A recent Wall Street Journal article by Jason Zweig discussed some recent data measuring risk attitudes, which showed attitudes varying hugely from state to state in the US. This was interesting enough as a demonstration of how heavily investors’ momentary risk attitudes are influenced by the surrounding context. But, precisely for this reason, the article should have come with a health warning: any measure of risk tolerance that displays such instability and sensitivity to the immediate context should absolutely never be used to guide the construction of long-term portfolios.

Imagine a nationwide survey suggested that a certain popular personality trait known to be stable over the long term, such as one’s degree of extraversion or introversion, were found to vary dramatically based on geography. Do you start to theorise about what breed of psychologically inclined inhabitant one area attracted that another did not? Or do you question the validity of the survey?

Risk tolerance, measured correctly, is a stable, long-term psychological trait. Robust and reliable measures of it should never display the sort of instability that those data show. If they do, they’re simply not measuring what they purport to measure: the investor’s long-term, reasoned willingness to accept the possibility of lower long-term returns for the chance of achieving higher ones. They are instead measuring momentary emotional responses to risk today. These are no doubt interesting but should never be used as a foundation for portfolio construction.

This matters because the results of unreliable measures are being used to determine the risk level to bake into a portfolio an investor might hold for the next few decades. The stakes are high. Life savings and the life goals they seek to fund are on the line.

It is right to imply that risk tolerance is only one aspect of determining the right level of risk to take with one’s investible assets (for example, for most investors the largely neglected component of risk capacity is far more important). However, long-term psychological willingness should never be mixed in with short-term “emotional contagion” in determining this right level of risk; “gossip, news and beliefs” are dangerous grounds for long-term portfolio construction.

The Harvard study mentioned in the article found one in three investors had bought a stock because someone other than a financial professional had talked with them about it probably strikes no one as the pinnacle of sound financial judgement. And yet this is exactly what a measure that varies so much depending on one’s neighbours does: it just formalises the process! Moreover, such unstable measures are inherently procyclical, registering high levels of “tolerance” when times are good, and lower when times are bad. Building “buy high, sell low” into an investment process is self-evidently a bad idea.

Valid and reliable measures of risk tolerance and the short-term behaviours investors exhibit in seeking emotional comfort with the investment journey both have crucial roles to play in a risk-profiling process. The second they start talking over each other, the guidance they were designed to give is lost.