As anyone who has been in the pensions industry for a while can vouch for, investment strategies are as much ruled by fashion as by theory. This reflects the complexities of an environment with many different stakeholders and a slow but inexorable tightening of legislation turning promises by pension sponsors into legal liabilities, without a detailed map of where such a road leads to.
The cult of the equity introduced by into UK institutional pension funds during the 1950s replaced the previous orthodoxy of using bonds as the main investment of pension funds. Having seen immense growth in equity holdings by pension funds during the next four decades, the arguments are now swinging back in favour of more bonds, as pension funds seek benchmarks relevant to their future liabilities rather than arbitrary market average or peer group comparisons.
Turning a pension promise into a legal liability leads to the argument that pension liabilities are no different form other corporate liabilities and should be brought onto the balance sheet as indicated by the accounting standard FRS 17. Matching liabilities with assets that behave in a similar manner has led to some arguing that there should be a complete replacement of equities by a combination of conventional and index-linked bonds. However, there are theoretical and practical issues that will make such a move unlikely in the short term, whilst even in the longer term, there are strong reasons for ensuring a diversified portfolio, albeit with debt instruments having a major and probably better planned role. The key difference in the future should be that all stakeholders in a pension plan are aware of the role that the debt investments are delivering, and the reasons why they are combined with investments in a range of other assets in a controlled and risk budgeted manner.
One of the problems that exist in deciding how to match liabilities is in deciding what the minimum risk portfolio would actually be. The UK’s actuarial profession had a working party that produced the idea of a Liability Benchmark Portfolio (LBP) that acts as a proxy for accrued liabilities and suggested that this in most cases could be approximated by a portfolio of conventional and index-linked bonds, although even the working party itself was not able to agree in totality on this with one dissenting member submitting a separate proposal. Whilst a benchmark of liability cashflows can be constructed, they can involve cashflows going out to 80 years in the most extreme cases (a 20 year old who lives to a 100). There are few bonds over 30 years and whilst it is possible to trade nominal swaps out to 70 years, liquidity reduces after the 30 year mark and for inflation linked swaps, 40 years is the absolute maximum. Constructing a benchmark “minimum risk” asset portfolio that does not adequately match cashflows beyond 30 years needs to be handled with great care if it is not to confuse trustees even more by giving a misplaced sense of objectivity and security. It also raises the issue as to whether it is possible at all to construct a matched portfolio of fixed interest and index-linked bonds that is the theoretical minimum risk portfolio.
Debt market instruments
At the core of the debt markets are government bonds, which define the risk free rates of return. Issuance of government debt during the 1990s slowed down after the period of rapid growth of borrowings during the 1980s, whilst there has been large growth in issuance of other forms of debt, notably through securitisations, privatisation and also increased issuance of debt directly by corporates as an alternative to bank borrowings. This has also been accompanied by the explosion in growth of the debt derivative markets, which can enable tailoring of cashflows to be undertaken in a cost effective manner, as well as widening the universe of investment opportunities.
Whilst ideal in terms of matching liabilities with no credit risk components, the total size of the government debt market is limited. The UK gilt market for example, at the end of 2002 was around £300bn (E450bn) compared to the total UK pension fund market estimated at around £600bn. With around £130bn in sterling bonds, if the pension market moved towards an average 50% bond allocation, it would have created demand for another £170bn. Despite the substantial increase in gilt issuance in 2003, there is clearly, a large mismatch between potential demand for long dated bonds in particular, and the actual supply.
The use of derivatives can aid in both extending the universe of potential investments and also in tailoring specific cashflows if required. Futures based on government bonds, can and often are used to adjust durations of portfolios and for cashflow management. Cashflows can be tailored using interest rate swaps where the floating and fixed payments are exchanged on a notional amount whilst foreign currency bonds can be purchased and hedged back into the domestic currency using currency swaps where the principal amounts in two different currencies are exchanged and interest paid on the swapped amount. Whilst it is possible to construct very complex derivative-based structures to hedge guarantees embedded in liability payments, there is usually a price to be paid in terms of price and illiquidity. However, a striking example where such structures would have been useful is the demise of the Equitable Life Assurance which, like its competitors, had not hedged out its exposure to guaranteed annuities sold during the periods of high inflation in the 1970s. In contrast, CGU (now part of AVIVA) managed to hedge its remaining exposures through a series of forward starting options on swaps during the late 1990s, although the size of the transaction probably dried up the total capacity in the market for such transactions for some time, closing the market to its competitors.
