One of the key duties for pension scheme trustees is to invest the contributions in order to have sufficient funds to meet the benefits as they fall due. Accordingly, this is an ‘asset liability’ problem (not just a question of maximising returns) and so in setting investment strategy, trustees first need to consider the scheme’s liabilities.
In the case of active members these liabilities are mostly ‘real’ – being related to wage and price inflation. Intuitively, real liabilities require ‘real’ assets (equities, property, index-linked gilts) whereas fixed liabilities (eg, fixed pensions) require ‘cash’ investments, such as bonds. Other liabilities (eg, inflation-linked pensions but with a cap of 5%) have elements of both. The problem is that most real assets are not only low-yielding but also their prices are volatile.
Pension scheme trustees must nevertheless have both a long and a short-term perspective. The reporting of the long-term funding position involves a degree of smoothing to ride out the bad times with the good. But pension schemes can be wound up at short notice, which dramatically changes the liabilities from ‘real’ to ‘fixed’ – there is no longer any linkage to wage inflation. The short-term position is also where the statutory obligations lie under the minimum funding requirement (MFR).
An AL study is used to determine the extent to which short-term constraints impact on investment policy and in particular the balance between equities and bonds. In general, the more a scheme is mindful of its pensioner liabilities, the more of its assets need to be in bonds. Maturity impacts on investment strategy in a number of ways:
q Pensions need to be met from investment income and contributions to minimise the risk of selling assets at unfavourable prices.
q Deficits will have an increasing impact on the employer’s contribution rate (as a percentage of pensionable pay).
q The date of eventual wind-up is closer so the trustees need to have greater regard to the winding-up liabilities.
AL studies have historically been shrouded in actuarial mystique but they are really just a risk assessment. For well-funded schemes the short-term (or MFR) risk is further away and so there is less of a short-term constraint on the long-term desire to maximise returns – surplus generates investment freedom. For less well-funded schemes, how we express this short-term constraint (or ‘risk measure’) will be a matter for discussion. For example, it could be that we are concerned that:
q the MFR solvency level should not fall below 90%; or
q the employer’s contribution rate should not exceed the regular cost by more than 5% of pensionable salaries.
We also need to agree the constraint (or ‘risk criterion’). That is, there must be no more than a 10% chance, say, of our risk measure failing. The definitions of risk and return may differ, depending on whether one is considering the trustees, the shareholders or the company directors.
A key advantage of the AL tool is that it allows trustees to see the potential gains, as well as the risks, in adopting a mismatched position. This means that trustees are better able to take informed decisions on investment strategy.
Although the AL process can be useful, trustees should not accept blindly the output of such models – there are health warnings! These include:
q Proving that one equals one! If one asset class is assumed to return more than the ‘valuation rate of interest’ this will generate surplus in the long term, regardless of what happens in the short term. This can lead to anomalous conclusions.
q Investment models tend to work better in the long term. But if we are investigating the short-term position, our model should reflect a reasonable set of short-term scenarios (the long term is a succession of short terms!).
q Avoid spurious accuracy. In this part of the investment review we are only concerned with the broad asset allocation for cash, bonds, index-linked bonds and ‘general equities’ to the nearest 10% or perhaps 5%.
q Do the projections relate to the real world? Would we come to a different conclusion if the AL study was made at October 1999 instead of 1998?
The AL process is really quite simple in structure. First, we consider how the scheme’s liabilities are affected by certain parameters such as inflation, salary growth, investment returns, etc. This is usually done by projecting the cashflows out of the pension scheme (pensions, death benefits, transfers etc). Against these are placed the positive cashflows in respect of contribution income, etc.
Next we look at the interaction of capital markets and how they influence those same parameters of inflation, salary growth, etc. These relationships can be established by analysing historic data (it is said that about 75% of history has a habit of repeating itself) and a number of clever economic models have been developed for this purpose. These generally fall into two types – ‘random walk’ and ‘autoregressive’ (or ‘mean reverting’ – an up is more likely to be followed by a down than another up!).
Once our cashflow projection is complete we test a large number (5,000 or more) of random scenarios. Each scenario generates projected values for our parameters and returns for each asset class. From these one calculates the projected funding levels of the scheme and how they are influenced by each asset class. The whole process is as illustrated in Figure 1.
By putting all this together, so that we are analysing assets and liabilities together, we can see how different mixes of assets suit the liabilities. That is why we call the process an “asset-liability study”. We are not interested in how one asset class behaves relative to another per se, but how they interact in the context of the liabilities.
It isn’t all investment models and interaction of capital markets! Some AL studies can be at a very simple level. For example, consider a scheme where the trustees are happy with their overall strategy but are concerned that the assets will produce insufficient income to meet the expected benefit outgo. This needs little more than a broad-brush cashflow projection.
The more conventional AL study looks at the trade-off between maximising long-term returns and minimising short-term risk. There are many ways to present the results. In the simplest case, suppose we have just two asset classes: equities and bonds. We can plot on a graph the behaviour of the different portfolios (Figure 2).
Our ‘reward’ here is a high median (average) long-term funding rate. Our ‘risk’ is a low MFR solvency level. Extra caution is warranted so we consider a ‘worse-than-average’ result (the 33rd percentile). Eleven portfolios have been considered, ranging from 100% in bonds to 100% equities, in 10% steps.
As the equity content increases, so the long-term ‘reward’ increases and the short-term ‘risk’ decreases. This is ideal! However, in due course, the balance changes and whilst we continue to get better long-term returns, the risk of an MFR problem increases. The trustees might conclude here that an equity content of 60–80% draws a reasonable balance between the long-term and short-term.
Whichever way the results are presented, the conclusions of the AL study should be a recommendation for asset allocation; a strategy to accompany it; and a sensible benchmark for the scheme to use as its financial target. May all trustees find the ideal investment strategy for their scheme!
Mark da Silva is a partner at actuarial consultants Barnett Waddingham in London