There has been a growing emphasis on pension fund liability management and creating a portfolio of assets to match these liabilities.
The popular press has published several recommendations to protect pension fund solvency through the immunisation of liabilities.
Recommendations have ranged from the approach of the Boots Pension fund, which converted the entire portfolio of assets into fixed income assets (since revised to include a small allocation to non-fixed income assets), to a complete abandoning the strategic asset allocation policy in favor of a totally tactical policy, to duration extension combined with ‘portable alpha’.
However, the principal challenge to developing an optimal policy for managing the assets-to-liabilities ratio (called the funded ratio) has probably more to do with (a lack of) understanding of the liabilities than in developing innovative investment policies. Through a series of short notes, we explore how innovative investment policies can be used to cause the funded ratio to grow over time, but first we develop a simple process by which liabilities can be understood and monitored on an intra-year basis.
This renewed interest in asset-liability matching is mainly caused by the rapid decline in interest rates, which led to an increase in most pension fund liabilities on a mark-to-market basis. Unfortunately for most pension funds, this decline in rates coincided with a decline in many stock markets. The fall in the value of assets at a time when the value of liabilities was increasing led to a dramatic decline in the ratio of assets to liabilities.
This ratio is often what regulators, CFOs and members are looking at to judge whether a plan is “safe”.
Further, new accounting standards imply that pension fund losses can affect the corporate pension fund sponsor. For these three reasons, many pension plans worldwide are adjusting their investment and hedging strategies to reduce the impact of a declining (and volatile) funded ratio.
Pension funds often find it difficult to express the desired approach and implementation of liability-driven investments to their asset managers.
The main problem pension managers encounter is in setting an appropriate investable benchmark which reflects the liabilities.
Here we describe a simple, stable and accurate benchmark to mimic liabilities based on daily available Lehman swap indices. The only input needed is the projected annual liability cash flows. The website www.nftk.nl provides a free trial of software to help pension funds make effective decisions. We use the work done at the €17bn Bedrijfstakpensioenfonds Metalektro in the Netherlands, also known as the PME Pension Fund (www.pmeonline.info), to clarify our views.
Chart 1 presents a typical actuarial example of projected pension benefit cash flows. We restrict ourselves to the nominal liabilities.
Now the board appointed executive faces the question: How to manage assets against this set of cash flows?
To effectively manage a pension fund, one needs to measure and monitor these nominal liabilities. A related question is: how to measure and monitor liabilities relevant to investment decisions that need to be made? Previously, an annual approach was sufficient, but in today’s world a more frequent valuation is needed, especially if one wants to manage the funding ratio effectively on an intra-year basis. In turn, that leads to the challenge of translating the above cash flows to quoted market instruments.
The standard approach to develop as a benchmark is relatively simple, but it is harder to implement in practice.
It forces managers to accept either a high tracking error, or a very costly implementation.
Effectively, each future cash flow is nothing but a zero coupon bond with a given maturity. When viewed this way, the manager needs to invest in such bonds in the appropriate notional amount. For example, if the estimated cash flow due in 2016 is €750,000 , then the manager needs to find a zero coupon bond with that maturity and invest the relevant amount to earn a payout equal to €750,000. The relevant amount to invest today to achieve such a payout in December 2016 would be determined by the zero coupon bond rate for that maturity.
Using the cash flows demonstrated in chart 1, it can be concluded that the present value of liabilities is €15bn with a duration of 15-14 years, as shown in the table. However, the difficulty of this approach is that such pure instruments do not exist, and, hence either a theoretical zero coupon portfolio can be constructed but not implemented, or an alternative portfolio can be constructed from zero coupon as well as coupon bearing bonds (or by boot-strapping the zero coupon curve implied by these bonds).
It is not that hard to construct such a portfolio for any set of liability cash flows from the universe of Eurozone government bonds, though inter/extra polation techniques have to be used for the 30+ years cash flows. The theoretical bond portfolio can be marked to market daily and, hence, give clients an effective estimate of their liability performance intra-year. The bigger problem is that with many cash flows, over multiple years and of moderate amounts, the liability benchmark involves small allocations to multiple securities, making the benchmark unwieldy and difficult for the average board member to grasp, so an alternative approach using swaps (see below) is more appropriate.
