Liability Index: The True Objective
By Ronald Ryan

Ryan argues that pension plan portfolio managers are benchmarking against market indexes when they should be benchmarking against what is happening to the present value of liabilities they eventually have to meet. Infrequent and delayed liability data are part of the problem. One answer: basing a liabilities index on zero-coupon Treasurys, a standard approved by the FASB for valuing liabilities. That would allow managers to update their investment bogey daily.

The absolute size of liabilities is enormous - have you heard of Social Security? (Now, there's an asset/liability dilemma in progress.) Just the pension fund industry alone has over four trillion dollars in pension assets (defined benefit). If the purchasing power of pensions were viewed as the GNP of a country, it would be the third largest GNP in the world and quickly gaining on the second. For most pension plan sponsors, the pension asset is one of their largest assets, and the pension liability is one of their largest liabilities. Legally, the pension liability is supposed to rank senior to all debt.

However, most institutional asset managers are given a generic index as their index benchmark, which may have no similarity to the behavior pattern of liabilities. Furthermore, actuaries and consultants price liabilities differently than either the market or the Financial Accounting Standards Board (FASB) would confirm as proper. It has become common practice to price all liabilities at the same interest rate regardless of their maturity or payment date. Yet have you ever seen a flat yield curve, at least one that lasted more than a few days or weeks? Too often, this discount rate is much higher than the market rate, causing an underpricing of liabilities. The higher the assumed yield is above market, the lower the present value of liabilities. Just like in golf, the pencil (that keeps the score) may be the most important wood in the bag. Unlike golf, in investment management this can result in underfunding in which the round of drinks at the clubhouse won't be bought by someone else.

Moreover, liabilities are usually analyzed annually, with data up to three months old. Given infrequent and delinquent data, it is difficult, if not impossible, for the asset side to understand or key on the liability side. As a result, most asset managers manage assets versus generic market indices. But what do these generic indices have in common with liabilities? In fact, these indices may even violate the liability objective or risk/reward behavior pattern. What is happening is assets are managed vs. assets (e.g. generic asset indices) and not versus liabilities. By losing sight of the true liability objective, several problems, if not crises, may develop.

LIABILITIES: FASB DEFINITION

The FASB does a thorough job of explaining how any liability works and should be priced. FASB regulates corporate financial statement reporting and is quite clear on how to report and measure pension and medical liabilities. FASB 87, Paragraph 44 deals with pension liability pricing: Assumed discount rates shall reflect the rates at which the pension benefits could be effectively settled.... In making those estimates, employers may look to rates of return on high-quality fixed income investments currently available and expected to be available during the period to maturity of the pension benefits.

Therefore, according to the FASB, pension plans should use high quality, noncallable-for-life bonds that match the maturity date of the pension benefit date. Since virtually the only high quality for life, non-callable-for-life bonds are Treasuries, it follows... pension liabilities should be priced off the Treasury yield curve.

Furthermore, FASB 87, Paragraph 199 says: Interest rates vary depending on the duration of investments; for example, Treasury bills, 7-year bonds and 30-year bonds have different interest rates. The disclosures required by this Statement regarding components of the pension obligation will be more representionally faithful if individual discount rates applicable to various benefit deferral periods are selected.

Therefore, each liability should be priced individually off the Treasury yield curve with the same maturity as the liability benefit payment date.

In 1993, the SEC got involved when it questioned a registrant about the selection of discount rates under FASB 87. The SEC wrote a letter to corporations and the FASB, stating that the guidance provided in Paragraph 186 of FASB 106 (created for medical liabilities), should be used if there is confusion on FASB 87. Apparently, according to the SEC, the present value of each type of liability should be calculated the same way.

FASB 106, Paragraph 186 reads: The objective of selecting assumed discount rates is to measure the single amount that, if invested at the measurement date in a portfolio of high quality debt instruments, would provide the necessary future cash flows to pay the accumulated benefits when due. Notionally, that single amount, the accumulated post retirement benefit obligation, would equal the current value of a portfolio of high-quality zero coupon bonds whose maturity dates and amounts would be the same as the timing amount of the expected future benefits payments.

