Prior to this legislation, the Internal Revenue Service required use of the 30-year Treasury bond as the key interest rate for pricing liabilities. Under the new act, the discount rate will be based upon a blended composite of three corporate bond indexes, ranging in maturities from 10-plus to 15-plus years and ratings from A to AAA.
In essence, the act replaced one bond with more than 400 bonds. Among the major problems with this portfolio construction:
1. Each index is market weighted, not equal weighted, to create an average yield. This means that the yield of each bond has a different weight, even if it has the same maturity or duration.
2. The three indexes are equal weighted to create an average composite yield. So we have a blend of market-weighted indexes into an equal-weighted composite index. This seems to violate the index methodology rules — either market-weight or equal-weight, but not both. Moreover, this composite double and triple counts many securities that appear in more than one index.
3. Each index uses different price sources. Without a bond exchange or generally accepted closing price, there is a disparity in prices and yield on the same bonds on the same day. This is much more acute with corporate issues than Treasuries that are priced real time on broker screens and have large volume trades frequently. As a result, tracking error among these three indexes will be prevalent.
4. Financial Accounting Statement 87 states that high-quality bonds should be used for discount rate pricing. This blended composite will be 33% A rated. Such a rating is usually not considered high quality, but investment grade.
5. No zero-coupon bonds exist in any of the three indexes, which reduces their longest duration bonds to about 16 years. This shortcoming makes it difficult to price liabilities longer than 16 years. Only zero-coupon bonds connect present values to future values accurately. Only zero-coupon bonds have durations out to 30 years. The only high-quality zero-coupon bonds available today are government securities.
6. It is impossible to purchase this average index yield or index portfolio for most of the liability payment dates. There needs to be a rule that says, "If you can't buy it, you can't use it."
7. While the Treasury and prominent practitioners promoted removal of the four-year weighted-average methodology to a three-month or current market status, the new act made no changes. As a result, the blended index rate will be based on the same formula as before: a 48-month weighted average, giving the more recent of the four 12-month periods progressively higher weights. If rates stay at current levels, this blended corporate rate would calculate to 6.12% for the four years ended Dec. 31, 2004. The longest Treasury (5.375%, due Feb. 15, 2031) closed May 31 at a 5.4% yield. If interest rates rise by 115 basis points by year end, the long Treasury would out-yield this blended corporate rate! Skewing this calculation to old rates only provides relief when interest rates are in a secular bull market. If we go into a prolonged bear market, the current rate for long Treasury will out-yield this amalgamation of corporate bonds!
8. Pension liabilities are like snowflakes — no two are alike. The act continues the improper procedure of pricing all liabilities at one single discount rate. This is regrettable. Moreover, this action seems to violate FAS 87, which says to be faithful to this rule, use individual discount rates. The Financial Accounting Standards Board even gives examples to suggest you price each liability off the exact maturity/
duration on the spot curve that matches the liability payment date.