Iain Morse describes the benefits and pitfalls involved in implementing a pooled liability fund strategy
Will we all live to be 100? Increasing longevity has the concomitant effect of hugely amplifying interest and inflation rate risks for defined benefit pension schemes. Typical bond portfolios have no more than a five- year duration, typical pensioners somewhat more. Their deck chairs and ice creams must be paid for.
"After an asset liability study, trustees and pension boards are often unpleasantly surprised by the size of the interest and inflation rate risks they are running in their liabilities," warns Andrew Firth, senior investment consultant at Aon Consulting in the UK.
In some cases, these risks are larger than the risk in a scheme's equity portfolio. "More and more pension boards are now seeking to hedge out some or all of this risk. I would say 60% to 70% of our clients use some hedging," notes Gerard Roelofs, head of investment consulting at Watson Wyatt in the Netherlands.
There are three ways of hedging this risk. Larger schemes can go directly to an investment bank and arrange a forward programme of interest rate or inflation rate swaps. As an alternative, the trustees or pension board can ask an asset manager to arrange a programme on their behalf with a bank. The last, and increasingly popular, option is to invest into one or more of a wide range of so-called bucket funds, some leveraged, some not, which offer a pre-packaged solution.
Making the best choice between these is far from straightforward. A pension scheme dealing direct with a bank needs to effect due diligence and reach agreement on issues such as the main ISDA agreement and an annex dealing with collateral management. Bucket funds may cost more but governance and implementation is dealt with by the provider. Risk is thus offset to the provider by trustees or board, minimising time on implementation.
The permutations of the choice between bespoke and off-the-peg solutions are complex. Some schemes use more than one solution. The decision also needs to placed in context. In the last 10 years there has been a rapid progression in the paradigms used to capture optimal asset allocations within pension scheme portfolios. In the late 1990s balanced management gave way to core-satellite. Now that has been superseded by a distinction between return-seeking and liability-matching parts of a portfolio. There have been many digressions along the way; passive versus active, enhanced index, active quant, alternative and ‘new' asset classes.
Liability driven investing (LDI) is the latest. Increasing bond allocations and increasing demand for inflation-linked bonds were the initial solution to liability matching. This change is driven in part by a need to reduce or eliminate uncompensated portfolio risk. The difference between now and the 1990s is that portfolio risk is increasingly measured by a scheme-specific liability benchmark.
Over recent years, the price and duration of sovereign and investment grade bonds have made swaps relatively attractive for matching and since 2000 innately conservative pension trustees and boards have used them more widely.
ery large schemes were early adopters; mass-market bucket funds have only been available for three or four years but the growth in their use is rapid. Market leader Barclays Global Investors already has more than £7bn (€9.7bn) in its sterling and euro-denominated funds, State Street Global Advisors (SSGA) an estimated £3-£4bn, PIMCO probably a little less. But quite a few other providers, such as Robeco, have operated successfully in the Dutch market for several years.
There are no reliable estimates of the aggregate amounts in bucket funds or the extent of other interest rate and inflation rate hedging but the totals must run into tens of billions.
This sounds straightforward but setting the exact definition of liability benchmarks varies from country to country. In the UK, pensions in payment must increase in line with a measure of real price inflation, as must accruing deferred benefits. In the Netherlands, increases are based on an annual decision by each scheme's board based on their current funding position.
Elsewhere in Europe the situation is equally varied; some countries periodically set minimum fixed accrual rates expressed as nominal pension increases. Hedging in Holland is mainly against nominal rather than real rates of inflation. The difference between hedging nominal and real is significant; nominal is fixed in advanced, real will vary by a formula such as retail price inflation or LIBOR plus. The costs of the swaps also varies; hedging nominal is less expensive than real. It can also be difficult to find appropriate inflation hedges; Dutch pension schemes use Dutch RPI, which varies from euro-zone RPI, and so on.
There is also little consensus on the optimal hedge for liabilities. If two thirds of Dutch schemes hedge one or another interest or inflation risk, only a minority try to hedge 100% of this liability. "Most start at 50% then some go up to 75% or 80%," says Roelofs.