Credit exposures and derivatives
Extra yield beyond what is available from government bonds for liability matching can be obtained at low levels of risk by investing in investment grade corporate bonds and asset backed securities. Going further down the risk spectrum, to higher yielding debt with varying risks of default on interest and/or principal repayments leads to more volatile returns more akin to equity investment. Whilst many of these debt products may be attractive investments in their own right, they have less relevance to liability matching strategies. Emerging market debt for example, has been one of the best performing asset classes during the last decade or more, and a number of major European schemes have made substantial investments during the last couple of years. Sophisticated managers can obtain high returns through seeking illiquid issues and controlling risk through utilising credit default swaps which are akin to credit insurance, with one counterparty receiving a regular premium in exchange for making a specified payment in the event of a negative credit event such as a default. However, the resultant cashflows will certainly not be suitable for accurately matching defined liability cashflows. They may however, be suitable assets for giving inflation plus returns over long time periods. Managers such as Muzinich that specialise in active approaches to investing in high yielding corporate bonds would argue that their approach of investing in high coupon bonds with low durations is designed to beat inflation through a business cycle.
The fact that there is a strong inflation linked component to pensions led the UK government to introduce index-linked gilts in the 1980s and they have been followed by a number of other countries with outstanding volume of government debt in Europe now around $60bn. Although clearly designed to fulfil pension funds inflation liabilities, the limited amount of index-linked gilts in existence makes them expensive and illiquid, exacerbated by the lack of significant corporate issuance. There has been a certain amount of activity utilising inflation swaps where fixed and index-linked payments are exchanged on a notional amount. Whilst theoretically, it should be possible to find corporate entities willing to pay out inflation-linked cashflows as counterparties to pension funds wishing to receive them, there has not been much evidence of this actually happening so far and the indications are that inflation linked swaps have to be hedged using index-linked gilts, making them expensive and illiquid. The increase in demand for index-linked cashflows has led to the index-linked gilt market performing well over the last year, leading to less attractive return potential for the future.
One problem with matching attempting to match liabilities with index-linked bonds is that the cashflows can be complex and there may be a capping of inflation linking in the pensions to say 5%. It may be cheaper to calculate the worst-case scenario in terms of the inflation linking and just buy more conventional bonds or fixed interest swaps to cover that eventuality.
Cashflows and portfolio management
To understand the nature of the problem, an example of a typical cashflow projection is shown below.
The red line shows the actual cashflows whilst the blue line shows their value discounted at the appropriate rate, to give the net present value of each cashflow.
Matching the cash flows exactly to produce a minimum risk portfolio would require matching the fixed cashflows exactly going out as far as is required, together with matching any inflation linked elements with index-linked bonds and swaps. This is impossible to do exactly without using swaps as there are no bonds with long enough maturities. If you are taking mismatch and credit risks for the cashflows beyond that period, how perfectly should one hedge cashflows where there are instruments available but at increased cost?
What is immediately obvious from the chart is that whilst cashflows may extend almost to 80 years, the present values after 35 years or so are very small. The arrow shows the mid-point year where the sum of the present values before and after are the same. It represents the duration of the cashflows and a simple way of taking account of liabilities is to ensure that the asset portfolio has the same duration as the liability cashflow, although the details will be very different. This procedure known as duration matching is a popular way to capture the most important characteristics of a liability cashflow whilst leaving great flexibility on the choice of bonds in the asset portfolio. There are of course, margins of error in the assumptions used to calculate the cashflows so that target durations may need to be reassessed on a regular basis.
Duration levels off after 15 years or so with change in maturity and as it is not possible to match very long cashflows with appropriate bonds, managers sometimes attempt to get round this by buying the longest duration bonds available with the highest ‘convexity’, ie, the highest change in duration for given change in bond yields. This provides some measure of aligning changes in the value of the longest dated liabilities due to interest rate changes with the changes in the value of the shorter dated fixed income assets. Using a combination matching near term cashflows exactly and matching just duration for longer term liabilities is a pragmatic solution that is sometimes also advocated.