The FASB needs little or no translation here. Liabilities are to be priced as if they were zero-coupon bonds whose maturities match the liabilities benefit date and whose par values match the liability payment amounts. Since there are only government zero-coupon bonds, it follows: Liabilities = Treasury Zero-Coupon Bond Portfolio.

This should be no surprise, since to defease liabilities we buy only zero-coupon bonds to match liabilities (e.g. lotteries). Many companies appear to take great liberties with the FASB guidelines. Instead of pricing each liability at the market for Treasury zeroes (STRIPS) they use a higher and singular rate for all liabilities. Since horizontal or flat yield curves are not common, the mispricing of liabilities here seems careless or intentional.

Some companies interpret high-quality, zero-coupon bonds to be anything with a rating of AAor higher. Since zero-coupon corporate bonds do not exist, we are dealing with theoretical discount rates. Companies know that AAcorporate bonds have higher yields thereby creating a lower present value on liabilities. Beware of the pencil!

FIRST DILEMMA: ASSET ALLOCATION

The goal of traditional asset allocation is to create the optimal absolute return. As a result, stocks have been favored consistently. However, the true objective of asset allocation should be to create the optimal relative return versus liabilities. Each client has a unique liability term structure. As with snowflakes, no two client liability schedules are identical.

Logically, therefore, the liability term structure should dictate the shape of the asset allocation. Unfortunately, most asset allocation models have no input for clients' liability characteristics. They only analyze historical generic market indices. Without a client's liability structure, such generic models could give most clients similar asset allocations. In fact, this is a national commonplace. Yet it is hard to imagine that any generic asset allocation model can fit most liability schedules. The asset/liability ratio (surplus/deficit) and shape of liabilities differ from client to client, in fact, requiring dynamic asset allocation shifts.

Let's take, for example, client Aand client B. Client Ahas 82% in long liabilities, suggesting an allocation of 82% in long assets, (unless there is an unusual deficit/surplus situation). Client B has only 28% in long liabilities, suggesting an allocation of 28% to long assets, (unless there is an unusual deficit/surplus situation). The term structure of each clients' liability payout schedule should become the base for asset allocation. Asset allocation should be tailored to each client's unique liability term structure and not based on generic models.

Now let's look at the graph on page 32 showing the last 10 calendar years of assets versus liability risk/reward behavior.

The line is the Ryan Labs Liability Index where the present value of the one through 30 year liability payments are equal (equal weighted index). The dots are asset classes represented by the most popular generic market indices. Notice that the Lehman Aggregate has volatility similar to the four year STRIP (four year Liability); the S&P 500 like a 17 year STRIP and EAFE like a 24 year STRIP. The first step of asset allocation should be to define the risk/reward behavior of asset classes to determine what liabilities they should fund (e.g. short, intermediate, long, very long). For example, you wouldn't buy equities to fund the one-year liability because that would be a serious risk/reward mismatch. As the graph clearly indicates, there is a definite distinction of asset classes risk/reward behavior that determine what liabilities they should fund:

Notice that the Treasury STRIPCurve (Ryan Labs Liability Index) covers the complete spectrum of volatility for any asset class. Since the duration of a STRIPis equal to its maturity, the STRIP curve has durations out 30 years. The popular bond indices have durations of five suggesting the STRIPcurve (and very long liabilities) can be six times more volatile given the same interest rate movement. You can find a Treasury zero-coupon bond that has the volatility of any asset class! You would think the value added of any asset class is versus the STRIP with the same or similar volatility. If you lose to the Treasury STRIPwith no credit, event or liquidity risk over long time horizons your value is negative.

Once a Custom Liability Index is built, the term structure of such an index can be updated frequently and clearly. This becomes the foundation of the asset allocation process (e.g. 32% in very long liabilities = 32% allocated to very long assets). Those asset classes within each liability term structure cell now compete for the allocation to that term structure cell. The client may decide, for example, to give 100% of the long liability cell allocation (32%) to EAFE asset managers or a diversified blend of small cap plus EAFE managers. Moreover, a Liability Index Fund for that term structure could be the core asset management device or technique.

