In my last article, I looked at some important messages revealed by the results of the latest annual survey from the National Association of Pension Funds. I will continue this theme here by considering an issue which is not only becoming increasingly recognised as important to pension schemes but to investment strategies generally – asset allocation.
The NAPF has identified that pension schemes are paying much more attention (not least prompted by recommendations in the Myner's report) to the benefits of a structured strategy for asset allocation or portfolio construction. These principles were also a significant feature of the Sandler review and so are just as important to investment planning as they are to pensions.
Here, I will seek to outline a number of fundamental strategies and concepts which can be easily and very profitably applied by portfolio planners and advisers. The concepts and strategies at the core of this article are diversification and correlation, projection of investment returns and the measurement and use of volatility.
It is widely believed among advisers and clients that, to realistically hope for higher levels of reward from a portfolio, it is necessary to accept higher levels of risk. However, this perceived relationship between risk and return, while usually true of individual asset classes and sectors, must be called into question when planning and considering a well-structured and diversified portfolio.
The fundamental aim of this article is to discuss ways in which a portfolio might be constructed to produce lower levels of overall risk with maintained levels of returns or improved levels of returns with maintained levels of risk.
It should be noted that diversification might not reduce portfolio risk – or not by as much as anticipated – if the investments held in the portfolio might be expected to rise and fall in value in a similar way to each other. To ensure reduction in risk by diversification in a portfolio, it is essential to verify that the spread of investments are lowly correlated with each other, meaning the price behaviour of one of the constituent parts has little or no bearing on the price behaviour of one or more of the other parts.
If we consider a portfolio made up of entirely non-correlated holdings, it can be seen that total returns would be the average of the constituent parts but the risk profile of the portfolio as a whole would be significantly reduced as, when parts of the portfolio perform well, others will perform badly.
If the portfolio was wholly invested in just one asset class, the returns might be no better than our widely-diversified and lowly-correlated portfolio but the residual risk would probably be significantly higher.
This is a fundamental result of portfolios which give consideration to correlation between different asset classes and sectors. An investor can probably expect to achieve a higher rate of return for no greater level of portfolio risk or a lower level of residual risk with maintained levels of return.
I recommend advisers to obtain up-to-date correlation grids showing inter-correlation between the major asset classes (see below). The planning of a lowly-correlated mix of asset classes and sectors could be a convenient first step in portfolio construction as it aims to control or reduce risk profile. Controlling or reducing risk using correlation is one of the two main aspects of portfolio construction – maintaining or increasing returns is the other.
It is quite simple to reduce portfolio risk by using correlation factors but this might be achieved at the expense of a reduction in returns if the adviser is tempted towards more mundane assets such as cash and fixed-interest gilts.
Identifying likely returns from different asset classes and sectors is not simply a strategy of assuming that historical returns are likely to continue in future. Most portfolios will not, of course, invest in one asset class, so a sample consolidated projected rate can be calculated.
Having used correlation data to help restrict or reduce the level of portfolio risk and then used realistic growth rate projections to ensure this reduction of risk is not matched by a reduction in portfolio returns, there will be many instances where it is appropriate to enhance or reduce the overall risk/reward profile to either aim for increased returns or reduce risk further.
Important changes can be made to a portfolio by considering the relative volatilities of asset classes, sectors and funds. This is most conveniently achieved by noting the measurement of volatility of each potential or proposed investment in the form of standard deviations. These are a measurement of past volatility of an investment. The higher the standard deviation, the higher the past volatility of that fund or asset class. These measurements should not be taken to indicate a better or worse investment opportunity as differing investor circumstances will require different risk profiles.
If an adviser has determined a broad asset allocation model for a particular portfolio, how might standard deviations be used to assist the selection of appropriate funds and sectors within each asset class?
Comparing the standard deviations of asset classes and sectors will reveal which have had the highest and lowest volatility and by how much. Over the last five years, for example, equities have demonstrated the highest volatility, some four times higher than the asset class with the lowest volatility – property. More surprisingly, the averaged managed fund has had a volatility only 20 per cent less than equities and even defensive managed funds have only managed to halve the average equity volatility.
Beyond this comparison of asset class volatility, further important messages can be gleaned from comparing the respective volatilities of individual funds within each sector. Certain funds have demonstrated markedly lower volatility than their sector average and could be particularly considered in situations where it is desired to further reduce the risk profile of the portfolio.
Conversely, other funds have demonstrated a much higher volatility than the sector average and might be considered where an investor aims for above-average returns and is willing to wait to crystallise profits when his holdings are doing particularly well.
It should be understood that standard deviations are an historical measure and do not give a direct and precise indication as to whether certain assets or funds will demonstrate the same volatility in future.
However, volatility which has been significantly higher or lower than average generally has not happened by accident. There will usually be an underlying reason, such as management style or the precise nature of assets held within the fund.
I must stress that there are many techniques which may be used in planning a portfolio and I am not suggesting that, by omission, any of them are less useful than those on which I have concentrated.