophic withdrawal scenarios. As a result, GIC issuers charge little to provide this (catastrophic) benefit responsive protection. The cost of this approach to plan participants is that future returns are affected when maturities are used to fund withdrawals rather than being invested at current rates. Paying withdrawals from maturities maintains the stability of principal, but creates some uncertainty with regard to the blended rate's tracking of interest rates.
With an all-bond approach, the manager pays for participant withdrawals by simply selling bonds. The presence of a book value wrapper ensures stability of principal is maintained.
The real question is -- How will such withdrawals affect future returns? Depending on the design of the book value wrapper, this reinvestment risk may be the same, less than, or greater than the uncertainty of future returns in a conventional GIC ladder.
What should be obvious at this point is that there is no simple answer.
Both approaches can work. The actively managed bond strategy offers significant opportunities for diversification and return enhancement. However, GICs offer opportunities of their own and are likely to continue to do so.