SECOND DILEMMA: MEASUREMENT

Historically, assets are given generic indices (e.g., S&P500) as their objective. Their performance is measured by comparing the total return of an asset class (e.g., stocks, bonds) to the generic market index for that asset class. Money managers are hired and fired based on their performance versus an asset index.

An important reason why this situation persists is that liabilities are traditionally calculated annually and reported months after the fact. More importantly, the present value pricing of liabilities is not calculated using market yields. It is extremely difficult, if not impossible, for an asset manager to manage appropriately against such an ill-defined opponent.

What is needed is a custom liability index that correctly prices clients' liabilities (in conformity with the FASB) as a tangible, frequently updated (e.g., daily) system for investment management. The asset side must know frequently the liability growth rate for each term structure cell (e.g. short, intermediate, long). Performance measurement can then be properly assessed as the growth rate of assets versus the growth rate of liabilities for each term structure cell as well as total assets vs total liabilities.

Areview of past performance is very revealing. As the table on the next page indicates, here is the 10 year history of asset growth (returns) versus liability growth. Let's again use the Ryan Labs Liability Index where the one through 30 year liabilities are equal weighted. Using a common and static asset allocation ratio of 5% Cash, 30% Fixed Income, 60% Equity, 5% International shows that assets lost to liabilities five out of the last 10 years. Moreover, 1995 was the worst year. Most practitioners rejoiced in 1995 when the S&P returned 37.57% and your favorite bond index 18%. Ryan Labs was a lone voice crying that 1995 was the worst year in pension history! If a five-year generic bond index grows at 18%, what do you think a 15-year pension liability grew at in that same year? Most plan sponsors witnessed 40% plus growth in liabilities in 1995. Most clients didn't know until after the fact. The truth is that the greatest bull market in American history (1982 - 1998) was also a bull market for liabilities.

THIRD DILEMMA: PRICING LIABILITIES

As explained earlier, liabilities are to be priced as if they were a zero-coupon bond portfolio whose par values match the liability payment amount and whose maturities match the liability payment date. This is not common practice. Traditionally, liabilities are priced at 100 basis points higher than the Treasury STRIPcurve. If you multiply the interest rate difference times the average duration of liabilities you get an estimate of the error in present value calculations:

 (STRIPYield - Yield Used) x Duration = Error in Present Value (5.00% - 6.00%) x 15 = -15%

Accordingly, there is evidence to believe that liabilities are undervalued in ranges of 10% to 30% per year. This is strictly a pricing problem of using inflated discount rates. In time, this will show up as added costs, lower credit ratings and/or deficits.

SOLUTION: CUSTOM LIABILITY INDICES

It is critical that each client's liability objective be properly measured and supported. In 1991, Ryan Labs created the first Liability Index after two years of development. This index is customized and tailored to accurately calculate liability present values and present value growth for any client liability schedule. Asset allocation and performance measurement can thus be properly fitted to the clients' liability term structure. Performance measurements can properly assess asset growth versus the liability growth these assets are funding (e.g. long assets vs. long liabilities).

The following table is a history of assets vs. liabilities using the Ryan Labs generic Liability Index (for educational purposes only) having an average duration of 15.5 years.

As you can see, the bull market for bonds and stocks has been a bull market for liabilities. The Ryan Labs generic Liability Index has grown 149.2% (9.56% annual) over the last 10 years. The volatility of liabilities is significant. The years 1994 and 1995 saw liability growth go from -12.46% to 41.16%.

If asset managers can better understand the risk/reward behavior pattern of the liability opponent, they can better strategize how to outperform such an opponent. If asset managers can understand the term structure of liabilities, they can build proper assets allocations for each term structure area (long assets versus long liabilities). Without accurate liability term structure measurements, clients face the greatest risk there is... mismatching assets versus liabilities by term structure. The S&Lcrisis is too vivid a memory of what can happen with mismatched term structure exposure. The S&L crisis of yesterday could be the liability crisis of tomorrow.