Anxiety, Overconfidence and Excessive Risk-Taking

I have been doing some work on money market codes of conduct. In my research, I came across a fascinating article on “Anxiety, Overconfidence and Excessive Risk-Taking” by Thomas Eisenbach and Martin Schmalz (FRBNY Staff Report No.71, March 2015).

The authors note that it is widely accepted and well substantiated that people tend to systematically underestimate risk. Consequently, they tend to be overconfident and exhibit a natural propensity to excessive risk-taking. The aim of their paper is to integrate such distortions in belief into economic theory by asking, what is the rationale for overconfidence?

The proposition is that overconfidence depends on the time horizon of decision-making. People are less inclined to take imminent risks and more inclined to take distant risks. We prefer risky gambles with high returns to safer alternatives as long as they are temporarily distant but when the future arrives, we tend to ‘chicken out’. The authors called this inconsistent risk aversion ‘anxiety-prone’ behaviour. The authors’ Preference-Based Model of Overconfidence’ basically seeks to establish an equilibrium level of overconfidence that balances excessive risk-taking due to overconfidence in some states of the world with insufficient risk-taking due to anxiety in other states.

When anxiety-prone behaviour is combined with an ability to forget, ignore, distort or, in some way, deprive ourselves of information about the future, we can become overconfident and more inclined to take higher (future) risks than we otherwise would. To my childish delight, the authors refer to ‘a preference for a biased posterior’. The concealment of information about the future is what reconciles a person’s current risk aversion with their future risk appetite.

Further, the authors demonstrate that, counterintuitively, people who are more prone to anxiety when facing immediate risk are the ones more likely to exhibit overconfidence about the future. In other words, anxiety-prone behaviour is worse in more anxious people! This makes intuitive sense if you consider that deluding oneself about the future would be the natural preference for an anxious person as it reduces their anxiety!

At first glance, it would seem unlikely that we would forget, ignore, distort or deprive ourselves of information about the future. But the authors explain that they are not necessarily describing the actual mental process of a decision-maker. Rather, the behaviour observed is the same as if the decision-maker consciously made the decision. And while it seems odd to suggest making a conscious decision to forget, the actual process may be one of not making the necessary conscious effort to recall something.

Moreover, what often happens in practice is that a decision-maker will implement, encourage or acquiesce in the creation of a management culture that keeps themselves misinformed about risks. One manifestation of such information selectivity is the much remarked upon scarcity of openly-expressed critical upward feedback. Lack of upward feedback is often said to be implicitly or explicitly mandated by the head of an organisation (‘killing the messenger’). The lack of formal and open upward feedback is the reason for the establishment of formal, anonymous upward feedback mechanisms, including whistle-blowing provisions. Such observations resonate with accounts of the management culture of banks before the recent crisis. There is also direct evidence on the biased selection of information from studies of financial decision-making that find that investors look up their portfolio performance less often after receiving a sign about increased risk, such as a price fall (the ‘Ostrich Effect’). And one study found that retail investors display little demand for unbiased advice, even though they are the ones who need it the most!

There is an interesting discussion by the authors of ‘self-manipulation with alcohol and drugs’ as a means of self-deception. This is particularly relevant to codes of conduct. It is well established that alcohol is associated with risky behaviour, including gambling, through boosting (over)confidence. But alcohol has also been shown to lead to forgetting, especially of negative signals. And there is evidence that drugs are used deliberately to induce performance changes, particularly by anxious individuals. Anecdotal evidence on the ‘widespread use of…cocaine by professional traders’ is consistent both with strategic self-manipulation and with overconfidence.

The authors also argue that, under assumptions of horizon-dependent risk aversion and self-delusion through the biased selection of information, a riskier environment is more conducive to overconfidence. The return on a risky strategy looks the sensible choice if one does not recognise its true risk. This leads to excessive risk-taking in risky environments and insufficient risk-taking in safe environments. As overconfident decision-makers have a greater appetite for risks, they sustain and reinforce excessive risk levels at the top of the cycle. Conversely, underconfidence helps sustain and reinforce the lack of action at the bottom of the cycle.

A final sobering thought. While our propensity to overconfidence and excessive risk-taking can damage financial markets, it is also the motive force that drives entrepreneurship. Get out of that one!

Leave a Reply