Allocating risk requires measuring both investments and investors

Allocating risk requires measuring both investments and investors

In last week’s superb FT article on matching investors to investment risk, (Risk, the retail investor and disastrous new rules) John Kay rightly points out the foolishness of using recent historical volatility as a proxy for (future) investment risk. However, whilst focussing on the problems with establishing the risk of investments, he only briefly touched on determining appropriate risk levels for investors.


Determining the right level of risk for each investor is akin to packing a suitcase for a trip, where the investments that go in the suitcase are bigger when they are riskier, and smaller when they are less risky. Kay’s key point is that the measurement of investment risk required by the key information document (Kid) fails to measure the risk that really matters to clients. No traveller wants to arrive at the airport only to discover that they must pay a massive fee for having an overweight suitcase, just as no investor wants to learn the hard way that they have taken much more risk than was right for them.

But, it is equally vital that each investor starts out with the right sized suitcase for their particular journey. Kay restricts his comments on this investor side of the problem only to mentioning that selecting the right luggage arises from “an elaborate questionnaire designed to ascertain your ‘risk appetite’”.

On Investors – how big should the suitcase be?

Risk appetite questionnaires need not be (indeed, should not be) “elaborate”. Over-engineered and superficially sophisticated ‘revealed preference’ approaches result in exactly the same problem for investors as Kids do for investments: highly unstable reflections of current short-term preferences, not long-term needs. Overwhelming evidence from behavioural science tells us that finding the solution that feels right for the present is unlikely to be the best for your long-term needs. Instead, advisers and wealth managers should use (non-elaborate) psychometric tests to establish the investor’s risk tolerance relative to the general population.

Risk tolerance cannot be accurately assessed by putting complex risk-return choices in front of people, or asking them to choose between possible portfolio outcomes in the distant future. As humans, we simply don’t know how to judge the level of risk we’re prepared to take over the long term. All such questions merely determine our current comfort levels with current conditions, and so provide a poor guide to long-term investing.

To successfully use a psychometric risk tolerance questionnaire, it is vital to have a robust methodology for mapping investors to investments—and one that isn’t based on just asking clients what they would choose. Such a methodology is largely lacking from most risk profilers, which is why so many erroneously rely on elaborate questionnaires to get investors to (unreliably) map themselves to a model portfolio. Other providers may offer investment risk bands for each investor profile, but usually without a robust rationale for doing so, and the resulting bands are usually so broad that one could put almost anything in the portfolio and be within the boundaries.

Kay identifies that investment risk is a “property of an investment strategy taken as a whole, not a property of the individual components of a portfolio”. The same is true of an investor’s risk tolerance: it should be determined at the level of the individual, and not for a single ‘pot’, investment account, or portfolio component. Risk tolerance, as a personality trait, arises from the psychology of the person, not from an incoherent bundle of attitudes to component investments. This doesn’t mean that investors can’t have sub-components of their portfolio reflecting different risk levels, but risk profiling should be done at the highest level: the level of the individual.[1]

And whilst an individual’s risk appetite is important, it is far from everything that is needed to establish the right level of portfolio risk. Risk tolerance needs to be woven together with measures of risk capacity, which for most clients is the more important of the two. Required by the regulator, but frequently given no more than lip service by the industry, risk capacity requires integrating knowledge of an investor’s entire balance sheet (including assets as well as liabilities; and investible assets as well as non-investment holdings) with knowledge of their anticipated future income and expenditure. A well-designed risk capacity framework, of which there are few good examples, permits the appropriate investment risk profile to be updated dynamically as circumstances, plans, and goals change—even as risk tolerance remains constant (as it generally should, if measured correctly).

On Investments – how much risk does each investment add to the suitcase?

When it comes to the measurement of investment risk via historic volatility, we agree with Prof. Kay’s argument entirely, which we summarise in five elements. This approach fails to recognise that:

  • risk, as the concept is meaningful to investors, is about the chance of downside, not variability;
  • risk is about long-term outcomes, not fluctuations of value on the journey;
  • risk is not adequately reflected by information about the recent past—indeed, there is no such thing as risk in the past, only volatility;
  • risk is a feature of the client’s portfolio, not individual products therein; and
  • risk is particular to the individual investor, and so needs to account for their particular combination of risk tolerance, risk capacity, and financial personality.

If organisations are to ensure they are providing the right level of risk for their clients, they need to stop pretending that it is possible to arrive at accurate measures of risk for individual investments ‘bottom-up’. The more granularly we try to measure risk, the less credible the measures become. Instead we should turn our attention to robust frameworks for assessing the suitability of clients’ portfolios, though a combination of top-down examination of the overall asset allocation, and bottom up tests to identify concentration risks that could indicate the top-down measure is inadequate.

Assessing risk is difficult, but it is vital that on both the investor or investment side we don’t reach for superficially sophisticated solutions that end up getting things very precisely wrong, rather than approximately right. The solutions developed by Oxford Risk offer best-in-class outcomes precisely by being robust but not over-engineered:

  1. A simple, yet empirically validated psychometric risk tolerance scale that doesn’t rely on spurious revealed preferences or over-engineered quantitative choices;
  2. A dynamic and quantitative risk capacity model capable of coping consistently with variable levels of completeness of client information;
  3. An academically sound algorithm for blending risk tolerance and risk capacity, and dynamically updating the resulting risk profile to respond to changing circumstances and preferences;
  4. A practically and theoretically robust methodology for mapping investor risk profiles onto appropriate investment risk levels (when measured in a Kay-appropriate manner);
  5. A suite of psychometric behavioural scales to deepen client understanding, improve client engagement, and manage emotional comfort along the investment journey;

A framework for ongoing monitoring of client portfolios, combining top-down and bottom up tests to establish suitability of the client’s actual holdings without relying on spurious granular risk measurement of individual assets.

[1] This is even true for goals-based investing, where it is meeting the overall portfolio of goals that truly matters, not each goal in isolation – see A Behavioural Perspective on Goals-Based Investing