Market participants have generally come to interpret the concept of liability-driven investing (LDI) as being equivalent to liability matching investing (LMI). This interpretation is one that is forcing more and more pension plans to make decisions that do not make economic sense – decisions that may damage the ability of ostensibly well-funded pension schemes to deal with future events with respect to liability valuation or a downturn in markets. In addition, the total value of US defined benefit pension assets is about 3.5 times the USD investment-grade long-duration bond market, making execution of LDI or LMI initiatives difficult. This article proposes an alternative to the LDI approach: investing not with a view to match liabilities, but with a view to explicitly minimize a funding gap (or maximize a surplus). This approach, which we call funding gap-driven investing (FGDI), also incorporates the use of assets other than long-duration fixed income securities. The aim here is that diversification of return opportunities should lead to greater stability in long-term returns as well as greater scope for active management to add alpha in order to provide greater absolute return.

The current historically low levels of rates, while causing the widening of funding gaps as they lowered, now offer an attractive opportunity to implement an asymmetric hedging structure that will automatically narrow a funding gap if rates rise. The current tight level of investment-grade spreads bolsters the argument for incorporating other sources of absolute return. Plan sponsors should aim to hedge losses but not gains. Furthermore, it simply makes sense to diversify sources of return, even in a normalized rate and spread environment.

Liability-driven investing (LDI)
Liability-driveninvesting has developed into the core framework for management of defined benefit pension plans. LDI is based on the premise that pension assets exist to service pension obligations, and this concept has shifted the focus of pension asset management away from a “return at all costs” aim to one that more effectively matches pension liabilities. LDI requires that assets are invested with the characteristics of the liabilities or obligations (e.g., liquidity, duration etc.) in mind.

There are two primary objectives for the investing of pension assets:

1. Maximize the likelihood that the beneficiaries receive their stated obligations in full across all economic scenarios; and
2. Minimize the cost to the plan sponsor of funding these obligations across the entire duration of the obligations.

While not mutually contradictory, these aims are in tension, and the investing of a plan’s asset portfolio within the LDI framework can create an inherent compromise between the two objectives.

Funding gap-driven investing (FGDI)
The aim of FGDI is to engineer an entire portfolio to structurally minimize the funding gap over time while making it robust in a wide range of investing environments. Classic LDI advocates investing in assets with a high correlation to the liabilities, which begs the question: why consciously allocate assets to lose money alongside the liabilities when rates rise when you could instead allocate assets to capitalize on an opportunity to narrow the funding gap?

The needs of the company sponsor and the needs of the pension beneficiaries are relatively simple: the company sponsor wants to know whether there will be a need for a future contribution; the pension beneficiary wants to know if they are going to get paid what was promised. Essentially, what both interested parties are asking about is the funding gap, not whether assets match liabilities. This may seem a distinction without a difference, but it is not.

Management of pension assets only matters in as much as it narrows the funding gap or, ideally, generates an enduring funding surplus between plan assets and obligations. The goal is to ensure that, over the long run, pension plan assets will be sufficient to fund the required cash flows to beneficiaries.

Investors who limit their approach to matching liabilities lock in any current funding gap and ignore the possibility that the funding gap may increase due to unhedgeable events. This is not to say one should ignore the needs of cash flow matching, but rather that the plan’s asset allocation should not be driven by the need to fund short-term cash outflows to the possible detriment of managing the funding gap in the long term.

Rather than simply trying to match the first order determinants of the liabilities, the focus in FGDI is on minimizing the risk of a widening funding gap while maximizing the possibility of narrowing the funding gap. It is not focused on liability matching, but rather on these three ideas:

1. Hedge the hedgeable: duration asymmetry;
2. Minimize sponsor risk: low market exposure; and
3. Prepare for the unhedgeable: diversification of risk premia.

FGDI should ideally also take into account the relation of the funding gap to the corporate sponsor. A portfolio that is structured such that it has a wide funding gap just when the corporate sponsor can ill afford to take a hit to its finances is dangerous for both the corporate sponsor and the beneficiaries. Hence equity risk and credit risk are not very sensible as they correlate to general economic performance, and therefore the ease with which the average company can afford to make top-up payments to the pension plan.

Hedge the hedgeable: duration asymmetry
The duration of pension obligations is so large that even small changes in rates have a large effect on presentvalue calculations. In liability matching investment, changes in value due to duration on the obligation side are matched by a similar change on the bond investments side. The concept of duration asymmetry uses the multiplier of the long duration obligation against itself. The range of outcomes for a swaption hedge is asymmetric and generally favourable, especially in an upwards rate environment when compared with a swap-based hedge.

