It is surprisingly easy to overlook the point that pension funds exist to provide the necessary resources to fund payments to members. Funding allows investment returns to play a role in providing those resources. In an ideal world, investment returns would fulfil three criteria: to ensure that the assets do not lose value over time relative to liabilities; to provide a buffer against unforeseen risks such as a rise in longevity; and to reduce the need for contributions.

Put this way, it is clear that investment returns outperforming either the average pension fund investment return or certain market indices is irrelevant to the question of whether payments can be made to a scheme’s members. Instead, the key requirement is that investment returns exceed the growth of a fund’s liabilities or its return on those liabilities.

In a regulatory environment, where the liabilities are discounted at a fixed rate, the return on the liabilities will (appear to) be predictable. It therefore follows that an LDI approach will add little value. However, there is a clear global trend among financial regulators toward market-realistic valuations of assets and liabilities, as well as the use of risk-based capital principles for the purpose of assessing solvency and financial requirements. This overlaps with a similar trend in financial reporting standards that will also impact some sponsors. With discount factors driven by market yields, the liability ‘return’ becomes far from predictable and a high degree of volatility becomes a factor.

This is the essence of liability driven investment - the recognition that investment strategy should be constructed relative to the liabilities of a fund, and that assessment of risk and return must take place on a holistic basis to combine assets and liabilities.

Consequently, there is strictly speaking no such a thing as an LDI product. Instead, LDI is really a process to determine strategy. In developing that strategy, a whole range of possible investment solutions and products can have a role to play. Some of these will already be in some, some will be familiar - albeit refined for use in an LDI context - and others will be totally new.

The liability proxy

First, consider the nature of the liability returns that the investment returns must outperform. The average defined-benefit pension fund promises to make payments that extend many years hence. In the same way that the value of a fixed-interest bond is sensitive to changes in long-term interest rates, so too is the present value of a fund’s liabilities very sensitive to such changes.

Given the significant reductions in European interest rates in the last decade, the present value of pension liabilities has generally grown quite aggressively - even after allowing for the impact of increasing longevity and contributions, which in some cases have been lower than is actually necessary for the long term.

Different regulatory environments impose different inflation indexation requirements. However, even where indexation is not a regulatory or legal requirement, indexing against inflation can become a further source of volatility among a fund’s liabilities.

In LDI, the interest rate and inflation sensitivity characteristics of the liabilities are addressed by developing a liability proxy - the combination of assets that would provide the lowest risk-investment strategy relative to liabilities. Given the nature of pension liabilities, the liability proxy is usually a combination of fixed- and inflation-linked government bonds. One noteworthy limitation is that without any effective investment that can be used as a longevity hedge, the liability proxy cannot at present encompass longevity risk. However, there have been some indications that some banks are looking at issuing an instrument with these characteristics, so it might also over time become possible to incorporate this risk.

Although the liability proxy is a very simple concept, it becomes a very powerful tool in LDI. Because - by definition - it specifies the lowest-risk investment strategy, it becomes a powerful comparator for the assessment of the relative risk and return characteristics of different investment strategies. In particular, the risk arising from asset–liability combinations can be quite different from the risks arising from assets alone.

Once risk tolerance (relative to liabilities) has been assessed, a risk budget can be fixed. Each strategy under consideration is reviewed to make sure that total risk is consistent with the risk budget, and also that there is no undue sensitivity to any single risk such as, for example, global equity markets.

LDI approaches

Approaches to LDI typically fit three broad categories.

Close matching: If a scheme is well funded, it is possible for it to adopt the lowest-risk investment strategy available to it. The scheme would, therefore, invest in fixed- and index-linked bonds designed to match the cashflows of corresponding liabilities. In practice, the precision of matching can vary between an annual cashflow match on the one hand, to broader match of average exposure on the other. If it is accepted that the liability projections are fairly accurate, and if the local bond market is liquid and well developed, then a more precise match is possible. However, in less favourable circumstances, increased transaction costs can outweigh the risk-reduction benefits of a closer match. A close-match strategy can either be passively managed or it can be actively managed within agreed risk limits - allowing the asset manager to take active positions, for example, on the shape of the yield curve or corporate bond exposures or even overseas bonds exposures.

Where the local bond market is either illiquid or poorly developed, the interest-rate swaps market might still allow a closer match (if desired) than physical bonds. For Finnish funds, membership of the euro offers a much wider range of government bonds and swaps than is available to other Nordic countries. However, the potential risk between local inflation and eurozone inflation must be considered.

In practice, having determined the scope of the securities to be considered, it is helpful to use an optimiser program to determine the best fit between the available assets and liabilities. So, especially where the maximum maturity is shorter than the liabilities, it is common to see the significant exposures to the maximum maturity to balance the ultra-longterm exposure.

Optimised conventional solutions: Unless a pension fund is very substantially over funded, a narrow match will only deliver the liability return and will not exceed it significantly. So many funds are willing to bear greater risk in order to increase their expected return. Using an optimiser, it is possible to estimate the efficient frontier of a range of investment strategies, determining for different levels of targeted return over the liability proxy the optimal investment mix that has the lowest expected risk relative to liabilities. Optimiser models can be quite sensitive to the quality of the input assumptions, and in practice there is usually a range of broadly efficient strategies clustered around the optimal strategy that will all be effective.