The marginal benefit of hedging the first 50% seems to be greater than the marginal benefit of hedging the remainder of this risk. "The reasons for this relate to the matching characteristics of assets in the growth component of the portfolio," Roelofs continues.
The initial hedge dilutes the overall risk of the portfolio and proportionately increases the matching characteristics of equities and other growth investments over longer investment horizons. "The extent of any hedge ratio must be judged on a scheme-specific basis, according to issues such as their overall asset allocation," he adds.
However, it is still possible to form a generic picture of optimal hedge ratios for a hypothetical scheme. Suppose a scheme has a nominal liability duration of 15 years, an asset allocation of 40% equities and 60% bonds and a nominal coverage ratio of 125%. "It is then fairly easy to calculate the required duration for the fixed income portfolio if the scheme does not want to increase the allocation to fixed income," says Alexander van Ittersum, senior product manager at Robeco in Rotterdam.
The required duration for the assets equals the duration of scheme liabilities divided by the coverage ratio multiplied by the fixed income portfolio allocation. Fifteen would therefore be divided by 125% times 60%, which results in a duration of 20 years. "This is the required duration for the assets if the pension scheme wants to fully hedge the interest rate risk," he adds.
If the pension fund wants to fully hedge the interest rate risk it can invest all of its fixed income portfolio in bucket funds, with a duration of 20 years unleveraged, or 55% of this in bond portfolio with a five-year duration and 45% in a leveraged fund with a duration of 40 years. "There is flexibility in these solutions," adds Van Ittersum.
"The attraction of using swaps is that they can cost less," argues Joseph Moody, head of liability driven investing at SSGA.
The obvious alternative is to increase fixed income allocations. But most under-funded schemes, particularly in the UK, cannot afford this response because of the concomitant fall in expected returns after an increase in bond allocations has been made - although it needs to be emphasised that this is
never a ‘one-sized-fits-all' solution, Moody points out.
The strength of a sponsor's covenant, of the same sponsor's balance sheet, trustees attitude to risk and more all make this a scheme-specific decision.
"Suppose a scheme has nominal mismatch between assets and liabilities of 15 years, and long interest rates, and therefore also swap rates fall by 0.5%," he adds. As swaps represent the credit risk of AA corporate bonds they also yield a higher return over government bonds and are used to value the liabilities of many schemes. Hence a fall in the discount/swap rate means a commensurate rise in pension liabilities. The impact on the relative value of the assets versus liabilities would be about -7.5%, changing a funded scheme into one in deficit.
This kind of consideration helps explain the growing popularity of swap-based solutions to hedging interest and inflation risk. But what about the relative costs of a segregated programme of swaps versus the use of bucket funds?
The charges for each are clear in outline but opaque in detail. This is because both the investment banks and the fund providers vary their charges on a client-specific basis. Expect an annual management fee ranging from 5-6 basis points up to 20 basis points on bucket funds. These may or may not also include an initial joining fee, also typically expressed as a basis point charge on assets under-management. "We certainly would negotiate on behalf of our clients not to pay this kind of initial charge," warns Roelofs.
The cost of rolling swaps on a segregated basis should be much less, a few basis points, but this ignores the additional very real costs of running collateral, ensuring that cash flows occurring from the swap programme are fully accounted for and much else. In a conventional asset class like equities or bonds, manager charges would be set against the likelihood of a manager meeting or beating a performance benchmark, but this measure is not applicable to these funds.
There are a large number of bucket funds now available and the number of their providers easily exceeds the half dozen or so large banks usually used for segregated programmes. For instance, BGI has more than 30 separate funds, and State Street not many fewer, designed to mature in successive years so hedging can be built out by a scheme using the funds on a mix and match basis.
Providers such as Robeco have found an ingenious solution to their larger rivals; Robeco offers only four funds but combines these with a range of bond funds that offer varying duration.
"The choice can come down to issues such as quality of reporting," adds Roelofs, "and sometimes familiarity, such as whether you are investing into other funds with the same provider."