How do your tools account for loss aversion?

Loss aversion and Risk Tolerance are very different things. A calculation of loss aversion has no place in a Risk Tolerance assessment.

Loss aversion is, by definition, an in-the-moment decision bias. It derives from immediate context and framing effects. This is at odds with Risk Tolerance, which (when suitably assessed) is a measure of someone’s long-term stable willingness to trade off risk and return.

Risk Tolerance should absolutely not reflect immediate framing and context. There would be an argument for measuring loss aversion separately to Risk Tolerance if: a) there were a clear way to use this knowledge; and b) if it were possible to measure it in a stable way (such that the investor’s score in the assessment were transferable to the broader investment context). Neither of these is the case.

As above, loss aversion is not stable, and is therefore unsuited for establishing what level of risk an investor wants in the medium to long term. A client that says they are willing to accept, say, a 10 per cent loss on an investment doesn’t give you a Risk Tolerance score. The evidence is clear that answers to such quantitative questions are heavily context dependent, more a test of numeracy than personality, and do not reliably translate to what investors would really be willing to accept when the situation actually arises.

Our present assessment of our future preferences is one of the most pervasive biases running through decades of behavioural literature. An investor’s tendency towards being influenced by short-term reference points, framing, and immediate gain/loss outcomes is important. We measure this through our other psychometric scales, not as part of Risk Tolerance.

In particular, our Composure scale offers a highly stable measure of an investor’s proclivities towards short-term behavioural responses (for example, low Composure investors will have a higher tendency towards loss aversion). But measures of loss aversion itself are simply not stable.

The academic literature shows individual measures in experimental conditions will differ from one assessment to the next even when the questions are identical. Moreover, questions on loss aversion are highly sensitive to question design and the magnitude of values used in any examples.