In practice, the only direction in rates that an asset portfolio should be concerned about is when rates go down, creating a need to keep pace with the rapid increase in the discounted value of liabilities. When there is an increase in the discount rate, the discounted value of the liabilities tumbles, offering an opportunity for an uncorrelated asset portfolio to narrow the funding gap. Using an asymmetric hedge such as receiver swaptions can allow the portfolio to maintain an existing funding gap when rates go down and to structurally narrow a funding gap as rates rise (to a certain degree irrespective of the performance of the other assets in the portfolio). In the current low-yield environment, combining this with an appropriate swaption roll mechanism allows the funding gap to be structurally narrowed should rates rise or become volatile.[1]

Minimize sponsor risk: low market exposure
Relying solely on one single risk factor – the return due to credit risk – is not a sufficiently diversified way to build a cushion of returns to hedge against future changes in the value of liabilities. Furthermore, poor returns from investing in credit (or equity) risk are very likely to be correlated to a weakened financial position on the part of the average fund sponsor (because of the likely overall poor economic environment). Using this as the sole market exposure risk premium does not make sense.

Prepare for the unhedgeable: diversification
Not being able to hedge against non-duration-related changes in liability value is not an excuse to abandon basic investing principles such as diversification. Relying on a single driver of returns is going to provide a wider range of both good and bad outcomes as opposed to diversifying and therefore narrowing the range of expected outcomes using a wider range of return drivers. In the context of seeking to structurally minimize the funding gap, relying on the single driver does not make sense, as investors are basically widening the expected range of returns and, hence, the range of possible bad outcomes. If all investors do is invest in other risk premia with the exact same expected return but some correlation benefits, this will dramatically reduce the scale of the outcomes, if not the number, in which funding gaps widen. Further, they will start to position the entire portfolio such that duration asymmetry can structurally minimize the funding gap without taking directional asset allocation risk.

Implementation of FGDI
It is clear that even the most orthodox implementation of LDI requires some degree of active management, if only to buy bonds that are added to the benchmark and to sell bonds that fall outside of its requirements. Most implementations would include a level of credit picking to try and optimize risk and return. However, the amount of return that could possibly be added by active management is linked to the size of the opportunity set. By one measure[2] there are around just 1900 USD investment-grade bonds in the long duration investable universe. By any measure it is harder to add alpha by active management with this limited opportunity set versus the investable set of securities offered by an actively managed hedge fund exposure which numbers well above 100,000.

One approach to implementing FGDI is to rely on a diversified portfolio of low market-directional, actively managed hedge fund exposures to secure an alpha or skill-based return. Alpha due to active management, while controlling market directional exposure, has shown over time to be a highly uncorrelated source of return. Over market cycles, these exposures have been strong absolute return performers, allowing for steady compounding over time. The successful execution of this approach requires that the portfolio be actively invested, at all levels, as manager skill can itself be volatile. Very tight position-level risk monitoring is also a vital function in this approach; this ensures that the portfolio incorporates as many different risk premia as possible and allows for control of any possible drift into directionality. Constructing and monitoring a portfolio of this type – which can provide a low-volatility, low-market-exposure set of returns that is also robust across a wide range of market scenarios – is a very large undertaking.

Using this as a basis for the absolute return portfolio reduces the need to generate returns by timing the exposure to different asset classes, as such investing is tremendously hard to do consistently. But timing exposure to other asset classes is also a form of risk premium as the fund can provide liquidity in a market drawdown or panic. The successful execution of timing exposure to different asset classes has to be done in a controlled fashion, with a strong understanding of what it means to the distribution of funding gap possibilities in the future. In other words, timing directional market exposure adds a large level of risk to the asset return and needs to be used very cautiously and done when the odds or pricing are very much in the investor’s favour.

Obligations are bond-like but not bonds
A pension obligation has long duration and steady cash flows far out into the future – just like a very long duration bond; however, when you go further out, the actual size of the cash flows gets harder and harder to estimate. The actual duration of these cash flows is also not certain or fixed. Amongst the factors that influence the current estimated value of this obligation are long-term interest rates, mortality risk and inflation (particularly wage inflation).

For long-dated bonds, there are fixed steady and contractual payments that stretch out neatly into the future. The contractual size of the cash flows is the same now as it will be 20 years from now. The principal risk factors of the value of this bond now are long-term interest rates, credit spread/default risk and inflation.