The basic process is similar to the traditional strategy-setting mechanism. But because of the focus on the position relative to liabilities, the outcome is likely differ as follows:

• the strategy will have a lower exposure to equities and a greater exposure to bonds;

• the strategy will be highly diversified into a broad range of alternative asset classes;

• long-maturity bonds will be favoured over shorter maturity bonds because they have a lower risk relative to liabilities, yet provide a higher expected return.

• a swap overlay permits management of the interest rate and inflation exposures and allows greater flexibility in the strategy for the underlying assets; and

• the strategy will not rely on competitor benchmarks and will place less reliance on index benchmarks, which will be as broad as possible.

The strategy will seek to make use of all available sources of outperformance by asset managers within their market. Clearly, if added value can be generated by active managers, there is less need to take market risk in order to meet the required returns. It follows that any market risk will be a smaller proportion of total risk. This is particularly advantageous because any overperformance or underperformance by an asset manager tends not to correlate very strongly with the market risk, thus providing diversification benefits. Techniques that have aroused interest in the marketplace are developing in the realm of portable alpha and permit the independent selection of market exposure and manager skill.

Figure 1 Illustrative optimised cashflow match using eurozone bonds

Holistic solutions

If, for example, the liability return is broadly equivalent to long-duration bond returns, investment objectives can be expressed as ‘bond returns plus a bit’. Holistic-solution approaches are where this type of objective is directly defined as an explicit benchmark for the asset manager. The assets are then managed in a process designed to deliver this benchmark.

Figure 2 Efficient frontier in an LDI context

Such a process could be operated in a wide range of different ways. Assuming that the liability proxy has been properly defined, the important factors to consider in assessing such an approach are:

• the process is truly aligned to generate the target (and is not merely a re-branded process that aims to beat an irrelevant market index); and

• the number and quality of sources from which performance exceeding the benchmark will be derived.

Issues to consider

There is an old military adage that generals will always attempt to perfect the fighting of their previous war until they are forced by events to recognise that the new war is different. The fall in global long term interest rates has caused liabilities to rise significantly so that rates are now at historically low levels. Hedging interest rate exposure now could result in missing out on the fall of liabilities that a rise in interest rates would bring. However even if the world has not changed in some way and rates do revert toward their long term average, there is no requirement that rates will not fall further in the meantime. As Maynard Keynes famously said, markets can remain irrational for longer than one can remain solvent.

The decisive issue is whether further falls in rates are affordable. If exposures seem to exceed what can be tolerated, it is possible to hedge part of the exposure now and to continue bearing the remainder of the risk until a more favourable time arrives. This decision should be made in the context of the holistic risk budget as of a number of risks taken in order to increase the expected return. If this issue is approached separately, there is a danger of retaining too high a degree of interest rate exposure, which would unbalancing the overall strategy.

Economic risk or regulatory risk?

Since Denmark pioneered the introduction of the traffic light system in 2002, regulators elsewhere have adopted the philosophy of using market values and shock-based risk analysis. We expect regulation to continue to evolve in this direction, as has been the case in the Netherlands, where debate continues in respect of the choice of discount rate (government bond yields or swap curve yields) and the formulation of shock tests for asset liability combination. However, as regulations currently stand, different countries are either more or less aligned with a pure holistic risk analysis.

One further factor is the differences between the extent to which member expectations (for example, in respect of inflation indexation) are legally enforceable or can be disregarded. If expectations are legally enforceable, there is only one set of liability cashflow predictions; however if not, there will be a totally different view of the liabilities that turns on whether the expectations are built in or not.

These factors introduce a tension between whether the solution should be optimised for the real economic risks or optimised for specific regulatory risk assessments. It will rarely be possible in practice to meet both objectives effectively, and steering a middle way can result in the worst of both worlds. Whereas it might be it possible to take advantage of regulations where the regulatory risk assessment is not aligned with real economic risks, this will tend to lead to a less than optimal solution in real economic risk terms.

So on balance, focusing primarily on economic risks is not to be recommended - with the possible exception of those cases where the regulations appear to be unduly penal and where optimisation at the behest of the regulations can take place without detriment to the economic risk strategy. Regulations will clearly continue to develop over time, and so current rules are no strong justification for failing to address real economic risks.

Start with risk tolerance

The risk tolerance of a fund should determine a risk budget to limit choice to those strategies that offer a derived maximum relative return. However, even during LDI reviews, the following variations can be observed:

• some clients merely set their risk levels relative to their current risk and maximise their returns;

• other clients determine their current expected outperformance relative to liabilities and then decide whether to bear the risk that accompanies an optimised version of that return; and

• and yet others have as their starting point the view that the outperformance which they believe is necessary and decide to bear the risk that accompanies an optimised version of that return

In each of these cases, the level of risk is a result of the process not an input. This undermines the logic behind LDI; although the client might be better informed about the level of their risk exposure - the solution might also be better optimised than it was before (better ratio of risk to return) - this is a more superficial answer and can still lead to bearing risk that cannot really be afforded.

Proper assessment of risk tolerance requires substantial discussion between the fund and the sponsor, as well as an understanding of the shape of utility curves, which might different, and in particular any inherent limits on capacity for loss. Equally, It is important to consider the appropriate timescale against which to assess risk and again this may be different between trustees and sponsor. One side effect of the move toward market-value accounting is that sponsors are less likely to tolerate short-term risk.

Printer friendly version

Email this article to a friend