Managing a tailor-made portfolio of 50 zeros does not seem to exceed current computing power. The real issue is that the zero coupon bond approach does not lead to clean and transparent pricing and would also be costly to implement. Furthermore, the obtained benchmark relying on the valuation of (zero coupon) bonds, may contain a credit spread element (as there is no unique Eurozone government yield curve, because different issuers have (slightly) different credit qualities) and is hence less objective than standard benchmarks. For that reason, PME adopted a different approach using the most liquid fixed income markets instruments: swaps.
The swap curve is the pricing benchmark in the euro market and therefore the best starting point for index construction. The plain vanilla (coupon bearing) interest rate swaps market offers excellent benchmarks.
For instance, each of the Lehman Bellwether indices has a long track record. If the above cash flows could be replicated by using a set of these indices, an objective and investable benchmark could be achieved.
Under the swap method, an optimisation technique is used to determine optimal weights to a portfolio of Lehman indices such that this swap portfolio will mimic, to a large degree, the performance of liabilities.
In such a situation, a chosen set of indices is used to find an optimal mix, and typically the swaps selected are of the standard plain vanilla coupon bearing variety and maturity (ie, one-year, two-year, five-year, 10-year, and 20-year). The procedure to determine the weight of each Lehman index in the liability benchmark is to use historical yield curve data from 1999-2005 and find the portfolio that gives sufficiently low tracking error relative to the valuation of liabilities. In the case of pension cash flows modelled in the chart, optimal weights to the various indices are as follows:
First, this is a very simple portfolio of some key liquid instruments and, hence, lends itself to acceptance by a board. This portfolio of Lehman indices has an annualised tracking error of 0.30% relative to the liabilities over the historical simulation period. An alternative check is to compare the durations of both portfolios – and as shown in table 1, the difference is minimal.
One can argue that it is better to optimise the weights over some future simulations, using a Monte Carlo technique, but such a proposal only changes the method, not the approach to estimating an appropriate liability benchmark. Chart 2 shows how liabilities evolved from 1999-2005 with changes in interest rates, and chart 3 demonstrates the close tracking of the portfolio of swaps over this period.
There are other caveats; sometimes such an optimisation can provide negative weights to certain swap maturities (eg, the 12-month or one-year recommended allocation) and sophisticated clients can see this as either a short position or a forward starting swap. This is a very simple portfolio to monitor and track and mark-to-market as frequently as other assets. Effectively, one can now track, on an intra-year basis, the annualised growth in the surplus as the difference between the annualised return on assets minus the annualised return on liabilities.
An additional benefit of using the swap-based approach is that very little capital is required to hedge the investable liability index through swaps or other derivative instruments.
Many pension funds mandate setting aside capital for hedging liabilities (through fixed-income investments) or return generation (through equity or alternative investments). Then, because they feel there is a leverage constraint (ie, the weights allocated to these instruments have to add up to 100%), they act as if there is tension between these objectives. Of course, this is not the case. The real concern should be overall funding risk, and the ability to effectively hedge liabilities with derivatives is an attractive proposition that frees up capital for return generation.
This note set out to demonstrate how innovative pension plans can convert actuarial projections of pension benefits into an appropriate investable benchmark, allowing plan sponsors a better measure of the performance of their liabilities (and assets) on an intra-year basis.
In future notes we extend our Metalektro case to explain how one can control a €17bn portfolio in a very cost efficient way by using advanced web-enabled techniques. In the past three and a half years, PME has used this approach to ensure one of the highest returns of all Dutch pension funds, which include over 800 pension funds, in a very competitive market. More important, using this approach of tracking liabilities effectively has allowed PME to control and improve its overall solvency.
Jan Willem Stuijvenberg is founder of Stuijvenberg Financial Services and Arun Muralidhar is chairman of Mcube Investment Technologies