When it comes to asset allocation, the focus has shifted from beating benchmarks to the requirement that liabilities are met, writes Richard Skelt.
Asset allocation has never been more important than in this post-financial crisis period. Determining a sensible strategic asset allocation is the key determinant of long-term performance, but it can be daunting.
A retail investor selecting an appropriate portfolio for retirement savings or an institution investing on behalf of others is faced today with a bewildering array of choice. The options have expanded beyond recognition since a portfolio of government bonds was the norm.
The fundamental challenge when setting allocation policy is the lack of certainty about returns across the investment landscape. If we consider a simple stocks, bonds and cash portfolio, what can we say about likely returns for the next five years? Or the next 25?
A key point is that our assumptions must reflect reasonable beliefs about the future rather than simply mirroring the experience of the past. It is all too easy to take the last fifteen or twenty years of market data and feed it into an optimiser to come up with a recommended portfolio.
However, unless we see an exact repeat of the same financial conditions and returns, this approach is highly unlikely to be optimal.
The performance of bonds in recent decades illustrates this neatly. Most of the major bond indices have their start date in the early 1980s – a time when government yields were in double digits.
Since then, we have seen yields fall to low single digits. The drop in yields has led to a steady tailwind of capital appreciation so that bonds have shown very strong returns both in absolute terms and relative to other asset classes.
But with government yields of 2 per cent or lower in the major economies, there is precious little prospect of matching those past returns from this starting point. A reversion of yields to their long-term average would result in low or negative returns.
Were we to use the last thirty years of the Government bond index as a predictor of returns over the next thirty years, we would be doomed to disappointment.
If we can’t use historical returns as the input for our modelling then what can we do? This is where some hard thinking is needed to create a robust framework for estimating future returns for asset classes and also their statistical distributions.
Many people choose to use a risk premium approach when addressing the question of returns. The idea is that investors are compensated for holding risky assets, so that over the long run they receive a greater return than they would achieve in a risk free asset. This additional return is termed a risk premium.
Historical analysis over a period of more than a hundred years and over several geographies suggests a defensible risk premium assumption for equities would be something like 4 per cent a year over cash.
Does this mean that we should expect equity returns to be 4.5 per cent over the next 12 months given a current cash rate of 0.5 per cent?
No. The concept of risk premium only makes sense when considering very long periods of time. Equity markets are volatile and the chance of us seeing a 4 per cent excess return in any 12-month period is slim.
However, the longer the holding period the more confident we should be of achieving an annualised excess return over cash of this order of magnitude.
History tells us that buying equities when they are expensive results in a significantly lower five to 10 year return than if we buy them when they are cheap.
However, we do not think it is right to factor current market valuations into long-term return assumptions. It is a good discipline to think of long run return assumptions on the basis of money that will be invested five or 10 years from now when the situation is likely to be very different from today.
This is helpful, because it allows us to separate opportunistic thinking from long-term strategic thinking. If we take a long enough time horizon we can ignore short-term market dynamics.
In practice, we model strategic allocations over 40 years. While we can reduce the length of the modelling period, we have to recognise that this increases the variability of outcomes.
Forty years is a period consistent with the lifecycle of a typical worker making contributions into a pension scheme or with the sort of time horizon a sovereign fund storing wealth to distribute to future generations may have in mind.
If one of the asset markets in our allocation is currently judged to be at an extreme, then we aim to address this through a tactical asset allocation discipline which explicitly seeks to take positions according to shorter term inputs in order to generate additional return or to protect capital.
This approach allows us to be very clear about the proportion of the total portfolio held in a particular asset class that derives from long-term considerations and how much derives from short-term tactical positioning.
Attempts to create a less well-separated approach led to a lack of clarity about exactly what role each percentage holding is playing in the portfolio and how its performance should be judged.
One area where timescale is at the forefront of the asset allocation question is ‘target date’ funds. These are funds where the investment manager explicitly manages the asset allocation to match the investor’s time horizon.
An investor in their twenties saving for retirement should be able to tolerate a lot of equity risk. Their largest asset, when projected to the expected retirement date, is the value of contributions yet to be made.
A bear market in equities would allow them to buy at lower prices and profit from the recovery, even if this takes a long time.
An investor in their sixties will not have the same tolerance for equity risk.
If they have 100 per cent exposure to equities just before retirement and markets fall significantly, there is a real risk they will be unable to make good the damage done – either through a recovery in the market or through increased contributions.
This argues for a far more conservative allocation.
So over a 40-year time period, the asset allocation should change from a more growth oriented portfolio to a much more conservative one. Many investors struggle to implement such a strategy unaided.
Asset allocation is the key factor in determining the performance and volatility of long-term investments. Deciding on an appropriate strategic asset allocation requires the consideration of a wide range of factors.
In a time when markets seem particularly unpredictable, disciplined asset allocation provides a way to generate more consistent returns at reduced volatility by taking calculated risks.
The importance of experience and common sense as well as a robust and well considered approach to modelling should not be underestimated.
Richard Skelt is chairman of asset allocation at Fidelity Worldwide Investment.
