Multi-asset
12 min read 3 Oct 24
In financial markets, the existence of ‘bubbles’ and ‘crashes’ seem to be the clearest indication that behavioural forces can even take hold in a field dominated by detailed research, algorithms, and high-powered risk models. And yet years of being aware of bias does not seem to have made markets more ‘rational’.
Most of the biases examined in academic behavioural finance focus on the mistakes that individuals make, without necessarily linking them to the types of collective behaviour that is needed to create opportunities in markets. This is important; evidence that individuals can display bias is not enough to prove that it is possible to ‘beat the market’. Indeed, many believe that the vast variety of individual biases ultimately cancel each other out, are corrected by quantitative strategies, or are quickly exploited by ‘smart money’.
The Global Macro Team do believe that human emotion can drive exploitable market inefficiencies, but that the key is identifying collective errors. Financial markets are collective manifestations of human emotion and are subject to bouts of panic, bubbles or overconfidence. These instances, and their subsequent corrections, are what the team refer to as ‘episodes’.
The following examples outline some significant episodes from recent years. We examine how they are derived from biases identified in psychological literature and demonstrate how they can be identified with an appropriate framework.
Myopic loss aversion actually combines two concepts from the psychology of decision making1.
The first – ‘loss aversion’ is the observation that human beings are more sensitive to losses than gains of the same amount. The cause of this potentially lies in the brain itself since studies have shown that the pain of financial loss is felt in the same part of the brain as physical pain2.
Even outside the world of investment losses and gains, there is evidence that negative news has a greater impact on humans than positive3.
The second observation is myopia – a tendency for investors to behave as if they are operating with a shorter time horizon than they really are, and for human beings to give more weight to recent information (‘recency bias’4).
It has long been an adage in the media that bad news is more likely to sell newspapers than good news: “if it bleeds, it leads”.
In 2023, academics building on the ever-growing ability to crunch huge amounts of data and study human language actually provided evidence of this effect in a study of online news consumption. Their study found that, although there were more news headlines with positive words than negative, negative words in news headlines increased readership while positive words had the opposite effect5.
Loss aversion on its own is often used to explain why investors are unwilling to close positions that have lost money (the pain of acknowledging the loss means that they are desperate to ‘get even’).
However, loss aversion also has a forward-looking element, particularly when combined with myopia. We are less likely to take risks when the potential for loss looms large in our consciousness6. Unfortunately, the risk of loss looms largest when we have just experienced it, and when combined with our inbuilt bias to respond more to negative than positive news, it is easy for panic to ensue.
Sometimes the market can be right: maybe short-term losses can be associated with a deterioration in long-term prospects for returns. However, short-term losses are often a source of opportunity, and there are several symptoms that can at least suggest when a market reaction is a knee-jerk emotional response rather than careful consideration of long-term risk.
First, the speed of market declines in itself can be an indicator that investors are not taking time to consider genuine changes to long-term return prospects, and instead are more focused on protecting what they already have. Second, correlation across different sectors and regions can also indicate a “sell everything” mentality, and desire for safety at all costs.
In conjunction with observing the nature of price moves themselves, there are a growing number of techniques for assessing signs of widespread negative sentiment and panic. A global pandemic is clearly a terrifying event outside of what is happening in markets, but advances in analysing news flow meant that a number of studies in and around Covid itself were able to identify panic as a driver behind market moves7.
Conservatism bias is used to explain the observation that human beings are often too slow to change their beliefs when presented with new information.
There can be many reasons for this. In the realms of politics, for example, ideologies often make up an individual’s sense of self, and so people can be unwilling to change what is essentially a fundamental element of how they identify themselves in the world. However, in markets and other areas where beliefs are (somewhat) less about identity, it is often suggested that ‘anchoring’ and conformity (or ‘herding’) play a stronger role.
Anchoring is a cognitive bias that causes us to rely heavily on the first piece of information we are given, while conformity is our tendency to be influenced by the crowd in social environments and our aversion to feeling like an outsider. In financial markets it can feel very risky and embarrassing to take views that are at odds with experience or the beliefs of the majority.
The most famous experiment demonstrating conformity was conducted by Solomon Asch in the 1950s. Asch would test subjects by placing them in groups where all other members were actors, paid to give a clearly incorrect answer to a straightforward question. Under the pressure of being in the group, many test subjects would end up conforming to the group view even when (as they later admitted) they believed it to be obviously incorrect8.
When it comes to conservatism and anchoring, most examples focus on tests of people’s ability to adapt their views of probability when given new information. In the world of finance, quantitative methods can do much to correct for these errors. However, the realm of market forecasting is an area where the pressure to conform (both social and career driven) can have a far greater influence.
By the end of 2021, US government bond yields were still below where they had been prior to the pandemic. This was in spite of measures of core US inflation hitting levels close to 6%.
Despite the clear evidence that inflation had risen, most forecasters predicted that inflation in 2022 and beyond would return to the same 2% level they had been forecasting – correctly – for many years previously.
Indeed, even when forecasters were shown to be wrong about their 2022 inflation expectations, and had to revise up their 2022 number from 2.2% at the start of 2021 to 8.1% by the end of 2022 (the orange line below), their forecast for 2024 still stood at just 2.5%.
Charts like the one above can show us how forecasters are updating their predictions – or not. In the case of US inflation, you can see how after more than a decade of 2% forecasts, it might be more difficult for forecasters to change their anchor. This very likely accounted for the exceptionally low bond yields at the time.