Liabilities and asset allocation
Actuaries report on liabilities only once every three years, but asset prices can be seen every day so trustees have no conception of how assets and liabilities move together daily. The traditional approach has been for investment consultants to recommend a strategic bond/equity mix and then allocate mandates and set benchmarks to fund managers without any further acknowledgement of the liability profile. Typical bond benchmarks have durations far lower than those of liabilities, and it is the questioning of the high exposure to equities in particular of more than 50% that is leading to interest in liability driven benchmarks.
Only sophisticated companies with large treasury departments can look at liabilities in detail on a regular basis. Some FTSE 100 companies have started to examine liability cashflows and mark-to market valuations and are looking at taking these accurately into account in managing pension funds. Fund management groups that are part of larger financial services conglomerates are in some cases, able to use a combination of investment banking and fund management specialists to offer a consultancy-led investment approach that can appeal to more sophisticated clients. As Dominic Delaforce from Deutsche Asset Management finds, it can be helpful to have the back up of investment banking colleagues to be able to discuss more complex analyses of liability cashflows. The universe of possibilities can be expanded quite dramatically through the appropriate use of derivatives for example. Corporate bonds are a good case in point, with UK credits being a limited pool of assets, they trade expensively. At times, it is possible to buy an identical credit, for example, Vodafone, in US dollars, with a 40bp pick-up in yield and use the swap market to create essentially better yielding synthetic sterling assets.
In a long term industry, evolutionary changes are generally more preferable to revolutionary and this applies strongly in an industry where ultimate decision making lies with trustees who are unlikely to be investment experts and therefore reliant on outside advice and the comfort of market practice. The concept of the Liability Benchmark Portfolio in this respect, may be a useful tool, although it has to be recognised that a portfolio of conventional and index-linked gilts will not eliminate risk, particularly given the extended maturity of liabilities and therefore even for the most risk averse schemes with liabilities extending beyond 30 years or so may recognise that equities and other asset classes may have a part to play.
How does one reconcile the different strands into a framework that is solid yet flexible? One approach may be ‘risk budgeting’. As Delaforce would describe it, this could follow the steps of first deciding the liability-based benchmark cashflow; then deciding on an estimate of a “minimum risk market exposure” likely to be a combination of index linked and conventional bonds taking into account the yield curve and reinvestment risk after the 30-year cut-off. Given a benchmark portfolio, is this likely to generate sufficient return i.e. be at a yield that is high enough to be acceptable? If not, then diversification would be required to get more return after determining a risk budget available to be invested in equities, property and alternatives. The questions to be answered here would be; what return needs to be achieved; what mixture of managers and market risk do we want to take and finally what assets do we need?
Another way of looking at a liability driven approach is based around the chart showing the cumulative sum f the net present value of the liability profiles shown below.
Here it can be seen that 80% of the present value of the cashflows in question are paid out after around 32 years and so having a portfolio of bonds and index linked going out to around this period would leave just 20% of the liabilities that need to be matched by other long term assets. Even the most risk averse investor may feel comfortable with the idea that equities over periods of 30 years may be able to match bond returns even without assuming an equity risk premium of 2% or more. Those funds that are less risk averse can shorten the time period that they require cashflows to be matched, leaving more flexibility in the choice of assets beyond this. This can be looked at in terms of a ‘core/satellite model’ – the core here should be index-linked and fixed interest and the rest could be in assets that are ‘highly sweated’ – eg, a range of equities, property and alternatives. This can lead on to even more sophisticated strategies whereby the ‘satellite’ portfolios consist of portable alpha strategies, ie, a combination of specialist managers utilising long only strategies in individual markets with the market exposures removed using derivatives or else market neutral hedge funds.
However, much analysis one can undertake, human nature may ultimately be the driving force. As Delaforce points out, once a scheme becomes over funded, the incentive for the employer to take equity risk dramatically decreases but if the equity market rally continues and drives schemes to an over funded position, employers may become less concerned about risks and continue with high exposures to equities. So by the time they can afford to switch more into bonds, they won’t want to.