David Norman explains how assessing risk is one of the most important aspects of any investment decision.
The information contained in this page is for professional Financial Adviser use only.
Most advisers and planners will be using a risk profiling tool to assist with their investment decision making for clients. In simple terms there are two discrete steps: first to assess the risk profile of the client (attitude to risk, capacity for loss, need for return over inflation etc.), then to match this risk profile with a suitable asset mix.
Risk tools typically use some form of optimisation model (e.g. mean variance optimiser) to construct an efficient frontier from which the risk and return can be derived. The tools use long term asset class return, volatility and correlation assumptions to attempt to build the optimum portfolio for each level of risk (as measured by volatility).
The assumptions used by most optimisers are based on some form of historic data – long term return data for example. These may be adjusted for expected future returns to make them more forward looking than just backward facing.
There are many criticisms of these models – that they are backward looking, that volatility is a poor proxy for risk – but there is also wide acknowledgement that they represent the best available option. It is also true that most optimisers have some form of additional adjustments applied to their results, otherwise poorly diversified portfolios or portfolios which are commercially unacceptable may result.
Clearly the data that is used as inputs to these tools is critical to the output. Small changes to long term return assumptions can make substantial asset class changes for a given risk profile. There is also concern that liquidity risk is growing (particularly in bond markets) and that models do not take this into account at all. History also tells us that assets tend to be more correlated in times of market stress – just when it is most needed (probably linked to the lack of liquidity).
It is also important that advisers understand whether the optimiser outputs are real (after inflation) returns or nominal.
The elephant in the room.
Before 2008 inputs to models were straightforward "long term equities beat bonds". A balanced portfolio would therefore likely be skewed to equities - probably a 60:40 split in favour of equities.
This would have been validated by the Barclays Equity Gilt Study 2014 (using 2013 data) (real annual returns %pa, after inflation) for UK asset classes.