In terms of conformity and herding, one can also look at the distribution of beliefs. It could have been that the 2% forecasts for 2022 were just the average result and that there was actually wide disagreement among forecasters. However, this wasn’t the case. In July 2021, as the Cleveland Fed inflation measure was reaching 4%, the vast majority of forecasters (>45%) expected inflation in 2022 to be between 2.3% and 2.8%. Only 5% of forecasters thought it would be above 3.7%.
This ‘clustering’ illustrated just how confident and widely held the view was. In such market conditions any shock to these confidently held beliefs are likely to create huge amounts of volatility, as we subsequently saw across all assets in 2022.
‘Representativeness’ is a not a bias in its own right. Rather it is a heuristic, or mental short cut, that helps us to make decisions quickly9. It refers to the human tendency to judge a situation based on whether or not it has similarities to past experience.
In most walks of life the representativeness heuristic is highly useful and sensible, however it can also result in the creation of stereotypes. And in financial markets, it can drive forecasters to dangerous overconfidence.
Many studies have shown the adverse effects that can occur when humans assume that the current situation is the same as past experiences just because certain salient aspects are the same.
Several studies on medical and psychology professionals, for example, have shown evidence of misdiagnosis based on the stereotyping of patients or similarity of symptoms with other illnesses that had been experienced more recently or more often10.
By the start of 2023, there was a high degree of confidence that there would be a US and global recession, with an associated policy ‘pivot’ to lower interest rates.
There are a number of possible reasons for this view, and why it was subsequently proved incorrect. A key issue however was the inversion of the US yield curve. In the past such an inversion has been an effective indicator, and because it resembled the past, investors confidently asserted that a recession was inevitable this time. Some quantitative models were even putting the odds of recession within a year at 100%. You would be hard-pressed to find a similarly strong example of overconfidence, and it also illustrates that using quantitative approaches does not always correct for error.
Some will point to the fact that the extent of US government spending that was to come could not have been foreseen at the time, nor the effect this would have in stopping a recession. However, this argument only reinforces the need for huge amounts of humility when it comes to the future and a further challenge to overconfidence in forecasts.
Perhaps more significant than the spending argument is that few asked why the curve was inverted and how this differed from past occurrences. A world in which inflation was elevated due to supply issues that were thought very likely to abate (albeit not to the 2% of many anchored forecasts) is very different to past cases where the inversion of the curve revealed a belief that tighter policy would itself dramatically lower future growth.
“Disaster myopia” is the idea that the longer time passes without a disaster, the more human beings understate the risk of a shock occurring in the near future. It was initially put forward to explain cases where banks appeared to be more likely to take on extra risk the longer time passed without an economic shock to the system.
The academics who coined the phrase suggested that it was a result of human use of the ‘availability heuristic’ (where we think things that can be more easily remembered – or imagined – are therefore more likely) and the threshold heuristic (where things that are simply very unlikely are treated as effectively impossible)11.
There are plenty of examples in our everyday lives where we find ourselves being cautious only after an adverse event has happened. For example, there is evidence showing that people are more likely to purchase insurance after natural disasters12.
It has also been suggested that social organisations like businesses can display similar behaviour. The passage of time and the turnover of staff can have the same impact as an adverse event ‘fading from the memory,’ causing more cautious procedures to be abandoned and the same mistakes repeated13.
After more than a year of the forecasted US recession not occurring (and with the associated strong returns from many assets), investor sentiment today seems to suggest a high degree of comfort with all kinds of risk.
Investors are less worried about a recession that seemed so obvious just a year and a half ago, less worried about the “higher for longer” cash rates or inflation rates that caused a panic in late 2023, and less worried that the high cash rates we already have will have lagged negative impacts on growth.
This seems to be reflected in equity and credit valuations today, as well as in lower implied volatility in options. The latter has been suggested as a result of “hedging exhaustion”: the tendency for investors to give up on hedging portfolios because they are tired of paying for insurance that is not (or hasn’t been) needed.
Looking back at historic equity market volatility up to the present day shows us how suppressed current levels of volatility are on a long-term basis.
This in itself is not a cause to be bearish. The market can be right about risks being low, or become more positive rather than simply relaxed. However, it is likely that risky assets could be more than usually vulnerable to any disappointment.
A key way of trying to avoid disaster myopia is to start the investment process by thinking about how well one is compensated for a wide range of risks, rather than what you believe is most likely to happen or how you feel about risk today. Valuation rather than forecasting thus becomes the key influence on decision making and helps instil some degree of humility.
We should be wary of taking evidence of behavioural bias as giving carte blanche to the idea that generating excess return is easy, or that any pricing differing from our own view is obviously the result of a behavioural error. The market is indeed often wrong, but it is also often the best guess we have given the information available at the time.
However, history does suggest that investors as a whole are typically overconfident about the macro future, in no small part due to group dynamics and consensus thinking. This can mean that, as in the case of bond markets in 2022, investors are not well compensated from any surprise to this consensus (in that case the view that inflation was transitory).
Moreover, these periods of overconfidence are frequently punctuated by phases of panic, when investors are reminded of just how uncertain the world really is. In these phases, such as the early days of Covid, the consensus is no longer a source of confirmation for confident views, but a tide to be fought as investors ‘rush to the exit’.
The Global Macro Team believes that there is scope to exploit both phases of overconfidence, and episodes of panic (or return chasing), through a combination of value signals and careful assessment of the potential behavioural forces that lie behind market pricing.
The value of investments will fluctuate, which will cause prices to fall as well as rise and investors may not get back the original amount they invested. Past performance is not a guide to future performance. The views expressed in this document should not be taken as a recommendation, advice or forecast.