It would be easy to simply line up bonds on the asset side to fund obligations with the first level similarities and to basically claim that everything has been matched up. Unfortunately, there are a number of problems with this approach:

• The discount rate used as the principal determinant to discount the obligations is an idealized version of reality; if an investor seeks to replicate the universe that it is based on, he or she will experience trading costs and credit deterioration that the discount rate does not. This will lead to widening of the funding gap even with perfect replication.
• Credit risk (credit spreads and default risk) are principal drivers of value for any bond portfolio, yet they have next to no impact on the underlying valuation of the liabilities. The liabilities are valued by estimating factors such as the expected range of lifespans of beneficiaries. The performance of credit risk can obviously have a huge impact on how the asset side of the equation performs.
• Inflation will increase the amount payable by pension schemes because pension payments are loosely based on salary level. Salaries will likely increase in response to inflation, which will have a substantial impact on the value of the pension liabilities.

A key consideration when using bonds is that there is a reliance on one risk factor– credit spreads – as the absolute return engine to compensate or provide insurance against the non-discount rate-related changes in the present value of the pension obligations, both now and in the future.

100% funding is not 100% likelihood
The work that goes into the measurement and calculation of a pension obligation is very detailed, extremely complex and, in the end, nothing more than a well-calculated estimate of the future based on the information available today. There are known uncertainties with regard to a plan’s ability to meet its pension obligations, despite what appears to be a current fully funded or overfunded status. In addition, the risk of future factors that cannot be currently known will have an impact on a plan’s ability to meet its future obligations. For example, the impact of science in medicine (e.g., recent research on the advancement in creating stem cells from blood) will dramatically affect the mortality rate and the ability of a pension to forecast its future obligations. Just note the recent debate around the adoption of RP-2014 and MP-2014.[3] As they stand it could increase the value of liabilities by up to 7%, transforming a fully funded plan to one with a 93% funded status literally overnight, with no corresponding increase in the assets.

You can’t get out of the game
Achieving 100% or more funded status is not the endgame. Although this is better than the alternative, the problem is the same; just as a plan has to beat the expected rate of return to achieve its fully funded status, it has to find a way to maintain that rate of return. A typical approach to achieving this goal is to “de-risk” the portfolio, which is essentially switching from presumably “riskier” assets to presumably “safer” assets to “lock in” those gains and secure the funded status. This would make sense if the obligation valuation were secure (and certain) and is expected to remain so across the life of the plan. It would also make sense if return on these “safer” assets (usually fixed income) were guaranteed to be essentially equivalent to whatever the assumed rate of return is going forward or if the required return on assets is made to be the same as the discount rate of the liabilities. Unfortunately, achieving a return close to 6-8% with “safe” assets in the current rate environment could mean taking large amounts of credit risk, thereby rendering those same assets a lot less “safe.”[4] So even at 100% funded, plan sponsors would still have to pay someone (usually an insurance company) a significant premium to take on the plan’s risk, to hedge against the possibility of the plan’s changing liability valuation, and finally to grant that insurer some possibility of profit for bearing this risk.

Conclusion
With historically low rates and the zero bound not too far away, it is both intuitive and practical to structure a pension plan portfolio such that the very long durations of pension obligations are used against themselves. This can be done by only hedging a decline in rates and allowing the funding gap to narrow naturally as rates rise. Adding that approach to a broadly diversified portfolio of low market exposure risks should lead to a situation in which the funding gap narrows on a compounded basis from the absolute return portfolio and on a structural basis from the duration asymmetry. This approach alone can remove the need to make very difficult and risky macro allocation decisions between differing asset classes at the top fund level. One need only make these decisions when they are truly obvious, necessary and likely to be fully rewarded.

Ronan Cosgrave, CFA, CQF is a managing director and sector specialist for the Convertible Bond Hedging strategy at PAAMCO. He also serves as the portfolio manager and main point of contact for certain institutional investor relationships.

Footnotes

1. Volatility in rates can minimize funding gaps. Using an example when rates rise and fall by 1%, employing a liability matching approach will result in the loss (when rates rise) to be equal to the gain (when rates fall). In the same example, when using an asymmetric hedge with a roll, as rates rise the investor would sell the old swaptions and buy new ones with higher strikes; the investor will lose money as rates increase but the loss will be less than if the investor had employed the liability matching approach. Additionally, because the swaptions have been re-struck at a higher strike price, as rates decrease, more money is made than was lost on the way up, thus structurally narrowing the gap despite rates ending up where they started.
2. Barclays Long Government Credit Index as of 30 April 2014.
3. Society of Actuaries, RP-2014 Mortality Tables and Mortality Improvement Scale MP-2014, February 2014. Retrieved from http://www.soa.org/Research/Experience-Study/Pension/research-2014-mort-tables.aspx.
4. If you define “safe” as being a reasonable likelihood of making your return target without the possibility of significant losses.